platform/upstream/gstreamer.git
6 years agoadaptive: extract variant info
Eunhae Choi [Fri, 8 Dec 2017 02:23:38 +0000 (11:23 +0900)]
adaptive: extract variant info

- extract variant info
- apply the user-agent, cookie info

Change-Id: I38ec2e0b0248924716a010b3781b6939eb790e36

6 years agoadaptive: avoid pad switching
Eunhae Choi [Mon, 4 Dec 2017 10:01:02 +0000 (19:01 +0900)]
adaptive: avoid pad switching

Change-Id: I29fae423a03e4d642866a218058ca2efa9486f06

7 years agoMerge tizen patch based on 1.12.2
Eunhae Choi [Wed, 1 Nov 2017 13:14:06 +0000 (22:14 +0900)]
Merge tizen patch based on 1.12.2

except waylandsink,
tizen will use previous waylandsink which is included gst-plugins-tizen.

Change-Id: I9c47a204c953ecaec4f1ae6396126228d666d0bb

7 years agoMerge "Add rotation of spatial audio (first order ambisonics)" into tizen
eunhae choi [Fri, 20 Oct 2017 04:31:57 +0000 (04:31 +0000)]
Merge "Add rotation of spatial audio (first order ambisonics)" into tizen

7 years agoWaylandsink : Add checking MMVideoBuffer for GstBuffer of Codec Bufferpool
Hyunil [Wed, 18 Oct 2017 08:05:01 +0000 (17:05 +0900)]
Waylandsink : Add checking MMVideoBuffer for GstBuffer of Codec Bufferpool

Change-Id: Iafcbcd827682a4000a0f7283454ecdee1f94d142
Signed-off-by: Hyunil <hyunil46.park@samsung.com>
7 years agowaylandsink: Use handle num and plane num of MMVideoBuffer
Hyunil [Tue, 10 Oct 2017 01:53:42 +0000 (10:53 +0900)]
waylandsink: Use handle num and plane num of MMVideoBuffer

Change-Id: If2e5296125fa2e5a17f198ccb00b8796914e5729
Signed-off-by: Hyunil <hyunil46.park@samsung.com>
7 years agoAdd rotation of spatial audio (first order ambisonics)
Mykola Alieksieiev [Wed, 27 Sep 2017 15:55:48 +0000 (18:55 +0300)]
Add rotation of spatial audio (first order ambisonics)

Change-Id: I11d28fa80d652411233a0a631af482620196b0fd
Signed-off-by: Mykola Alieksieiev <m.alieksieie@samsung.com>
7 years agowldisplay: change version of tizen_policy from 1 to 7.
JunSeok, Kim [Mon, 18 Sep 2017 05:20:49 +0000 (14:20 +0900)]
wldisplay: change version of tizen_policy from 1 to 7.

Changing version of tizen_policy from 1 to 7 because of tizen_extensions
version up.

Change-Id: I9e4cd83219ae9af10760cab2915e50202d4ab44c

7 years agodashdemux/hlsdemux/adaptivedemux: Fix build warning and apply coding rule
Gilbok Lee [Tue, 12 Sep 2017 04:19:25 +0000 (13:19 +0900)]
dashdemux/hlsdemux/adaptivedemux: Fix build warning and apply coding rule

Change-Id: Ic5606190134d9b62e2b82121bf1e8ac3111b123f

7 years agoMake gstfragmented use OpenSSL instead of Nettle
Lukasz Pawelczyk [Tue, 4 Jul 2017 14:06:26 +0000 (16:06 +0200)]
Make gstfragmented use OpenSSL instead of Nettle

Change-Id: I3f180f5ba3e9a2eb593f1183d845ee1050a9f52f

7 years agohlsdemux: Fix crash when NETTLE is not used
Gilbok Lee [Mon, 4 Sep 2017 08:31:05 +0000 (17:31 +0900)]
hlsdemux: Fix crash when NETTLE is not used
hlsdemux: Fix bug for getting variant stream info

Change-Id: Ibe39e0749723d5ef41b927beb6066f91336e35b0

7 years agoRelease 1.12.2
Sebastian Dröge [Fri, 14 Jul 2017 11:03:37 +0000 (14:03 +0300)]
Release 1.12.2

7 years agoUpdate .po files
Sebastian Dröge [Fri, 14 Jul 2017 10:35:02 +0000 (13:35 +0300)]
Update .po files

7 years agopo: Update translations
Sebastian Dröge [Fri, 14 Jul 2017 10:23:03 +0000 (13:23 +0300)]
po: Update translations

7 years agoadaptivedemux: Workaround for live seek ranges when advancing
Edward Hervey [Thu, 13 Jul 2017 10:57:12 +0000 (12:57 +0200)]
adaptivedemux: Workaround for live seek ranges when advancing

This is a workaround for a regression introduced by
f4190a49c04f1d5d174cebba0bc9a03a7ec721c2
 ( adaptivedemux: Check live seeking range more often )

The goal of the previous commit was to be able to cope with non-1.0
rates on live streams which have a "seeking window" (i.e. the server
keeps around quite a bit of the live stream so you can seek back into
it).

Without that commit, two different kind of issues would happen:
* When doing reverse playback, you would never check whether you
  are outside of the seekable region. And would then continuously
  try to download fragments that are no longer present.
* When doing fast forward, you would end up requesting fragments
  which are not present yet.

In order to determine whether one was *really* outside of the seekable
window, we check whether the current stream position is still
within the seekable region.

The *problem* though with that commit is that it assumes that subclasses
will return continuously updated seeking ranges (i.e. dependent on the
current time), which is *NOT* the case.

For example:
* dashdemux does use the current UTC to determine the seekable region
* hlsdemux uses the values from the last updated manifest

Therefore if one downloads fragments faster than realtime, for HLS
we would end up at the end of the last manifest seekable range, and
the previous commit would consider the stream as being ended... which
is not the case.

In the long run, we need to figure out a way to cope with non-1.0
rates on live streams for all types of stream (including HLS).

https://bugzilla.gnome.org/show_bug.cgi?id=783075

7 years agoadaptivedemux: small refactor to avoid repeated code
Thiago Santos [Sat, 13 May 2017 22:17:57 +0000 (15:17 -0700)]
adaptivedemux: small refactor to avoid repeated code

Move segment event update to a function

https://bugzilla.gnome.org/show_bug.cgi?id=773159

7 years agoadaptivedemux: Handle prepared streams on seeks
Edward Hervey [Tue, 9 May 2017 09:41:49 +0000 (11:41 +0200)]
adaptivedemux: Handle prepared streams on seeks

This is a race that was exposed by the {hls|dash}.scrub_forward_seeking
validate test.

The "race" is that a subclass might want to change format, causing
a new stream to be created (but not exposed/switched yet) and put on the
prepared_streams list. That stream will have values (including pending
segment) from the pre-seek state.

Before the stream is exposed/switched, a new seek comes in and the stream
values get updated ... but the ones that will be changed don't get updated
causing them to push out wrong segments once they are exposed.

https://bugzilla.gnome.org/show_bug.cgi?id=773159

7 years agomxfdemux: Create index table arrays with a big enough size
Sebastian Dröge [Tue, 11 Jul 2017 17:54:25 +0000 (20:54 +0300)]
mxfdemux: Create index table arrays with a big enough size

7 years agomxfdemux: Initialize index table entries completely in all cases
Sebastian Dröge [Tue, 11 Jul 2017 17:53:22 +0000 (20:53 +0300)]
mxfdemux: Initialize index table entries completely in all cases

When seeking backwards into a previously unseen location, we wouldn't
fully initialize them and playback would fail later.

7 years agodashdemux: update manifest streams correctly if pads aren't exposed
Thiago Santos [Tue, 30 May 2017 05:47:10 +0000 (22:47 -0700)]
dashdemux: update manifest streams correctly if pads aren't exposed

In some cases, it is possible that we need to update the manifest before
pads have been exposed at all. If there are no current pads, just expose
the next prepared streams. This doesn't handle the case where a manifest
update would happen while a live streams is changing periods, which is a
type of use case that we're unaware of real usages yet.

https://bugzilla.gnome.org/show_bug.cgi?id=783028

7 years agodashdemux: Implement a default presentation delay
Edward Hervey [Tue, 30 May 2017 13:27:37 +0000 (15:27 +0200)]
dashdemux: Implement a default presentation delay

Have a fallback presentation delay is great (if not present in the
manifest), having an actual default value is better.

https://bugzilla.gnome.org/show_bug.cgi?id=783244

7 years agowaylandsink : update last_sample in paused state when signal_handoffs is set to zero
Hyunil [Wed, 12 Jul 2017 01:40:42 +0000 (10:40 +0900)]
waylandsink : update last_sample in paused state when signal_handoffs is set to zero

Change-Id: I640d8f10d54c573a615d7b1126374ac5ad62c515
Signed-off-by: Hyunil <hyunil46.park@samsung.com>
7 years agodirectsoundsrc: Properly pass pLockedBuffer by reference when calling IDirectSoundCap...
Philippe Renon [Mon, 10 Jul 2017 15:35:32 +0000 (17:35 +0200)]
directsoundsrc: Properly pass pLockedBuffer by reference when calling IDirectSoundCaptureBuffer_Lock

https://bugzilla.gnome.org/show_bug.cgi?id=784755

7 years agokmssink: Don't leak GEM primed from DMABuf
Nicolas Dufresne [Thu, 6 Jul 2017 21:20:56 +0000 (17:20 -0400)]
kmssink: Don't leak GEM primed from DMABuf

This otherwise breaks DMABuf reclaiming. This is not visible from
userspace, but inside the kernel, the DRM driver will hold a ref to the
DMABuf object. With a V4L2 driver allocating those DMABuf, it then
prevent changing the resolution and re-allocation new buffers.

https://bugzilla.gnome.org/show_bug.cgi?id=782774

7 years agoadaptivedemux: add streaming variant control
Eunhae Choi [Thu, 6 Jul 2017 08:04:57 +0000 (17:04 +0900)]
adaptivedemux: add streaming variant control

- add property to set max bandwidth/width/height limit
- add msg posting to share the current streaming variant

Change-Id: I9d58c3331a5cab6b2fc09a7b0cfdf25fb84a39be

7 years agomxfdemux: For intra-only streams, always output DTS=PTS
Sebastian Dröge [Wed, 5 Jul 2017 14:04:46 +0000 (17:04 +0300)]
mxfdemux: For intra-only streams, always output DTS=PTS

7 years agomxfdemux: Fix integer overflow in partition position comparison function
Sebastian Dröge [Wed, 5 Jul 2017 12:48:57 +0000 (15:48 +0300)]
mxfdemux: Fix integer overflow in partition position comparison function

7 years agomxfmetadata: Set field-order correctly for interlaced video
Sebastian Dröge [Wed, 5 Jul 2017 10:59:12 +0000 (13:59 +0300)]
mxfmetadata: Set field-order correctly for interlaced video

7 years agomxfmetadata: Use display width/height instead of stored width/height
Sebastian Dröge [Wed, 5 Jul 2017 10:52:25 +0000 (13:52 +0300)]
mxfmetadata: Use display width/height instead of stored width/height

Stored values contain padding and alignment, and should only be used as
fallback if neither display values nor sampled values exist.

7 years agomxfdemux: Don't include any KLV packets between header metadata and index table segme...
Sebastian Dröge [Tue, 4 Jul 2017 21:48:36 +0000 (00:48 +0300)]
mxfdemux: Don't include any KLV packets between header metadata and index table segments in calculations

Minor fixup of last commit.

7 years agomxfdemux: Allow filler and other KLV packets between index table segments
Sebastian Dröge [Tue, 4 Jul 2017 17:01:37 +0000 (20:01 +0300)]
mxfdemux: Allow filler and other KLV packets between index table segments

While only filler packets should be allowed, for good measure also skip
any other KLV packets in the range where there could be index table
segments.

This fixes parsing of partitions with multiple index table segments,
which are separated by a filler packet, or other packets.

7 years agomxfmux: Write temporal offset and correct keyframe offset into index table
Sebastian Dröge [Tue, 27 Jun 2017 12:01:22 +0000 (15:01 +0300)]
mxfmux: Write temporal offset and correct keyframe offset into index table

https://bugzilla.gnome.org/show_bug.cgi?id=784027

7 years agomxfdemux: Take temporal reordering from the index table into account
Sebastian Dröge [Tue, 27 Jun 2017 07:47:44 +0000 (10:47 +0300)]
mxfdemux: Take temporal reordering from the index table into account

This is needed to know the PTS, without that we only know the DTS and
using that also for the PTS is wrong unless we have an intra-only codec.

If we can't get the temporal reordering from the index table, don't set
any PTS for non-intra-only codecs and let decoders figure out something.

https://bugzilla.gnome.org/show_bug.cgi?id=784027

7 years agomxfdemux: Include duration in response to SEEKING query
Sebastian Dröge [Tue, 27 Jun 2017 12:41:48 +0000 (15:41 +0300)]
mxfdemux: Include duration in response to SEEKING query

7 years agoconfigure: Add --with-moc/uic/rcc options
Jan Schmidt [Wed, 21 Jun 2017 15:01:40 +0000 (01:01 +1000)]
configure: Add --with-moc/uic/rcc options

For cross-compiling, it's easier to be able to specify the
actual paths to the tools

7 years agottml: fix linking to libgstvideo
Tim-Philipp Müller [Mon, 3 Jul 2017 09:11:42 +0000 (10:11 +0100)]
ttml: fix linking to libgstvideo

https://bugzilla.gnome.org/show_bug.cgi?id=784439

7 years agovtenc: fix caps leak
Eunhae Choi [Fri, 30 Jun 2017 02:08:34 +0000 (11:08 +0900)]
vtenc: fix caps leak

Change-Id: Id805bc620ff72b3fa113cd8e85d12e693493cf60

7 years agogl: reset gl->DrawBuffer to the necessary values
Matthew Waters [Wed, 28 Jun 2017 04:45:18 +0000 (14:45 +1000)]
gl: reset gl->DrawBuffer to the necessary values

GL_COLOR_ATTACHMENT0 when a framebuffer is bound
GL_BACK if no framebuffer is bound

https://bugzilla.gnome.org/show_bug.cgi?id=784210

7 years agoglmemory: reset the draw buffer to GL_BACK
Hyunjun Ko [Wed, 28 Jun 2017 03:17:37 +0000 (12:17 +0900)]
glmemory: reset the draw buffer to GL_BACK

The draw buffer should be reset to GL_BACK since the framebuffer is already
unbound.

https://bugzilla.gnome.org/show_bug.cgi?id=784210

7 years agoFix memory leak (wrong usage gst_caps_from_string)
Gilbok Lee [Tue, 27 Jun 2017 05:50:03 +0000 (14:50 +0900)]
Fix memory leak (wrong usage gst_caps_from_string)

Change-Id: I8575e7aba0ea2e2ed9706e0a0e9a8ff16ab2ab45
Signed-off-by: Gilbok Lee <gilbok.lee@samsung.com>
7 years agoWaylandsink : Removing unsupported video formats from video caps of TIZEN
Hyunil [Thu, 22 Jun 2017 01:10:02 +0000 (10:10 +0900)]
Waylandsink : Removing unsupported video formats from video  caps of TIZEN

Change-Id: I461c414be916e74caa50583447ff283bd9121b96
Signed-off-by: Hyunil <hyunil46.park@samsung.com>
7 years agogl/cocoa: Put gstglwindow_cocoa.h into noinst_HEADERS again
Sebastian Dröge [Tue, 20 Jun 2017 16:19:51 +0000 (19:19 +0300)]
gl/cocoa: Put gstglwindow_cocoa.h into noinst_HEADERS again

And remove gstgldisplay_cocoa.h instead, which got moved to the
installed headers earlier.

7 years agoRelease 1.12.1
Sebastian Dröge [Tue, 20 Jun 2017 09:06:59 +0000 (12:06 +0300)]
Release 1.12.1

7 years agoUpdate .po files
Sebastian Dröge [Tue, 20 Jun 2017 08:24:55 +0000 (11:24 +0300)]
Update .po files

7 years agopo: Update translations
Sebastian Dröge [Tue, 20 Jun 2017 08:09:34 +0000 (11:09 +0300)]
po: Update translations

7 years agortph265depay : fix mem leak
Eunhae Choi [Mon, 19 Jun 2017 01:55:58 +0000 (10:55 +0900)]
rtph265depay : fix mem leak

Change-Id: I6ed5269baca45d8ef2b11aa0647a65a72f33100f

7 years agodirectfb: fix mem leak
Eunhae Choi [Fri, 16 Jun 2017 09:40:29 +0000 (18:40 +0900)]
directfb: fix mem leak

Change-Id: Id5069d045f280114a18dda098659a0096f8cb301

7 years agodvb: src: fix use of wrong array index for tune signal
Reynaldo H. Verdejo Pinochet [Fri, 16 Jun 2017 16:43:35 +0000 (09:43 -0700)]
dvb: src: fix use of wrong array index for tune signal

Erroneous value made "TUNE" overwrite "TUNNING_FAIL" ....

7 years agomxfdemux: Do not try to serialize unresolved metadatas
Thibault Saunier [Thu, 15 Jun 2017 18:13:14 +0000 (14:13 -0400)]
mxfdemux: Do not try to serialize unresolved metadatas

When retrieving the `mxfdemux.structure` property, it leads to an
assertion as metadatas need to be resolved for the call to
mxf_metadata_base_to_structure to be valid.

7 years agompegpsmux: fix mem leak
Eunhae Choi [Fri, 16 Jun 2017 07:51:50 +0000 (16:51 +0900)]
mpegpsmux: fix mem leak

Change-Id: I6825deeac82524c18b666498a168707978e0ef80

7 years agoglupload: Fix DirectVIV uploader for formats with a single plane
Sebastian Dröge [Mon, 12 Jun 2017 19:29:01 +0000 (22:29 +0300)]
glupload: Fix DirectVIV uploader for formats with a single plane

We have to pass the "height" as height = vmeta->offset[1] / width to the
API, which of course does not work well for formats with only a single
plane. Use the whole memory size instead of the offset in that case.

7 years agoadaptivedemux: Clear "cancelled" on uridownloader before processing manifest
Seungha Yang [Sun, 4 Jun 2017 11:23:36 +0000 (20:23 +0900)]
adaptivedemux: Clear "cancelled" on uridownloader before processing manifest

Previous commit let demux call gst_uri_downloader_cancel() on _demux_reset().
Note that, _demux_reset() called during PAUSED_TO_READY and READY_TO_PAUSED.
And, it will set "cancelled" on uridownloader which blocks the use of
uridownloader. The issue is that, subclass can use the uridownloader not only
live streaming for manifest update, but also for fetching another manifests
such as variant and rendition m3u8 of hls streaming. So to unblock it,
demux should clear "cancelled" before processing initial manifest.

https://bugzilla.gnome.org/show_bug.cgi?id=783401

7 years agoadaptivedemux: release the manifest lock ...
Mathieu Duponchelle [Wed, 31 May 2017 00:46:01 +0000 (02:46 +0200)]
adaptivedemux: release the manifest lock ...

before broadcasting preroll.

The deadlock was as follows:

-> The subclass pushes a buffer on a newly-created stream in T1
-> We take the preroll lock in T1, to handle_preroll
-> The demuxer is stopped in T2, we take the MANIFEST_LOCK
-> T1 starts blocking because it received a reconfigure event
   and needs to take the MANIFEST_LOCK
-> T2 deadlocks because it now wants the preroll_lock.

https://bugzilla.gnome.org/show_bug.cgi?id=783255

7 years agoadaptivedemux: do not erase data while updates-loop is running
Thiago Santos [Tue, 30 May 2017 05:28:21 +0000 (22:28 -0700)]
adaptivedemux: do not erase data while updates-loop is running

Make sure the manifest update loop is stopped before proceeding with the
resetting of the manifest data. Otherwise, the updates loop will try to
use it and it leads to a segfault

https://bugzilla.gnome.org/show_bug.cgi?id=783028

7 years agompdparser: remove duplicate free of client data
Thiago Santos [Tue, 30 May 2017 05:26:09 +0000 (22:26 -0700)]
mpdparser: remove duplicate free of client data

https://bugzilla.gnome.org/show_bug.cgi?id=783028

7 years agoadaptivedemux: make sure to free all "old streams"
Mathieu Duponchelle [Wed, 31 May 2017 01:14:04 +0000 (03:14 +0200)]
adaptivedemux: make sure to free all "old streams"

As we release the MANIFEST_LOCK in stop_tasks,
demux->priv->old_streams can be set, we need to free these
otherwise we may end up trying to dispose elements in the
READY state.

https://bugzilla.gnome.org/show_bug.cgi?id=783256

7 years agodashdemux: Remove wrong assertion
Edward Hervey [Wed, 31 May 2017 08:58:39 +0000 (10:58 +0200)]
dashdemux: Remove wrong assertion

This is wrong because:
* If the rate is negative we should check for the *previous* period
* adaptivedemux already does the proper checks before calling this
  method

7 years agoadaptivedemux: Fix debugging message
Edward Hervey [Fri, 26 May 2017 15:30:10 +0000 (17:30 +0200)]
adaptivedemux: Fix debugging message

GstSegment position is a guint64 and not a gint64

CID #1409910

7 years agoadaptivedemux: Check live seeking range more often
Edward Hervey [Thu, 25 May 2017 07:48:53 +0000 (09:48 +0200)]
adaptivedemux: Check live seeking range more often

The live seeking range was only checked when doing actual seeks. This was
assuming that the rate would always be 1.0 (i.e. the playback would
advance in realtime, and therefore fragments would always be available
since the seeking window moves at the same rate).

With non-1.0 rates, this no longer becomes valid, and therefore we need
to check whether we are still within the live seeking range when advancing.

https://bugzilla.gnome.org/show_bug.cgi?id=783075

7 years agodashdemux: Initialize sidx_position to GST_CLOCK_TIME_NONE
Seungha Yang [Tue, 16 May 2017 14:20:44 +0000 (23:20 +0900)]
dashdemux: Initialize sidx_position to GST_CLOCK_TIME_NONE

If a manifest has non-zero presentation time offset
(i.e., earliest presentation time specified by sidx box is not zero),
the initial sidx position shouldn't be zero. Since we cannot define
exact sidx position until parsing sidx box, set the value to unknown.

https://bugzilla.gnome.org/show_bug.cgi?id=782693

7 years agoopencv: fix mem leak
Eunhae Choi [Mon, 12 Jun 2017 08:34:08 +0000 (17:34 +0900)]
opencv: fix mem leak

Change-Id: Id26dc8effeed06a789bda89f95aa6aed663fc5ab

7 years agoopenjpegdec: fix display artifacts for RGB 8-bit
Aaron Boxer [Sat, 10 Jun 2017 11:56:48 +0000 (07:56 -0400)]
openjpegdec: fix display artifacts for RGB 8-bit

https://bugzilla.gnome.org/show_bug.cgi?id=783626

7 years agojpeg2000sampling: fix critical when sampling is missing from caps
Vincent Penquerc'h [Mon, 5 Jun 2017 14:31:52 +0000 (15:31 +0100)]
jpeg2000sampling: fix critical when sampling is missing from caps

This can happen with real files

7 years agoopenjpeg: guard against invalid memory access on crafted files
Vincent Penquerc'h [Wed, 7 Jun 2017 15:17:50 +0000 (16:17 +0100)]
openjpeg: guard against invalid memory access on crafted files

7 years agoFix up package name and origin in some plugins
Tim-Philipp Müller [Thu, 18 May 2017 09:58:20 +0000 (10:58 +0100)]
Fix up package name and origin in some plugins

7 years agomsdk: enc: set pts and dts, fix inverted sync_point flag
Scott D Phillips [Thu, 18 May 2017 17:36:50 +0000 (10:36 -0700)]
msdk: enc: set pts and dts, fix inverted sync_point flag

Set the pts and dts on the frame that we receive from the msdk.
Also fix the inverted logic in setting sync points, previously we
were marking all frames as sync points except IDRs.

https://bugzilla.gnome.org/show_bug.cgi?id=782801

7 years agoglutils: Fix GValue leak in gst_gl_value_set_texture_target_from_mask()
Jeremy Hiatt [Thu, 1 Jun 2017 01:15:05 +0000 (01:15 +0000)]
glutils: Fix GValue leak in gst_gl_value_set_texture_target_from_mask()

7 years agoMerge "Added gst.h for undefined symbol problem" into tizen
eunhae choi [Fri, 9 Jun 2017 04:27:22 +0000 (04:27 +0000)]
Merge "Added gst.h for undefined symbol problem" into tizen

7 years agoAdded gst.h for undefined symbol problem
Sejun Park [Fri, 9 Jun 2017 03:55:40 +0000 (12:55 +0900)]
Added gst.h for undefined symbol problem

Change-Id: I6793516a18871d932e3bf29010b89b02417d1833

7 years agoWaylandsink: Add plane_size info to tbm_surface for s/w codec output
Hyunil [Fri, 9 Jun 2017 00:45:40 +0000 (09:45 +0900)]
Waylandsink: Add plane_size info to tbm_surface for s/w codec output
Signed-off-by: Hyunil <hyunil46.park@samsung.com>
Change-Id: I8579f88ea423dac399153ac66deb7e192cd0b154

7 years agoMerge "apply CVE patch for security weakness" into tizen
eunhae choi [Thu, 8 Jun 2017 03:42:22 +0000 (03:42 +0000)]
Merge "apply CVE patch for security weakness" into tizen

7 years agoWaylandsink : Add SN21 video format
Hyunil [Wed, 7 Jun 2017 10:38:52 +0000 (19:38 +0900)]
Waylandsink : Add SN21 video format

Change-Id: Icc7587b4c5aacf3d7d5f7380cdaaedf99075056f
Signed-off-by: Hyunil <hyunil46.park@samsung.com>
7 years agoWaylandsink : Add plane size info to tbm_surface
Hyunil [Wed, 7 Jun 2017 06:45:21 +0000 (15:45 +0900)]
Waylandsink : Add plane size info to tbm_surface

Change-Id: I042dc83f489bda38526a4b156c575f42eff4568a
Signed-off-by: Hyunil <hyunil46.park@samsung.com>
7 years agoapply CVE patch for security weakness
Sejun Park [Wed, 7 Jun 2017 07:18:10 +0000 (16:18 +0900)]
apply CVE patch for security weakness

Change-Id: Ie7c401f1424312e2e6c2193df990525d30439083

7 years agovideoparse: Fix property handling
Thibault Saunier [Tue, 6 Jun 2017 18:58:55 +0000 (14:58 -0400)]
videoparse: Fix property handling

Usage of GstValueArray was wrong and frame-stride does not exist in
rawvideoparse

7 years agopcapparse: endianness fix
Wim Taymans [Thu, 1 Jun 2017 14:00:50 +0000 (16:00 +0200)]
pcapparse: endianness fix

Also swap the linktype after we detected that we need to do
byteswapping. Fixes a problem with reading pcap files generated
on a machine with different endianness.

7 years agoapply CVE patch for security weakness
Sejun Park [Thu, 1 Jun 2017 07:20:56 +0000 (16:20 +0900)]
apply CVE patch for security weakness

Change-Id: Ie64713714465e8ea9a0e0cf2c1f3e55e42de6fa9

7 years agoglformat: fix the usage of GST_GL_RGB565
Haihua Hu [Thu, 25 May 2017 03:05:47 +0000 (11:05 +0800)]
glformat: fix the usage of GST_GL_RGB565

GL_RGB565 is sized internal glformat, the corresponding glformat
should be GL_RGB and type is GL_UNSIGNED_SHORT_565. Otherwise will
return GL_INVALID_ENUM when creating texture.

https://bugzilla.gnome.org/show_bug.cgi?id=783066

7 years agoadaptivedemux: Don't create invalid event
Edward Hervey [Thu, 25 May 2017 14:42:03 +0000 (16:42 +0200)]
adaptivedemux: Don't create invalid event

tags could potentially be NULL

7 years agoadaptivedemux: retry download MAX_DOWNLOAD_RETRY_COUNT times before erroring
Matthew Waters [Fri, 2 Dec 2016 06:51:57 +0000 (17:51 +1100)]
adaptivedemux: retry download MAX_DOWNLOAD_RETRY_COUNT times before erroring

What we want is to retry downloading the fragment on 4xx/5xx errors
however returning EOS will cause waiting for a manifest update for live
(which may be a really long time) or stop everything for non-live.

Change that to only return EOS/ERROR once we've reached the error limit.

https://bugzilla.gnome.org/show_bug.cgi?id=776609

7 years agoglcontext: keep a ref to the active thread
fvanzile [Thu, 23 Feb 2017 23:42:08 +0000 (15:42 -0800)]
glcontext: keep a ref to the active thread

With the macOS/iOS implementations, the active thread can change
multiple times over the life of a pipeline which would expose a race in
the thread tracking.

Fix by taking a ref on the active thread while the context is active.

https://bugzilla.gnome.org/show_bug.cgi?id=779202

7 years agoqt: Use GST_GL_HAVE_PLATFORM_CGL instead of GST_GL_HAVE_PLATFORM_COCOA
Simon Himmelbauer [Thu, 18 May 2017 12:23:14 +0000 (15:23 +0300)]
qt: Use GST_GL_HAVE_PLATFORM_CGL instead of GST_GL_HAVE_PLATFORM_COCOA

The latter is not used/available anymore since years. Also fix a typo
in the include path for the Cocoa GL display header.

7 years agoamcvideodec: Unref downstream caps after usage
Sebastian Dröge [Thu, 18 May 2017 11:34:04 +0000 (14:34 +0300)]
amcvideodec: Unref downstream caps after usage

https://bugzilla.gnome.org/show_bug.cgi?id=782771

7 years agoh265parse: Fix calculation of codec_data buffer size
Sebastian Dröge [Thu, 18 May 2017 08:42:17 +0000 (11:42 +0300)]
h265parse: Fix calculation of codec_data buffer size

7 years agoglformat: Add missing GST_GL_RGB565 in some switch statement
Haihua Hu [Wed, 17 May 2017 08:26:38 +0000 (16:26 +0800)]
glformat: Add missing GST_GL_RGB565 in some switch statement

https://bugzilla.gnome.org/show_bug.cgi?id=782736

7 years agodecklink: Always use the video stream time for audio too
Sebastian Dröge [Wed, 10 May 2017 13:59:42 +0000 (15:59 +0200)]
decklink: Always use the video stream time for audio too

The audio packet times can be completely unrelated to the video stream
time, depending on the card. While this looks like a bug in the driver,
just always using the video stream time (which is correct) works as a
workaround for now.

7 years agococoa: Install gstgldisplay_cocoa.h
Sebastian Dröge [Wed, 17 May 2017 14:38:01 +0000 (17:38 +0300)]
cocoa: Install gstgldisplay_cocoa.h

It's needed by e.g. qmlglsink.

7 years agogl/viv-fb: Fix user-choice string comparisons
Carlos Rafael Giani [Tue, 9 May 2017 21:59:04 +0000 (23:59 +0200)]
gl/viv-fb: Fix user-choice string comparisons

https://bugzilla.gnome.org/show_bug.cgi?id=782921

7 years agodecklinkvideosink: Copy min(gstreamer_stride, decklink_stride) per line
Sebastian Dröge [Mon, 15 May 2017 13:37:14 +0000 (16:37 +0300)]
decklinkvideosink: Copy min(gstreamer_stride, decklink_stride) per line

Instead of just bpp * width, which might be more than we can copy.

7 years agoh264parse: fix caps leak in renegotiation
Vincent Penquerc'h [Fri, 12 May 2017 09:01:10 +0000 (10:01 +0100)]
h264parse: fix caps leak in renegotiation

7 years agoh265parse: fix caps leak in renegotiation
Vincent Penquerc'h [Fri, 12 May 2017 09:00:56 +0000 (10:00 +0100)]
h265parse: fix caps leak in renegotiation

7 years agojpeg2000parse: fix negotiation with j2c and jpc both allowed upstream
Vincent Penquerc'h [Tue, 9 May 2017 09:32:05 +0000 (10:32 +0100)]
jpeg2000parse: fix negotiation with j2c and jpc both allowed upstream

If upstream supports both, but downstream supports only jpc, j2c
would have been selected as the first in the caps.

https://bugzilla.gnome.org/show_bug.cgi?id=782221

7 years agodirectsoundsrc: Use a GstClockID to wait instead of Sleep()
Nirbheek Chauhan [Tue, 9 May 2017 10:36:10 +0000 (16:06 +0530)]
directsoundsrc: Use a GstClockID to wait instead of Sleep()

The main advantage is that our sleeps can be interrupted in case of
an src_reset(). Earlier, we would need to wait for a read to complete
before we could do a reset, which could take a long time.

https://bugzilla.gnome.org/show_bug.cgi?id=781249

7 years agoglmemory: Only use glDrawBuffer if available
Sebastian Dröge [Tue, 9 May 2017 09:25:20 +0000 (11:25 +0200)]
glmemory: Only use glDrawBuffer if available

Otherwise fall back to glDrawBuffers. Also check if glReadBuffer exists
before using it.

glDrawBuffer does not exist for GLES, only glDrawBuffers does.

https://bugzilla.gnome.org/show_bug.cgi?id=782376

7 years agomeson: Ignore msvc warnings in C++ as well
Scott D Phillips [Mon, 8 May 2017 18:01:39 +0000 (11:01 -0700)]
meson: Ignore msvc warnings in C++ as well

We were only ignoring the listed msvc warnings for C language
files and not C++. This was working by the coincidence that we did
not have any instances of these warnings in C++ files. Lately the
build of decklink has been fixed on windows, and it has an
instance of one of these warnings in a C++ file.

https://bugzilla.gnome.org/show_bug.cgi?id=782345

7 years agodecklink: Fix linking on MinGW
Nirbheek Chauhan [Thu, 4 May 2017 23:00:59 +0000 (04:30 +0530)]
decklink: Fix linking on MinGW

MinGW does not provide comsupp.lib, so there's no implementation of
_com_util::ConvertBSTRToString. Use a fallback implementation that
uses wcstombs() instead.

On MinGW we also truncate the name to 100 chars which should be fine.

7 years agodecklink: Fix building on Windows
Nirbheek Chauhan [Thu, 4 May 2017 22:32:29 +0000 (04:02 +0530)]
decklink: Fix building on Windows

BSTR is already a pointer to a string

7 years agomeson: Fix decklink building on Windows
Nirbheek Chauhan [Thu, 4 May 2017 22:31:38 +0000 (04:01 +0530)]
meson: Fix decklink building on Windows

Needs comsuppw, and does not need libdl or pthread.

7 years agodirectsoundsrc: Use latency-time and buffer-time settings
Nirbheek Chauhan [Wed, 3 May 2017 17:20:27 +0000 (22:50 +0530)]
directsoundsrc: Use latency-time and buffer-time settings

Earlier, the plugin was ignoring those settings and blindly setting
buffer-time to 2 seconds and latency-time to 200ms, which forced all
pipelines to have a minimum latency of 200ms + sink latency.

The values of segsize and segtotal were also not derived correctly.
Now we obey these values, and you can get close to the previous
behaviour by setting buffer-time and latency-time manually. Note that
they are set in microseconds.

As a consequence, when we haven't received enough data from the
device, we now sleep for a time proportional to the data remaining.
However, Directsound is a deprecated API so it maintains its own
software ringbuffer which updates at arbitrary intervals. Hence we
might have to wait a full segsize to get the last 10% of data. To
avoid tight loops, we clamp our sleep floor at 10ms.

In my testing, this keeps the wakeups not-too-high (proportional to
the latency-time set on the source). Further improvements should be
made by fixing the WASAPI audio source plugin instead of this.
Directsound is deprecated and as the comments explain, it is
impossible to get low latency, decent quality, or good performance
from it.

Based on a patch by Sebastian Dröge <sebastian@centricular.com>

https://bugzilla.gnome.org/show_bug.cgi?id=781249

7 years agomsdk: use a dummy.cpp file to force c++ linking
Scott D Phillips [Fri, 21 Apr 2017 03:18:30 +0000 (20:18 -0700)]
msdk: use a dummy.cpp file to force c++ linking

Add a dummy.cpp file to force c++ linking as required by libmfx.a.
I'm certain that the meson build worked for me in the past, but I
have no idea how it could have, looking at it now.

https://bugzilla.gnome.org/show_bug.cgi?id=781561