vlc.git
3 years agocontrib: d3d9: rename IDL includes variable
Thomas Guillem [Fri, 4 Jan 2019 13:17:58 +0000 (14:17 +0100)]
contrib: d3d9: rename IDL includes variable

3 years agocontrib: d3d11: get more wine
Thomas Guillem [Fri, 4 Jan 2019 10:43:21 +0000 (11:43 +0100)]
contrib: d3d11: get more wine

On Debian Buster, the idl include path is located on /usr/include/wine/wine/

This fixes the "Unable to open include file oaidl.idl" error.

3 years agocontrib: d3d11: rename IDL includes variable
Thomas Guillem [Fri, 4 Jan 2019 10:42:18 +0000 (11:42 +0100)]
contrib: d3d11: rename IDL includes variable

3 years agoFortunes about QT and "hearing" video codecs
Jean-Baptiste Kempf [Thu, 3 Jan 2019 23:27:48 +0000 (00:27 +0100)]
Fortunes about QT and "hearing" video codecs

3 years agoxcb/xv: drop RGB support
Rémi Denis-Courmont [Thu, 3 Jan 2019 18:21:32 +0000 (20:21 +0200)]
xcb/xv: drop RGB support

The RENDER output supports RGB chroma better in all respects. Besides,
I have not seen an X server supporting RGB XVideo formats in a decade.

3 years agoavcodec: fix version checks
Zhao Zhili [Thu, 27 Dec 2018 08:08:44 +0000 (16:08 +0800)]
avcodec: fix version checks

Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>
3 years agobuild: avoid annoying _FORTIFY_SOURCE warnings (fixes #17341)
Janne Grunau [Wed, 2 Jan 2019 11:54:49 +0000 (12:54 +0100)]
build: avoid annoying _FORTIFY_SOURCE warnings (fixes #17341)

fixed patch with the duplicate "#if defined ..." removed

Janne
---8<---
Some toolchains predefine _FORTIFY_SOURCE resulting in countless
_FORTIFY_SOURCE is redefined warnings. Using _FORTIFY_SOURCE without
compiler optimizations also generates warnings.
_FORTIFY_SOURCE is a reserved identifier in C99 ("All identifiers that
begin with an underscore and either an uppercase letter or another
underscore are always reserved for any use.") so the toolchain is
perfectly free to predefine it.

Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>
3 years agoaarch64: Use ".arch armv8-a+feature" instead of ".arch_extension feature"
Martin Storsjö [Fri, 21 Dec 2018 22:13:29 +0000 (00:13 +0200)]
aarch64: Use ".arch armv8-a+feature" instead of ".arch_extension feature"

Clang doesn't support the .arch_extension directive for aarch64 at
the moment, but the same effect can be achieved by using .arch
instead.

Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>
3 years agokva: report mouse position in window coordinates
KO Myung-Hun [Wed, 2 Jan 2019 09:38:22 +0000 (18:38 +0900)]
kva: report mouse position in window coordinates

Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>
3 years agoRevert "demux:avi: use samples_from_vlc_tick for PTS conversion"
Francois Cartegnie [Wed, 2 Jan 2019 22:41:30 +0000 (23:41 +0100)]
Revert "demux:avi: use samples_from_vlc_tick for PTS conversion"

breaks compressed audio / adpcm

This reverts commit 6c3bef5f18bf1647c33fd495367a3192e33bb8ca.

3 years agodisplay: do not require a vout_thread object
Rémi Denis-Courmont [Wed, 2 Jan 2019 10:43:19 +0000 (12:43 +0200)]
display: do not require a vout_thread object

Any object is fine.

3 years agodisplay: move out vout thread event handler
Rémi Denis-Courmont [Wed, 2 Jan 2019 10:35:36 +0000 (12:35 +0200)]
display: move out vout thread event handler

3 years agodisplay: remove vout back-pointer
Rémi Denis-Courmont [Wed, 2 Jan 2019 10:29:54 +0000 (12:29 +0200)]
display: remove vout back-pointer

3 years agodisplay: avoid unnecessary special case
Rémi Denis-Courmont [Wed, 2 Jan 2019 10:27:02 +0000 (12:27 +0200)]
display: avoid unnecessary special case

Always provide a display "owner" structure in caller when creating a
vout display. Also use the owner data as a pointer to the owning video
output thread.

3 years agodisplay: relax invalid pictures memory ordering
Rémi Denis-Courmont [Wed, 2 Jan 2019 10:19:36 +0000 (12:19 +0200)]
display: relax invalid pictures memory ordering

3 years agodisplay: use function rather than event for invalid pictures
Rémi Denis-Courmont [Wed, 2 Jan 2019 10:13:57 +0000 (12:13 +0200)]
display: use function rather than event for invalid pictures

Handling is the same for all (two) cases and does not depend on the
owner.

3 years agodisplay: remove dead code
Rémi Denis-Courmont [Wed, 2 Jan 2019 09:20:12 +0000 (11:20 +0200)]
display: remove dead code

3 years agoposix: add suggested noreturn qualifier
Rémi Denis-Courmont [Wed, 2 Jan 2019 09:15:14 +0000 (11:15 +0200)]
posix: add suggested noreturn qualifier

3 years agokva: fix compilation with some cosmetics
KO Myung-Hun [Mon, 31 Dec 2018 17:25:21 +0000 (02:25 +0900)]
kva: fix compilation with some cosmetics

Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>
3 years agokva: replace deprecated APIs
KO Myung-Hun [Mon, 31 Dec 2018 17:25:20 +0000 (02:25 +0900)]
kva: replace deprecated APIs

Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>
3 years agoskins2: os2: fix compilation
KO Myung-Hun [Mon, 31 Dec 2018 17:25:19 +0000 (02:25 +0900)]
skins2: os2: fix compilation

-----
In file included from gui/skins2/os2/os2_factory.cpp:39:0:
gui/skins2/os2/os2_window.hpp: In member function 'virtual void OS2Window::setOSHandle(vout_window_t*) const':
gui/skins2/os2/os2_window.hpp:65:13: error: invalid use of incomplete type 'struct vout_window_t'
         pWnd->type = VOUT_WINDOW_TYPE_HWND;
             ^~
-----

Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>
3 years agoqt: fix compilation
KO Myung-Hun [Mon, 31 Dec 2018 17:25:18 +0000 (02:25 +0900)]
qt: fix compilation

-----
In file included from gui/qt/menus.cpp:40:0:
gui/qt/main_interface.hpp: At global scope:
gui/qt/main_interface.hpp:75:10: error: 'atomic_flag' in namespace 'std' does not name a type
     std::atomic_flag videoActive;
          ^~~~~~~~~~~
-----

Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>
3 years agoWelcome to 2019
Jean-Baptiste Kempf [Wed, 2 Jan 2019 06:58:14 +0000 (07:58 +0100)]
Welcome to 2019

3 years agodisplay: remove useless lock
Rémi Denis-Courmont [Mon, 31 Dec 2018 16:31:45 +0000 (18:31 +0200)]
display: remove useless lock

3 years agodisplay: remove mouse events
Rémi Denis-Courmont [Sun, 30 Dec 2018 21:57:22 +0000 (23:57 +0200)]
display: remove mouse events

3 years agodisplay: reroute legacy mouse events through window
Rémi Denis-Courmont [Mon, 31 Dec 2018 13:59:58 +0000 (15:59 +0200)]
display: reroute legacy mouse events through window

3 years agomsw: report mouse position in window coordinates
Rémi Denis-Courmont [Mon, 31 Dec 2018 13:52:55 +0000 (15:52 +0200)]
msw: report mouse position in window coordinates

3 years agomacosx: remove write-only parameter
Rémi Denis-Courmont [Mon, 31 Dec 2018 15:33:08 +0000 (17:33 +0200)]
macosx: remove write-only parameter

3 years agodisplay: remove redundant place parameter
Rémi Denis-Courmont [Mon, 31 Dec 2018 15:32:01 +0000 (17:32 +0200)]
display: remove redundant place parameter

3 years agodisplay: remove constant parameter
Rémi Denis-Courmont [Sun, 30 Dec 2018 21:53:20 +0000 (23:53 +0200)]
display: remove constant parameter

3 years agovout: remove vout_ManageDisplay()
Rémi Denis-Courmont [Sun, 30 Dec 2018 19:54:32 +0000 (21:54 +0200)]
vout: remove vout_ManageDisplay()

3 years agoopengl/vdpau: do not assume C and GL ptrdiff_t match
Rémi Denis-Courmont [Mon, 31 Dec 2018 16:44:16 +0000 (18:44 +0200)]
opengl/vdpau: do not assume C and GL ptrdiff_t match

This should fix compilation on some 32-bits platforms.

3 years agodisplay: process fullscreen event synchronously
Rémi Denis-Courmont [Mon, 31 Dec 2018 16:34:23 +0000 (18:34 +0200)]
display: process fullscreen event synchronously

3 years agodisplay: do not use is_fullscreen at start
Rémi Denis-Courmont [Sun, 30 Dec 2018 19:38:18 +0000 (21:38 +0200)]
display: do not use is_fullscreen at start

Just send the legacy fullscreen event right after probing if necessary.

3 years agodisplay: process window state synchronously
Rémi Denis-Courmont [Sun, 30 Dec 2018 19:39:07 +0000 (21:39 +0200)]
display: process window state synchronously

3 years agodisplay: set on-top state immediately
Rémi Denis-Courmont [Sun, 30 Dec 2018 19:38:18 +0000 (21:38 +0200)]
display: set on-top state immediately

Do not postpone until the first manage poll event.

3 years agodisplay: move vout_display_Control() to public header
Rémi Denis-Courmont [Sun, 30 Dec 2018 19:35:44 +0000 (21:35 +0200)]
display: move vout_display_Control() to public header

3 years agovout: process SPU, filter and owner hooks
Rémi Denis-Courmont [Sun, 30 Dec 2018 18:46:54 +0000 (20:46 +0200)]
vout: process SPU, filter and owner hooks

...in the video output code.

3 years agodisplay: reorient mouse pointer position
Rémi Denis-Courmont [Mon, 31 Dec 2018 13:38:50 +0000 (15:38 +0200)]
display: reorient mouse pointer position

Handle non-standard video orientation when translating the cursor
coordinates.

3 years agodisplay: add/use vout_display_TranslateMouseState()
Rémi Denis-Courmont [Sun, 30 Dec 2018 18:40:37 +0000 (20:40 +0200)]
display: add/use vout_display_TranslateMouseState()

3 years agoxcb/window: fix warning
Rémi Denis-Courmont [Sun, 30 Dec 2018 21:57:10 +0000 (23:57 +0200)]
xcb/window: fix warning

3 years agoavcodec: add an assertion
Rémi Denis-Courmont [Sat, 29 Dec 2018 09:59:24 +0000 (11:59 +0200)]
avcodec: add an assertion

3 years agoavcodec: move/balance picture allocation in callbacks
Rémi Denis-Courmont [Sat, 29 Dec 2018 09:56:33 +0000 (11:56 +0200)]
avcodec: move/balance picture allocation in callbacks

Allocate the picture in the DR and HW callbacks that actually "consume"
it. This makes no functional differences to this point.

3 years agovdpau/display: remove pool
Rémi Denis-Courmont [Tue, 25 Dec 2018 19:15:21 +0000 (21:15 +0200)]
vdpau/display: remove pool

3 years agovdpau/chroma: use filter pool
Rémi Denis-Courmont [Tue, 25 Dec 2018 19:15:09 +0000 (21:15 +0200)]
vdpau/chroma: use filter pool

3 years agovdpau/chroma: allocate a picture pool
Rémi Denis-Courmont [Tue, 25 Dec 2018 19:07:22 +0000 (21:07 +0200)]
vdpau/chroma: allocate a picture pool

3 years agodisplay: do not copy opaque pictures
Rémi Denis-Courmont [Sun, 30 Dec 2018 17:08:18 +0000 (19:08 +0200)]
display: do not copy opaque pictures

If the video format is opaque, the useful data is accessed through
picture_t.p_sys. As such, copying a picture into the display pool will
not work.

So do not copy pictures in that case, and instead assume that the
upstream converter or decoder agree with the display on the semantics
of picture_t.p_sys.

3 years agomissing: add noreturn, fix warnings
Rémi Denis-Courmont [Sun, 30 Dec 2018 16:48:54 +0000 (18:48 +0200)]
missing: add noreturn, fix warnings

3 years agodisplay: fix video context pointer aliasing
Rémi Denis-Courmont [Sun, 30 Dec 2018 17:48:17 +0000 (19:48 +0200)]
display: fix video context pointer aliasing

3 years agodisplay: fix initial window size
Rémi Denis-Courmont [Sun, 30 Dec 2018 17:37:41 +0000 (19:37 +0200)]
display: fix initial window size

3 years agomacOS build.sh: Detect local installation of python3
David Fuhrmann [Sat, 29 Dec 2018 17:55:28 +0000 (18:55 +0100)]
macOS build.sh: Detect local installation of python3

Since python3 is now mandatory for meson and co, try to detect
a packaged install and add it to the path.
Note that /usr/local/bin is not added completely, to avoid
interference with homebrew.

3 years agocontrib: srt: Remove downgrade of availability error
David Fuhrmann [Sat, 29 Dec 2018 16:16:33 +0000 (17:16 +0100)]
contrib: srt: Remove downgrade of availability error

This is not needed anymore (and did not really work in some
situations).

3 years agomacOS build.sh: Fix make clean for contribs
David Fuhrmann [Sat, 29 Dec 2018 11:16:52 +0000 (12:16 +0100)]
macOS build.sh: Fix make clean for contribs

If an old contrib folder exists, bootstrap recognizes those as
"system-available" libraries and thus does not recompile.

3 years agodisplay: free up vout_display_t.owner.sys
Rémi Denis-Courmont [Fri, 28 Dec 2018 19:10:36 +0000 (21:10 +0200)]
display: free up vout_display_t.owner.sys

3 years agodisplay: append core private data after vout_display_t
Rémi Denis-Courmont [Fri, 28 Dec 2018 19:08:03 +0000 (21:08 +0200)]
display: append core private data after vout_display_t

3 years agodisplay: inline vout_display_Delete()
Rémi Denis-Courmont [Fri, 28 Dec 2018 19:00:49 +0000 (21:00 +0200)]
display: inline vout_display_Delete()

3 years agodisplay: fix leak on error
Rémi Denis-Courmont [Fri, 28 Dec 2018 19:00:29 +0000 (21:00 +0200)]
display: fix leak on error

3 years agodisplay: inline vout_display_New()
Rémi Denis-Courmont [Fri, 28 Dec 2018 18:55:31 +0000 (20:55 +0200)]
display: inline vout_display_New()

3 years agopicture_pool: handle reserve pool pictures
Rémi Denis-Courmont [Thu, 27 Dec 2018 21:28:25 +0000 (23:28 +0200)]
picture_pool: handle reserve pool pictures

This adds yet another kludge to detect if a picture is from a pool
through another (reserve) pool. This avoids copying pictures allocated
by filters through the "private pool", until displays are fixed not to
rely on pools.

3 years agosplitter: remove write-only owner
Rémi Denis-Courmont [Thu, 27 Dec 2018 17:29:10 +0000 (19:29 +0200)]
splitter: remove write-only owner

3 years agodisplay: move prepare & display helpers
Rémi Denis-Courmont [Thu, 27 Dec 2018 10:23:18 +0000 (12:23 +0200)]
display: move prepare & display helpers

...to public headers.

3 years agowindow: missing VLC_API qualifiers
Rémi Denis-Courmont [Thu, 27 Dec 2018 10:23:01 +0000 (12:23 +0200)]
window: missing VLC_API qualifiers

3 years agovout: do not blend SPU early based on resolution
Rémi Denis-Courmont [Wed, 26 Dec 2018 08:33:05 +0000 (10:33 +0200)]
vout: do not blend SPU early based on resolution

In general, the best SPU quality should be achieved by rendering
directly to the intended display resolution rather than scaling.

This also avoids certain problems:

- If there is no blending for the source chroma, blending fails
  depending on the display resolution. This is a real problem now with
  opaque chromas and "weird" high-depth chromas.
- If the source resolution is really large, rendering to it is a waste
  of processing power.

3 years agopicture_pool: handle non-pool pictures
Rémi Denis-Courmont [Thu, 27 Dec 2018 12:04:53 +0000 (14:04 +0200)]
picture_pool: handle non-pool pictures

This works around picture_pool_OwnsPic() crashing on non-pooled
pictures. The brittle kludge works for the time being because only
pooling and cloning use non-NULL picture_priv_T.gc.opaque.

3 years agosplitter: remove picture deletion callback
Rémi Denis-Courmont [Wed, 26 Dec 2018 18:08:55 +0000 (20:08 +0200)]
splitter: remove picture deletion callback

Use picture_Release() directly.

3 years agosplitter: remove useless pool
Rémi Denis-Courmont [Wed, 26 Dec 2018 17:09:19 +0000 (19:09 +0200)]
splitter: remove useless pool

3 years agosplitter: remove useless struct
Rémi Denis-Courmont [Wed, 26 Dec 2018 17:02:21 +0000 (19:02 +0200)]
splitter: remove useless struct

Also remove xmalloc() usage.

3 years agosplitter: make p_owner a void pointer
Rémi Denis-Courmont [Wed, 26 Dec 2018 16:58:05 +0000 (18:58 +0200)]
splitter: make p_owner a void pointer

3 years agosplitter: make p_sys a void pointer
Rémi Denis-Courmont [Wed, 26 Dec 2018 16:58:05 +0000 (18:58 +0200)]
splitter: make p_sys a void pointer

3 years agovout: drop unnecessary pool check
Rémi Denis-Courmont [Wed, 26 Dec 2018 15:41:24 +0000 (17:41 +0200)]
vout: drop unnecessary pool check

Pictures are now systematically copied into the correct display pool if
necessary, so which pool the decoder allocates pictures from is
irrelevant.

Note though that this did not, and still does not, ensure that pictures
are in the expected format. A buggy decoder could and still can send
pictures in a wrong or old format.

3 years agovout: remove no-op (fix warning)
Rémi Denis-Courmont [Wed, 26 Dec 2018 16:15:03 +0000 (18:15 +0200)]
vout: remove no-op (fix warning)

3 years agosplitter: remove no-op manage function
Rémi Denis-Courmont [Wed, 26 Dec 2018 16:11:10 +0000 (18:11 +0200)]
splitter: remove no-op manage function

This does not do anything any longer.

Refs #21604.

3 years agovout: inline vout_ManageWrapper()
Rémi Denis-Courmont [Wed, 26 Dec 2018 16:08:32 +0000 (18:08 +0200)]
vout: inline vout_ManageWrapper()

3 years agodisplay: remove a constant return value
Rémi Denis-Courmont [Wed, 26 Dec 2018 14:43:05 +0000 (16:43 +0200)]
display: remove a constant return value

3 years agovout: reset "pictures" at right time
Rémi Denis-Courmont [Wed, 26 Dec 2018 14:36:03 +0000 (16:36 +0200)]
vout: reset "pictures" at right time

This sends the pictures reset control to display, and updates the
display conversion filters at the needed time. That is to say after the
display format control requests are processed.

This fixes a race whereby a picture is allocated and rendered with the
old format after the reset request is sent but before it is processed.

Fixes #21580.

3 years agovout: remove picture copy
Rémi Denis-Courmont [Wed, 26 Dec 2018 15:39:37 +0000 (17:39 +0200)]
vout: remove picture copy

This is now taken care of by the display.c conversion code in a more
generic fashion.

3 years agodisplay: copy picture into pool if necessary
Rémi Denis-Courmont [Wed, 26 Dec 2018 15:02:55 +0000 (17:02 +0200)]
display: copy picture into pool if necessary

If the display supports the source format, the converter chain will be
empty, and the "conversion" will be a no-op. Yet, the picture might not
be allocated from the display pool, e.g. because the pool is slow or too
small compared to the DPB size.

Some filters will also do this when they modify the pictures in place.
However conversion filters should not be able to do that.

Copy the picture manually when this corner case happens.

3 years agoes: add video_format_CopyCropAr()
Rémi Denis-Courmont [Wed, 26 Dec 2018 14:59:59 +0000 (16:59 +0200)]
es: add video_format_CopyCropAr()

Export it from vout code.

3 years agokva: zoom, fill change A/R, not display size
Rémi Denis-Courmont [Wed, 26 Dec 2018 14:19:48 +0000 (16:19 +0200)]
kva: zoom, fill change A/R, not display size

Changing the zoom and the fill flag is susceptible to changing the
intended "place" size, not the display size.

3 years agovout: remove useless struct
Rémi Denis-Courmont [Wed, 26 Dec 2018 11:56:39 +0000 (13:56 +0200)]
vout: remove useless struct

3 years agovout: remove transitional use_copy flag
Rémi Denis-Courmont [Wed, 26 Dec 2018 11:54:57 +0000 (13:54 +0200)]
vout: remove transitional use_copy flag

3 years agovout_wrapper: only retain display_pool if needed
Rémi Denis-Courmont [Wed, 26 Dec 2018 11:48:05 +0000 (13:48 +0200)]
vout_wrapper: only retain display_pool if needed

The display pool is only used by the video output core in the corner
case whereby pictures must be copied but not converted.

3 years agovout_wrapper: avoid pool pointer comparisons
Rémi Denis-Courmont [Wed, 26 Dec 2018 11:28:43 +0000 (13:28 +0200)]
vout_wrapper: avoid pool pointer comparisons

3 years agovout: remove the use_dr flag
Rémi Denis-Courmont [Wed, 26 Dec 2018 11:14:41 +0000 (13:14 +0200)]
vout: remove the use_dr flag

The value is no longer used beyond the function that computes it.

3 years agovout: add flag for copy-needing display plugins
Rémi Denis-Courmont [Wed, 26 Dec 2018 11:06:29 +0000 (13:06 +0200)]
vout: add flag for copy-needing display plugins

In the corner case that the display plugin supports the decoder output
format, but the display pool is too small or too slow, copy is needed.

This adds and uses an explicit flag for that case to help revector the
code. The actual picture copy is not changed.

3 years agovout_wrapper: inline NoDrInit()
Rémi Denis-Courmont [Wed, 26 Dec 2018 11:12:46 +0000 (13:12 +0200)]
vout_wrapper: inline NoDrInit()

3 years agovout: do not blend early on DR
Rémi Denis-Courmont [Wed, 26 Dec 2018 08:49:07 +0000 (10:49 +0200)]
vout: do not blend early on DR

Tihs only matters if do_dr_spu, info.is_slow and do_snapshot are all
false. There are then two cases:

1) If is_direct is true, there are no actual differences between early
   and later SPU blending.

2) If is_direct is false, then this will blend SPU into the small
   display pool after copying the picture into it rather than the
   decoder pool before copying the picture. Since both pools have the
   same formats (use_dr is true) and is_slow is false, this does not
   affect the final result or the performance.

3 years agovdpau/chroma: check output format capabilities
Rémi Denis-Courmont [Tue, 25 Dec 2018 18:41:32 +0000 (20:41 +0200)]
vdpau/chroma: check output format capabilities

3 years agovdpau/chroma: do not get early picture for context
Rémi Denis-Courmont [Fri, 21 Dec 2018 22:04:29 +0000 (00:04 +0200)]
vdpau/chroma: do not get early picture for context

3 years agovdpau/chroma: separate filter private data types
Rémi Denis-Courmont [Tue, 25 Dec 2018 18:25:26 +0000 (20:25 +0200)]
vdpau/chroma: separate filter private data types

Use one type for the render and another (tiny) one for the YUV surface
GPU->CPU transfer.

3 years agovdpau/chroma: use vlc_obj_malloc()
Rémi Denis-Courmont [Tue, 25 Dec 2018 18:12:56 +0000 (20:12 +0200)]
vdpau/chroma: use vlc_obj_malloc()

3 years agovdpau/display: fix use-after-free race (refs #21580)
Rémi Denis-Courmont [Tue, 25 Dec 2018 19:21:45 +0000 (21:21 +0200)]
vdpau/display: fix use-after-free race (refs #21580)

3 years agovdpau: factor pool creation
Rémi Denis-Courmont [Mon, 24 Dec 2018 14:40:05 +0000 (16:40 +0200)]
vdpau: factor pool creation

3 years agoopengl/vdpau: remove struct priv
Rémi Denis-Courmont [Mon, 24 Dec 2018 14:28:22 +0000 (16:28 +0200)]
opengl/vdpau: remove struct priv

3 years agoopengl/vdpau: avoid ugly cast
Rémi Denis-Courmont [Mon, 24 Dec 2018 17:19:14 +0000 (19:19 +0200)]
opengl/vdpau: avoid ugly cast

3 years agoopengl: remove redundant VdpDevice
Rémi Denis-Courmont [Mon, 24 Dec 2018 14:24:57 +0000 (16:24 +0200)]
opengl: remove redundant VdpDevice

3 years agovdpau: factor output surface allocation
Rémi Denis-Courmont [Mon, 24 Dec 2018 14:18:09 +0000 (16:18 +0200)]
vdpau: factor output surface allocation

3 years agoQt: fix PCM audio description
Francois Cartegnie [Mon, 24 Dec 2018 11:46:18 +0000 (12:46 +0100)]
Qt: fix PCM audio description

3 years agoQt: add AV1 in profiles codecs
Francois Cartegnie [Mon, 24 Dec 2018 11:40:43 +0000 (12:40 +0100)]
Qt: add AV1 in profiles codecs

3 years agocodec: jpeg: set row_pointers on context
Francois Cartegnie [Mon, 24 Dec 2018 11:05:13 +0000 (12:05 +0100)]
codec: jpeg: set row_pointers on context

fixes potential invalid deref on jpeg error/longjmp