On Tue, 4 Jul 2017 22:18:21 +0300
Martin Storsjö wrote:
> If using the winstore compat library, a fallback LoadLibrary
> function does exist, that only calls LoadPackagedLibrary though
> (which doesn't work for dynamically loading d3d11 DLLs).
>
> Therefore explicitly check
On Thu, 29 Jun 2017 17:28:51 -0400
Vittorio Giovara wrote:
> From: Anton Khirnov
>
> The new API is more extensible and allows for custom layouts.
> More accurate information is exported, eg for decoders that do not
> set a channel layout, lavc
On Thu, 29 Jun 2017 14:44:52 -0400
Vittorio Giovara <vittorio.giov...@gmail.com> wrote:
> On Thu, Jun 29, 2017 at 5:18 AM, wm4 <nfx...@googlemail.com> wrote:
> > On Wed, 28 Jun 2017 18:10:56 -0400
> > Vittorio Giovara <vittorio.giov...@gmail.com> wrote:
> >
On Thu, 29 Jun 2017 10:25:09 +0100
Mark Thompson wrote:
> This does actually work already by magic :)
>
> Both NV12 and P010 surfaces become UNORM R and RG planes, just with a
> different size of sample underneath. Use in OpenCL then sees them
> identically as planes of
On Wed, 28 Jun 2017 18:11:02 -0400
Vittorio Giovara wrote:
> Resampling or conversion to/from ambisonic audio are currently
> unsupported features.
>
> Signed-off-by: Vittorio Giovara
> ---
> libavresample/utils.c | 8
> 1 file
On Wed, 28 Jun 2017 18:11:01 -0400
Vittorio Giovara wrote:
> Signed-off-by: Vittorio Giovara
> ---
> libavutil/channel_layout.c | 86
> --
> libavutil/channel_layout.h | 33 ++
>
On Wed, 28 Jun 2017 18:10:56 -0400
Vittorio Giovara wrote:
> Since the request_channel_layout is used only by a handful of codecs,
> move the option to codec private contexts.
Not sure if that is justified...
___
On Wed, 28 Jun 2017 18:10:45 -0400
Vittorio Giovara wrote:
> From: Anton Khirnov
>
> The new API is more extensible and allows for custom layouts.
> More accurate information is exported, eg for decoders that do not
> set a channel layout, lavc
On Wed, 28 Jun 2017 13:51:26 +0100
Mark Thompson <s...@jkqxz.net> wrote:
> On 28/06/17 12:13, wm4 wrote:
> > On Tue, 27 Jun 2017 22:50:54 +0100
> > Mark Thompson <s...@jkqxz.net> wrote:
> >
> >> ---
> >> libavutil/hwcontext_d3d11va.c | 2
On Wed, 28 Jun 2017 13:49:28 +0100
Mark Thompson <s...@jkqxz.net> wrote:
> On 28/06/17 12:09, wm4 wrote:
> > On Tue, 27 Jun 2017 22:50:45 +0100
> > Mark Thompson <s...@jkqxz.net> wrote:
> >
> >> Supports all surface formats in common
On Wed, 28 Jun 2017 13:36:30 +0100
Mark Thompson <s...@jkqxz.net> wrote:
> On 28/06/17 12:03, wm4 wrote:
> > On Tue, 27 Jun 2017 22:50:44 +0100
> > Mark Thompson <s...@jkqxz.net> wrote:
> >
> >> ---
> >> configure |
On Tue, 27 Jun 2017 22:50:54 +0100
Mark Thompson wrote:
> ---
> libavutil/hwcontext_d3d11va.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/libavutil/hwcontext_d3d11va.c b/libavutil/hwcontext_d3d11va.c
> index 75f78d866..543f90d6c 100644
> ---
On Tue, 27 Jun 2017 22:50:49 +0100
Mark Thompson wrote:
> ---
> libavfilter/Makefile | 6 +
> libavfilter/opencl.c | 285
> +++
> libavfilter/opencl.h | 74 +++
> libavfilter/opencl/rgbyuv.cl | 117
On Tue, 27 Jun 2017 22:50:45 +0100
Mark Thompson wrote:
> Supports all surface formats in common between the two.
> ---
> configure | 6 +
> libavutil/hwcontext_internal.h | 3 +
> libavutil/hwcontext_opencl.c | 298
>
On Tue, 27 Jun 2017 22:50:44 +0100
Mark Thompson wrote:
> ---
> configure |5 +-
> doc/APIchanges |4 +
> libavutil/Makefile |2 +
> libavutil/hwcontext.c |4 +
> libavutil/hwcontext.h |1 +
>
On Wed, 28 Jun 2017 11:37:57 +0200
Luca Barbato <lu_z...@gentoo.org> wrote:
> On 6/28/17 11:36 AM, wm4 wrote:
> > Or maybe we should get back to your old nested arrays, but I'm worried
> > about excessive memory usage.
>
> The alternative is malloc it with all
On Sun, 18 Jun 2017 19:08:02 +0100
Mark Thompson wrote:
> ---
> The intent of this is to have a common structure which can be used in all
> cases where DRM objects need to be shared between components. It would be
> helpful if anyone familiar with specific drivers or use-cases
On Sun, 25 Jun 2017 14:50:32 -0400
Vittorio Giovara <vittorio.giov...@gmail.com> wrote:
> On Sat, Jun 24, 2017 at 12:18 PM, wm4 <nfx...@googlemail.com> wrote:
> > On Sat, 24 Jun 2017 15:51:20 +0200
> > Luca Barbato <lu_z...@gentoo.org> wrote:
> >
>
On Sat, 24 Jun 2017 15:51:20 +0200
Luca Barbato <lu_z...@gentoo.org> wrote:
> From: Anton Khirnov <an...@khirnov.net>
>
> ---
>
> Better subject welcome. It is currently folded in wm4 mix fixes set.
>
> libavcodec/decode.c | 8 ++--
> 1 file c
On Thu, 22 Jun 2017 16:02:10 +0300 (EEST)
Martin Storsjö <mar...@martin.st> wrote:
> On Thu, 22 Jun 2017, wm4 wrote:
>
> > Basically copied from VLC (LGPL):
> >
> > http://git.videolan.org/?p=vlc.git;a=blob;f=modules/video_output/win32/direct3d11.c;h=e9fcb83dcabfe
Helpful for debugging.
---
libavcodec/dxva2.c | 47 +++
1 file changed, 47 insertions(+)
diff --git a/libavcodec/dxva2.c b/libavcodec/dxva2.c
index 79b77150d0..a60604a84f 100644
--- a/libavcodec/dxva2.c
+++ b/libavcodec/dxva2.c
@@ -196,12 +196,59 @@
Affects in particular DXVA2/D3D11VA, which asks for 128 pixels
alignment. The AVHWFramesContext will return AVFrames with that larger
size set. We need to crop it back in the decoder.
---
libavcodec/hevc_refs.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/libavcodec/hevc_refs.c
Some existed since forever, some are new.
The cast in get_surface() is silly, but unless we change the av_log
function signature, or all callers of ff_dxva2_get_surface_index(), it's
needed to remove the const warning.
---
libavcodec/dxva2.c | 16 +++-
1 file changed, 7
It appears in this case, frames_ininit is called twice (once by
av_hwframe_ctx_init(), and again by unreffing the frames ctx ref).
---
libavutil/hwcontext_d3d11va.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/libavutil/hwcontext_d3d11va.c b/libavutil/hwcontext_d3d11va.c
index
Some devices (some phones, apparently) will support only this opaque
format. Of course this won't work with CLI, because copying data
directly is not supported.
Automatic frame allocation (setting AVCodecContext.hw_device_ctx) does
not support this mode, even if it's the only supported mode. But
Make supported codec profiles part of each dxva_modes entry. Every DXVA2
mode is representative for a codec with a subset of supported profiles,
so reflecting that in dxva_modes seems appropriate.
In practice, this will more strictly check MPEG2 profiles, will stop
relying on the surface format
Makes dealing with formats that can not be used for staging textures
easier (DXGI_FORMAT_420_OPAQUE). It also saves memory if the staging
texture is never needed, so this is a good thing.
---
libavutil/hwcontext_d3d11va.c | 46 ---
1 file changed, 34
Basically copied from VLC (LGPL):
http://git.videolan.org/?p=vlc.git;a=blob;f=modules/video_output/win32/direct3d11.c;h=e9fcb83dcabfe778f26e63d19f218caf06a7c3ae;hb=HEAD#l1482
On Mon, 19 Jun 2017 11:56:04 +0200
Luca Barbato <lu_z...@gentoo.org> wrote:
> On 6/9/17 4:27 PM, wm4 wrote:
>
> > #include "avassert.h"
> > #include "common.h"
> > #include "hwcontext.h"
> > @@ -446,6 +452,14 @@ static int d3d
s
> in the DLLs instead of trying to load them dynamically at runtime.
> ---
> Requiring both d3d11.dll and dxgi.dll now, simplifying it slightly.
>
> Moved the configure check as suggested by Diego, and renamed the
> createD3D function as suggested by wm4.
>
> Still untest
On Tue, 13 Jun 2017 15:59:00 +0300 (EEST)
Martin Storsjö <mar...@martin.st> wrote:
> On Tue, 13 Jun 2017, wm4 wrote:
>
> > On Tue, 13 Jun 2017 15:31:12 +0300
> > Martin Storsjö <mar...@martin.st> wrote:
> >
> >> +static AVOnce functio
On Tue, 13 Jun 2017 15:31:12 +0300
Martin Storsjö wrote:
> When targeting the UWP API subset, the LoadLibrary function is not
> available (and the fallback, LoadPackagedLibrary, can't be used to
> load system DLLs). In these cases, link directly to the functions
> in the DLLs
On Tue, 13 Jun 2017 12:33:07 +0200
Diego Biurrun wrote:
> ---
> libavcodec/mmaldec.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/libavcodec/mmaldec.c b/libavcodec/mmaldec.c
> index e64ddba..023ebe8 100644
> --- a/libavcodec/mmaldec.c
> +++ b/libavcodec/mmaldec.c
On Tue, 13 Jun 2017 10:59:31 +0200
Diego Biurrun wrote:
> On Mon, Jun 12, 2017 at 04:06:40PM +0200, Hendrik Leppkes wrote:
> > On Mon, Jun 12, 2017 at 3:29 PM, Diego Biurrun wrote:
> > > On Sun, Jun 11, 2017 at 11:49:45PM +0200, Hendrik Leppkes wrote:
> >
On Sat, 10 Jun 2017 23:44:32 +0200
Hendrik Leppkes <h.lepp...@gmail.com> wrote:
> On Sat, Jun 10, 2017 at 3:58 PM, Diego Biurrun <di...@biurrun.de> wrote:
> > On Fri, Jun 09, 2017 at 04:27:35PM +0200, wm4 wrote:
> >> The cast in get_surface() is silly, b
On Fri, 9 Jun 2017 17:26:22 +0200
Luca Barbato <lu_z...@gentoo.org> wrote:
> On 6/9/17 4:27 PM, wm4 wrote:
> > It appears in this case, frames_ininit is called twice (once by
> > av_hwframe_ctx_init(), and again by unreffing the frames ctx ref).
> > ---
>
Helpful for debugging.
Should I extend this (maybe with nice names in the dxva_modes[] list and
matching/printing them), or drop it?
---
libavcodec/dxva2.c | 47 +++
1 file changed, 47 insertions(+)
diff --git a/libavcodec/dxva2.c b/libavcodec/dxva2.c
Affects in particular DXVA2/D3D11VA, which asks for 128 pixels
alignment. The AVHWFramesContext will return AVFrames with that larger
size set. We need to crop it back in the decoder.
---
libavcodec/hevc_refs.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/libavcodec/hevc_refs.c
Some devices (some phones, apparently) will support only this opaque
format. Of course this won't work with CLI, because copying data
directly is not supported.
Automatic frame allocation (setting AVCodecContext.hw_device_ctx) does
not support this mode, even if it's the only supported mode. But
Basically copied from VLC (LGPL):
http://git.videolan.org/?p=vlc.git;a=blob;f=modules/video_output/win32/direct3d11.c;h=e9fcb83dcabfe778f26e63d19f218caf06a7c3ae;hb=HEAD#l1482
It appears in this case, frames_ininit is called twice (once by
av_hwframe_ctx_init(), and again by unreffing the frames ctx ref).
---
Not sure if that's correct. Or if it could happen with certain other
decoders.
---
libavutil/hwcontext_d3d11va.c | 2 ++
1 file changed, 2 insertions(+)
diff
Makes dealing with formats that can not be used for staging textures
easier (DXGI_FORMAT_420_OPAQUE). It also saves memory if the staging
texture is never needed, so this is a good thing.
---
libavutil/hwcontext_d3d11va.c | 46 ---
1 file changed, 34
Some existed since forever, some are new.
The cast in get_surface() is silly, but unless we change the av_log
function signature, or all callers of ff_dxva2_get_surface_index(), it's
needed to remove the const warning.
---
libavcodec/dxva2.c | 16 +++-
1 file changed, 7
Make supported codec profiles part of each dxva_modes entry. Every DXVA2
mode is representative for a codec with a subset of supported profiles,
so reflecting that in dxva_modes seems appropriate.
In practice, this will more strictly check MPEG2 profiles, will stop
relying on the surface format
wm4 (8):
dxva: add declarative profile checks
dxva: fix some warnings
hwcontext_d3d11va: fix crash on frames_init failure
hwcontext_d3d11va: allocate staging texture lazily
dxva: support DXGI_FORMAT_420_OPAQUE decoding
hwcontext_d3d11va: add option to enable debug mode
hevcdec: fix
On Fri, 9 Jun 2017 12:07:59 +0300
Martin Storsjö wrote:
> When targeting the UWP API subset, the LoadLibrary function is not
> available (and the fallback, LoadPackagedLibrary, can't be used to
> load system DLLs). In these cases, link directly to the functions
> in the DLLs
On Wed, 7 Jun 2017 14:03:30 -0400
Vittorio Giovara <vittorio.giov...@gmail.com> wrote:
> On Wed, Jun 7, 2017 at 12:26 PM, wm4 <nfx...@googlemail.com> wrote:
> > On Wed, 7 Jun 2017 11:46:17 -0400
> > Vittorio Giovara <vittorio.giov...@gmail.com> wrote:
&g
On Wed, 7 Jun 2017 11:46:17 -0400
Vittorio Giovara wrote:
> Hello,
> assuming the previous conversions are ok, this chunk starts
> modifying AVCodecParameters, in particular affecting libavformat.
> The first 10 or so containers are modified to support the new
On Wed, 7 Jun 2017 16:58:58 +0200
Diego Biurrun <di...@biurrun.de> wrote:
> On Wed, Jun 07, 2017 at 04:50:43PM +0200, wm4 wrote:
> > On Wed, 7 Jun 2017 16:41:21 +0200 Diego Biurrun <di...@biurrun.de> wrote:
> >
> > > On Tue, Jun 06,
On Wed, 7 Jun 2017 16:42:32 +0200
Steve Lhomme <rob...@gmail.com> wrote:
> On Wed, Jun 7, 2017 at 12:06 PM, wm4 <nfx...@googlemail.com> wrote:
> > On Wed, 7 Jun 2017 08:59:24 +0200
> > Steve Lhomme <rob...@gmail.com> wrote:
> >
> >> On Tue, Ju
On Wed, 7 Jun 2017 16:41:21 +0200
Diego Biurrun <di...@biurrun.de> wrote:
> On Tue, Jun 06, 2017 at 06:51:11PM +0200, wm4 wrote:
> > --- a/configure
> > +++ b/configure
> > @@ -4673,6 +4682,7 @@ check_builtin stdatomic_h stdatomic.h "atomic_int
> > foo; ato
On Wed, 7 Jun 2017 16:34:46 +0200
Steve Lhomme <rob...@gmail.com> wrote:
> On Wed, Jun 7, 2017 at 12:13 PM, Hendrik Leppkes <h.lepp...@gmail.com> wrote:
> > On Wed, Jun 7, 2017 at 9:10 AM, Steve Lhomme <rob...@gmail.com> wrote:
> >> On Tue, Jun 6, 2017 at
On Wed, 7 Jun 2017 13:28:16 +0200
Diego Biurrun <di...@biurrun.de> wrote:
> On Wed, Jun 07, 2017 at 01:18:19PM +0200, wm4 wrote:
> > On Wed, 7 Jun 2017 13:09:15 +0200
> > Diego Biurrun <di...@biurrun.de> wrote:
> >
> > > On Tue, J
On Wed, 7 Jun 2017 13:09:15 +0200
Diego Biurrun <di...@biurrun.de> wrote:
> On Tue, Jun 06, 2017 at 06:51:07PM +0200, wm4 wrote:
> > --- a/configure
> > +++ b/configure
> > @@ -2166,7 +2166,7 @@ zmbv_encoder_deps="zlib"
> >
> > # hardwa
On Wed, 7 Jun 2017 09:10:17 +0200
Steve Lhomme <rob...@gmail.com> wrote:
> On Tue, Jun 6, 2017 at 6:51 PM, wm4 <nfx...@googlemail.com> wrote:
> > +// (avctx->pix_fmt is not updated yet at this point)
> > +sctx->pix_fmt = avctx->hwaccel->pix_
On Wed, 7 Jun 2017 09:01:07 +0200
Steve Lhomme <rob...@gmail.com> wrote:
> On Tue, Jun 6, 2017 at 6:51 PM, wm4 <nfx...@googlemail.com> wrote:
> > do {
> > +ff_dxva2_lock(avctx);
> > #if CONFIG_D3D11VA
> > -if (ff_dxva2_is_d3d11(avctx))
On Wed, 7 Jun 2017 08:59:24 +0200
Steve Lhomme <rob...@gmail.com> wrote:
> On Tue, Jun 6, 2017 at 6:51 PM, wm4 <nfx...@googlemail.com> wrote:
> > +static int d3d11va_device_create(AVHWDeviceContext *ctx, const char
> > *device,
> > +
This is done by the generic code now. It was not removed earlier to
reduce the diff in the commit adding the new code.
---
avtools/Makefile | 1 -
avtools/avconv.h | 1 -
avtools/avconv_dxva2.c | 440 -
avtools/avconv_opt.c | 4
This also adds support to avconv (which is trivial due to the new
hwaccel API being generic enough). For now, this keeps avconv_dxva2.c as
"dxva2-old", although it doesn't work as avconv.c can't handle multiple
hwaccels with the same pixfmt. It will be removed in a later commit.
The new decoder
So a hwaccel can access avctx->hwaccel in init for whatever reason. This
is for the new d3d hwaccel API. We could create separate entrypoints for
each of the 3 hwaccel types (dxva2, d3d11va, new d3d11va), but this
seems nicer.
---
libavcodec/decode.c | 4 ++--
1 file changed, 2 insertions(+), 2
The actual hwaccel code will need to access an internal context instead
of avctx->hwaccel_context, so add a new DXVA_CONTEXT() macro, that will
dispatch between the "old" external and the new internal context.
Also, the new API requires a new D3D11 pixfmt, so all places which check
for the pixfmt
I want to make it non-mandatory to set a mutex in the D3D11 device
context, and replacing it with user callbacks seems like the best
solution. This is preparation for it. Also makes the code slightly more
readable.
---
libavcodec/dxva2.c | 46 --
1 file
To be used with the new d3d11 hwaccel decode API.
With the new hwaccel API, we don't want surfaces to depend on the
decoder (other than the required dimension and format). The old D3D11VA
pixfmt uses ID3D11VideoDecoderOutputView pointers, which include the
decoder configuration, and thus is
Just resending it again. I'm not aware of any other things to fix.
Although I think I fixed a botched av_freep(), which was probably
introduced due to a review comment saying av_freep() is "safer".
wm4 (6):
lavu: add new D3D11 pixfmt and hwcontext
lavc: set avctx->hwaccel befor
configure stage of a
> build. Hide those details behind a generically-named declaration for the
> TLS protocol to avoid leaking those details into the configuration stage.
> ---
>
> Now keeping the global ff_tls_(de)init functions according to wm4's
> preference.
>
On Wed, 31 May 2017 14:57:54 +0200
Diego Biurrun wrote:
> TLS is currently implemented over either OpenSSL or GnuTLS, with more
> backends likely to appear in the future. Currently, those backend libraries
> are part of the protocol names used during e.g. the configure stage of
On Wed, 31 May 2017 13:21:42 +0300
Martin Storsjö wrote:
> This makes the getaddrinfo functions visible, which aren't normally
> by default on legacy mingw.
>
> We already force __MSVCRT_VERSION__ to an XP version.
> ---
> configure | 2 ++
> 1 file changed, 2 insertions(+)
>
On Tue, 30 May 2017 08:13:38 +0200
Anton Khirnov <an...@khirnov.net> wrote:
> Quoting Luca Barbato (2017-05-29 15:31:34)
> > On 5/29/17 2:43 PM, wm4 wrote:
> > > On Mon, 29 May 2017 13:59:40 +0200
> > > Luca Barbato <lu_z...@gentoo.org> wrote:
> >
On Mon, 29 May 2017 17:53:59 +0200
Steve Lhomme wrote:
> LGTM.
>
> I haven't tested the patches so I'm not sure it works as expected. But
> just looking at the code it seems to do the job.
Thanks for reviewing.
> There doesn't seem to be support for decoding to
>
On Mon, 29 May 2017 13:59:40 +0200
Luca Barbato wrote:
> ---
> avtools/avconv.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/avtools/avconv.c b/avtools/avconv.c
> index 719d289ff9..c30e1953ed 100644
> --- a/avtools/avconv.c
> +++
On Mon, 29 May 2017 12:39:12 +0200
Diego Biurrun <di...@biurrun.de> wrote:
> On Mon, May 29, 2017 at 12:18:22PM +0200, wm4 wrote:
> > On Mon, 29 May 2017 12:03:26 +0200
> > Diego Biurrun <di...@biurrun.de> wrote:
> > > On Mon, May 29, 2017 at 11:32:49AM +0
On Mon, 29 May 2017 12:03:26 +0200
Diego Biurrun <di...@biurrun.de> wrote:
> *PLEASE* do not bottom-post.
>
> On Mon, May 29, 2017 at 11:32:49AM +0200, wm4 wrote:
> > On Mon, 29 May 2017 10:56:36 +0200
> > Diego Biurrun <di...@biurrun.de> wrote:
> &g
On Mon, 29 May 2017 10:56:36 +0200
Diego Biurrun wrote:
> TLS is currently implemented over either OpenSSL or GnuTLS, with more
> backends likely to appear in the future. Currently, those backend libraries
> are part of the protocol names used during e.g. the configure stage of
On Fri, 26 May 2017 19:30:40 +0200
Anton Khirnov wrote:
> Quoting Luca Barbato (2017-05-23 22:08:49)
> > On 5/17/17 7:46 PM, Vittorio Giovara wrote:
> > > +av_channel_layout_uninit(dst);
> > > +return av_channel_layout_copy(dst, channel_layout);
> >
> > Maybe put
On Fri, 26 May 2017 13:56:13 +0200
Diego Biurrun wrote:
> ---
> TLS is not handled like other protocols. Instead the implementation details
> of which crypto library is used get exposed to the user. Hiding those
> details allows simplifying and refactoring some code and
On Fri, 26 May 2017 13:50:11 +0200
Diego Biurrun wrote:
> Both libraries provide similar functionality and cannot be used together.
> When both are enabled one is used and the other ignored arbitrarily. Error
> out instead and have the user choose which library to use.
> ---
>
This also adds support to avconv (which is trivial due to the new
hwaccel API being generic enough). For now, this keeps avconv_dxva2.c as
"dxva2-old", although it doesn't work as avconv.c can't handle multiple
hwaccels with the same pixfmt. It will be removed in a later commit.
The new decoder
So a hwaccel can access avctx->hwaccel in init for whatever reason. This
is for the new d3d hwaccel API. We could create separate entrypoints for
each of the 3 hwaccel types (dxva2, d3d11va, new d3d11va), but this
seems nicer.
---
libavcodec/decode.c | 4 ++--
1 file changed, 2 insertions(+), 2
To be used with the new d3d11 hwaccel decode API.
With the new hwaccel API, we don't want surfaces to depend on the
decoder (other than the required dimension and format). The old D3D11VA
pixfmt uses ID3D11VideoDecoderOutputView pointers, which include the
decoder configuration, and thus is
This is done by the generic code now. It was not removed earlier to
reduce the diff in the commit adding the new code.
---
This also hopefully fixes all aspects of dxva2 in configure.
---
avtools/Makefile | 1 -
avtools/avconv.h | 1 -
avtools/avconv_dxva2.c | 440
I want to make it non-mandatory to set a mutex in the D3D11 device
context, and replacing it with user callbacks seems like the best
solution. This is preparation for it. Also makes the code slightly more
readable.
---
libavcodec/dxva2.c | 46 --
1 file
The actual hwaccel code will need to access an internal context instead
of avctx->hwaccel_context, so add a new DXVA_CONTEXT() macro, that will
dispatch between the "old" external and the new internal context.
Also, the new API requires a new D3D11 pixfmt, so all places which check
for the pixfmt
Hopefully addresses most of the review comments.
wm4 (6):
lavu: add new D3D11 pixfmt and hwcontext
lavc: set avctx->hwaccel before init
dxva: preparations for new hwaccel API
dxva: move d3d11 locking/unlocking to functions
dxva: add support for new dxva2 and d3d11 hwaccel APIs
avc
On Fri, 19 May 2017 11:25:59 -0400
Vittorio Giovara wrote:
> >> > Also, why is this only 1 byte per channel? Seems a little short-sighted.
> >> >
> >>
> >> Probably because even in the long run there won't be more than 255
> >> different channels so 1 byte should
On Thu, 18 May 2017 15:01:54 -0400
Vittorio Giovara <vittorio.giov...@gmail.com> wrote:
> On Thu, May 18, 2017 at 11:16 AM, wm4 <nfx...@googlemail.com> wrote:
> > On Wed, 17 May 2017 13:46:51 -0400
> > Vittorio Giovara <vittorio.giov...@gmail.com> wrot
On Wed, 17 May 2017 13:46:51 -0400
Vittorio Giovara wrote:
> From: Anton Khirnov
>
> The new API is more extensible and allows for custom layouts.
> More accurate information is exported, eg for decoders that do not
> set a channel layout, lavc
On Sat, 13 May 2017 12:09:45 +0200
Anton Khirnov wrote:
> Generally looks good, though I'm not very familiar with this hwaccel.
> Are you planning to deprecate the old-style hwaccels?
I would, but that's not up to me, but to important API users like VLC
and lavfilters.
> > +
On Tue, 9 May 2017 10:34:50 +0100
Mark Thompson wrote:
> > How so? For dxva2, we do use a magic memcpy for uncached memory.
>
> Well, I assume the copy from surface to staging texture is driven by the GPU
> (i.e. like the vaGetImage() case in VAAPI). Since magic memcpy sucks
On Tue, 9 May 2017 08:48:45 +0100
Mark Thompson <s...@jkqxz.net> wrote:
> On 09/05/17 06:07, wm4 wrote:
> > On Mon, 8 May 2017 22:59:11 +0100
> > Mark Thompson <s...@jkqxz.net> wrote:
> >> On 04/05/17 07:44, wm4 wrote:
> >>>
On Mon, 8 May 2017 12:28:45 -0400
Vittorio Giovara wrote:
> Signed-off-by: Vittorio Giovara
> ---
> libavformat/avformat.h | 6 ++
> libavformat/dump.c | 2 ++
> 2 files changed, 8 insertions(+)
>
> diff --git
On Mon, 8 May 2017 12:28:43 -0400
Vittorio Giovara wrote:
> This is currently an unsupported feature, only passthrough is allowed.
>
> Signed-off-by: Vittorio Giovara
> ---
> libavresample/utils.c | 8
> 1 file changed, 8
On Mon, 8 May 2017 12:21:28 -0400
Vittorio Giovara <vittorio.giov...@gmail.com> wrote:
> On Mon, May 8, 2017 at 11:58 AM, wm4 <nfx...@googlemail.com> wrote:
> > It would avoid the requirement for copy and free functions, so if the
> > reordering isn't seen as a big ad
On Mon, 8 May 2017 22:38:55 +0100
Mark Thompson <s...@jkqxz.net> wrote:
> On 04/05/17 07:44, wm4 wrote:
> > Radically rebased, and omits a few in-between commits that are
> > unnecessary for the end result. avconv_dxva2.c should probably
> > also be deleted, but for now
On Mon, 8 May 2017 23:45:24 +0100
Mark Thompson <s...@jkqxz.net> wrote:
> On 04/05/17 07:44, wm4 wrote:
> > This also adds support to avconv (which is trivial due to the new
> > hwaccel API being generic enough). For now, this keeps avconv_dxva2.c as
> > "dxva
On Mon, 8 May 2017 23:24:33 +0100
Mark Thompson wrote:
> > +ff_dxva2_unlock(avctx);
> > av_usleep(2000);
>
> Omgwtf? I think I preferred the illusion that this API was sane, which is
> now forever ruined.
Indeed, this is a sad thing.
On Mon, 8 May 2017 22:59:11 +0100
Mark Thompson <s...@jkqxz.net> wrote:
> On 04/05/17 07:44, wm4 wrote:
> > To be used with the new d3d11 hwaccel decode API.
> >
> > With the new hwaccel API, we don't want surfaces to depend on the
> > decoder (other than
On Mon, 8 May 2017 11:41:57 -0400
Vittorio Giovara <vittorio.giov...@gmail.com> wrote:
> On Fri, May 5, 2017 at 11:13 PM, wm4 <nfx...@googlemail.com> wrote:
> > On Fri, 5 May 2017 22:20:18 -0400
> > Vittorio Giovara <vittorio.giov...@gmail.com> wrote:
&g
This is mostly for convenience, so that API users (including possibly
libavcodec) don't have to duplicate native format mapping tables.
Not sure if these should really have a device_ctx as argument. hw_format
is probably enough, but I wasn't sure whether future implementations
might possibly
On Sat, 6 May 2017 16:21:14 +0200
Luca Barbato <lu_z...@gentoo.org> wrote:
> On 5/6/17 3:46 PM, wm4 wrote:
> > I'd still like to see it.
>
> have a look at github, the set should be there already.
And it can't be posted as a patch here why?
How are we supposed to di
On Sat, 6 May 2017 15:30:35 +0200
Luca Barbato <lu_z...@gentoo.org> wrote:
> On 5/6/17 5:13 AM, wm4 wrote:
> > On Fri, 5 May 2017 22:20:18 -0400
> > Vittorio Giovara <vittorio.giov...@gmail.com> wrote:
> >
> >
> >> +enum AVChannelOrder {
>
101 - 200 of 810 matches
Mail list logo