2013-11-05 21:06:48 +00:00
|
|
|
/*
|
|
|
|
* This file is part of mpv.
|
|
|
|
*
|
2016-01-07 10:22:54 +00:00
|
|
|
* mpv is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
|
|
* License as published by the Free Software Foundation; either
|
|
|
|
* version 2.1 of the License, or (at your option) any later version.
|
2013-11-05 21:06:48 +00:00
|
|
|
*
|
|
|
|
* mpv is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
2016-01-07 10:22:54 +00:00
|
|
|
* GNU Lesser General Public License for more details.
|
2013-11-05 21:06:48 +00:00
|
|
|
*
|
2016-01-07 10:22:54 +00:00
|
|
|
* You should have received a copy of the GNU Lesser General Public
|
|
|
|
* License along with mpv. If not, see <http://www.gnu.org/licenses/>.
|
2013-11-05 21:06:48 +00:00
|
|
|
*/
|
|
|
|
|
|
|
|
#include <stddef.h>
|
|
|
|
#include <assert.h>
|
|
|
|
|
vo_opengl: refactor into vo_gpu
This is done in several steps:
1. refactor MPGLContext -> struct ra_ctx
2. move GL-specific stuff in vo_opengl into opengl/context.c
3. generalize context creation to support other APIs, and add --gpu-api
4. rename all of the --opengl- options that are no longer opengl-specific
5. move all of the stuff from opengl/* that isn't GL-specific into gpu/
(note: opengl/gl_utils.h became opengl/utils.h)
6. rename vo_opengl to vo_gpu
7. to handle window screenshots, the short-term approach was to just add
it to ra_swchain_fns. Long term (and for vulkan) this has to be moved to
ra itself (and vo_gpu altered to compensate), but this was a stop-gap
measure to prevent this commit from getting too big
8. move ra->fns->flush to ra_gl_ctx instead
9. some other minor changes that I've probably already forgotten
Note: This is one half of a major refactor, the other half of which is
provided by rossy's following commit. This commit enables support for
all linux platforms, while his version enables support for all non-linux
platforms.
Note 2: vo_opengl_cb.c also re-uses ra_gl_ctx so it benefits from the
--opengl- options like --opengl-early-flush, --opengl-finish etc. Should
be a strict superset of the old functionality.
Disclaimer: Since I have no way of compiling mpv on all platforms, some
of these ports were done blindly. Specifically, the blind ports included
context_mali_fbdev.c and context_rpi.c. Since they're both based on
egl_helpers, the port should have gone smoothly without any major
changes required. But if somebody complains about a compile error on
those platforms (assuming anybody actually uses them), you know where to
complain.
2017-09-14 06:04:55 +00:00
|
|
|
#include "video/out/gpu/hwdec.h"
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
#include "ra_gl.h"
|
2013-11-05 21:06:48 +00:00
|
|
|
#include "video/vdpau.h"
|
2014-04-29 13:11:26 +00:00
|
|
|
#include "video/vdpau_mixer.h"
|
2013-11-05 21:06:48 +00:00
|
|
|
|
2013-11-09 22:31:12 +00:00
|
|
|
// This is a GL_NV_vdpau_interop specification bug, and headers (unfortunately)
|
|
|
|
// follow it. I'm not sure about the original nvidia headers.
|
|
|
|
#define BRAINDEATH(x) ((void *)(uintptr_t)(x))
|
|
|
|
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
struct priv_owner {
|
|
|
|
struct mp_vdpau_ctx *ctx;
|
|
|
|
};
|
2013-11-05 21:06:48 +00:00
|
|
|
|
|
|
|
struct priv {
|
|
|
|
struct mp_vdpau_ctx *ctx;
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
GL *gl;
|
2013-11-05 21:06:48 +00:00
|
|
|
uint64_t preemption_counter;
|
vo_gpu: hwdec_vdpau: remove direct_mode
As we are less and less interested in vpdpau, with nvdec and vaapi
being better choices in general on nvidia and AMD respectively, we
might consider removing direct_mode, where we bypass the vdpau
mixer and work directly with yuv textures. Normally, working with
yuv textures would be great, but vdpau built in an assumption that
all frames are delivered as separate fields, causing us to have
to re-interleave them.
nvidia then introduces a new OpenGL extension that can return the
yuv frames as frames, but we can't just unconditionally switch to
that as we'd want to keep supporting older hardware where the drivers
are no longer getting new features. The end result is that we
wouldn't be able to get rid of the old code paths.
Removing direct_mode means we always use the mixer, and work with
rgba frame textures. There are some theoretical limitations to
this, but in practice they probably don't matter much - unsupported
colourspaces don't matter because without 10bit decoding support,
we can't use them anyway, and apparently we're not doing separate
chroma scaling these days, so scaling the rbga doesn't really lose
anything (and the vdpau hq scaling option remains available).
2019-11-16 23:10:13 +00:00
|
|
|
GLuint gl_texture;
|
2016-02-08 12:39:33 +00:00
|
|
|
bool vdpgl_initialized;
|
2013-11-05 21:06:48 +00:00
|
|
|
GLvdpauSurfaceNV vdpgl_surface;
|
|
|
|
VdpOutputSurface vdp_surface;
|
2014-04-29 13:11:26 +00:00
|
|
|
struct mp_vdpau_mixer *mixer;
|
2019-05-07 04:25:45 +00:00
|
|
|
struct ra_imgfmt_desc direct_desc;
|
2015-03-14 21:36:59 +00:00
|
|
|
bool mapped;
|
2013-11-05 21:06:48 +00:00
|
|
|
};
|
|
|
|
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
static int init(struct ra_hwdec *hw)
|
|
|
|
{
|
2022-03-12 18:54:23 +00:00
|
|
|
struct ra *ra = hw->ra_ctx->ra;
|
|
|
|
Display *x11disp = ra_get_native_resource(ra, "x11");
|
|
|
|
if (!x11disp || !ra_is_gl(ra))
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
return -1;
|
2022-03-12 18:54:23 +00:00
|
|
|
GL *gl = ra_gl_get(ra);
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
if (!(gl->mpgl_caps & MPGL_CAP_VDPAU))
|
|
|
|
return -1;
|
|
|
|
struct priv_owner *p = hw->priv;
|
|
|
|
p->ctx = mp_vdpau_create_device_x11(hw->log, x11disp, true);
|
|
|
|
if (!p->ctx)
|
|
|
|
return -1;
|
|
|
|
if (mp_vdpau_handle_preemption(p->ctx, NULL) < 1)
|
|
|
|
return -1;
|
|
|
|
if (hw->probing && mp_vdpau_guess_if_emulated(p->ctx))
|
|
|
|
return -1;
|
|
|
|
p->ctx->hwctx.driver_name = hw->driver->name;
|
2022-03-24 21:18:59 +00:00
|
|
|
p->ctx->hwctx.hw_imgfmt = IMGFMT_VDPAU;
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
hwdec_devices_add(hw->devs, &p->ctx->hwctx);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void uninit(struct ra_hwdec *hw)
|
2016-06-08 20:16:32 +00:00
|
|
|
{
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
struct priv_owner *p = hw->priv;
|
|
|
|
|
|
|
|
if (p->ctx)
|
|
|
|
hwdec_devices_remove(hw->devs, &p->ctx->hwctx);
|
|
|
|
mp_vdpau_destroy(p->ctx);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void mapper_unmap(struct ra_hwdec_mapper *mapper)
|
|
|
|
{
|
|
|
|
struct priv *p = mapper->priv;
|
|
|
|
GL *gl = p->gl;
|
2016-06-08 20:16:32 +00:00
|
|
|
|
2017-08-14 10:17:34 +00:00
|
|
|
for (int n = 0; n < 4; n++)
|
|
|
|
ra_tex_free(mapper->ra, &mapper->tex[n]);
|
|
|
|
|
2016-06-19 17:58:40 +00:00
|
|
|
if (p->mapped) {
|
2016-06-08 20:16:32 +00:00
|
|
|
gl->VDPAUUnmapSurfacesNV(1, &p->vdpgl_surface);
|
2016-06-19 17:58:40 +00:00
|
|
|
}
|
2016-06-08 20:16:32 +00:00
|
|
|
p->mapped = false;
|
|
|
|
}
|
|
|
|
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
static void mark_vdpau_objects_uninitialized(struct ra_hwdec_mapper *mapper)
|
2013-11-05 21:06:48 +00:00
|
|
|
{
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
struct priv *p = mapper->priv;
|
2013-11-05 21:06:48 +00:00
|
|
|
|
|
|
|
p->vdp_surface = VDP_INVALID_HANDLE;
|
2015-03-14 21:36:59 +00:00
|
|
|
p->mapped = false;
|
2013-11-05 21:06:48 +00:00
|
|
|
}
|
|
|
|
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
static void mapper_uninit(struct ra_hwdec_mapper *mapper)
|
2013-11-05 21:06:48 +00:00
|
|
|
{
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
struct priv *p = mapper->priv;
|
|
|
|
GL *gl = p->gl;
|
2014-03-19 18:57:08 +00:00
|
|
|
struct vdp_functions *vdp = &p->ctx->vdp;
|
2013-11-05 21:06:48 +00:00
|
|
|
VdpStatus vdp_st;
|
|
|
|
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
assert(!p->mapped);
|
2015-03-14 21:36:59 +00:00
|
|
|
|
2013-11-05 21:06:48 +00:00
|
|
|
if (p->vdpgl_surface)
|
|
|
|
gl->VDPAUUnregisterSurfaceNV(p->vdpgl_surface);
|
|
|
|
p->vdpgl_surface = 0;
|
|
|
|
|
vo_gpu: hwdec_vdpau: remove direct_mode
As we are less and less interested in vpdpau, with nvdec and vaapi
being better choices in general on nvidia and AMD respectively, we
might consider removing direct_mode, where we bypass the vdpau
mixer and work directly with yuv textures. Normally, working with
yuv textures would be great, but vdpau built in an assumption that
all frames are delivered as separate fields, causing us to have
to re-interleave them.
nvidia then introduces a new OpenGL extension that can return the
yuv frames as frames, but we can't just unconditionally switch to
that as we'd want to keep supporting older hardware where the drivers
are no longer getting new features. The end result is that we
wouldn't be able to get rid of the old code paths.
Removing direct_mode means we always use the mixer, and work with
rgba frame textures. There are some theoretical limitations to
this, but in practice they probably don't matter much - unsupported
colourspaces don't matter because without 10bit decoding support,
we can't use them anyway, and apparently we're not doing separate
chroma scaling these days, so scaling the rbga doesn't really lose
anything (and the vdpau hq scaling option remains available).
2019-11-16 23:10:13 +00:00
|
|
|
gl->DeleteTextures(1, &p->gl_texture);
|
2013-11-05 21:06:48 +00:00
|
|
|
|
|
|
|
if (p->vdp_surface != VDP_INVALID_HANDLE) {
|
|
|
|
vdp_st = vdp->output_surface_destroy(p->vdp_surface);
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
CHECK_VDP_WARNING(mapper, "Error when calling vdp_output_surface_destroy");
|
2013-11-05 21:06:48 +00:00
|
|
|
}
|
2014-04-29 13:11:26 +00:00
|
|
|
p->vdp_surface = VDP_INVALID_HANDLE;
|
2013-11-05 21:06:48 +00:00
|
|
|
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
gl_check_error(gl, mapper->log, "Before uninitializing OpenGL interop");
|
2013-11-05 21:06:48 +00:00
|
|
|
|
2016-02-08 12:39:33 +00:00
|
|
|
if (p->vdpgl_initialized)
|
|
|
|
gl->VDPAUFiniNV();
|
2013-11-05 21:06:48 +00:00
|
|
|
|
2016-02-08 12:39:33 +00:00
|
|
|
p->vdpgl_initialized = false;
|
|
|
|
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
gl_check_error(gl, mapper->log, "After uninitializing OpenGL interop");
|
2013-11-05 21:06:48 +00:00
|
|
|
|
2014-04-29 13:11:26 +00:00
|
|
|
mp_vdpau_mixer_destroy(p->mixer);
|
2013-11-05 21:06:48 +00:00
|
|
|
}
|
|
|
|
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
static int mapper_init(struct ra_hwdec_mapper *mapper)
|
2013-11-05 21:06:48 +00:00
|
|
|
{
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
struct priv_owner *p_owner = mapper->owner->priv;
|
|
|
|
struct priv *p = mapper->priv;
|
2013-11-05 21:06:48 +00:00
|
|
|
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
p->gl = ra_gl_get(mapper->ra);
|
|
|
|
p->ctx = p_owner->ctx;
|
|
|
|
|
|
|
|
GL *gl = p->gl;
|
2014-03-19 18:57:08 +00:00
|
|
|
struct vdp_functions *vdp = &p->ctx->vdp;
|
2013-11-05 21:06:48 +00:00
|
|
|
VdpStatus vdp_st;
|
|
|
|
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
p->vdp_surface = VDP_INVALID_HANDLE;
|
|
|
|
p->mixer = mp_vdpau_mixer_create(p->ctx, mapper->log);
|
|
|
|
if (!p->mixer)
|
|
|
|
return -1;
|
2013-11-05 21:06:48 +00:00
|
|
|
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
mapper->dst_params = mapper->src_params;
|
2013-11-05 21:06:48 +00:00
|
|
|
|
2016-01-25 18:40:35 +00:00
|
|
|
if (mp_vdpau_handle_preemption(p->ctx, &p->preemption_counter) < 0)
|
2013-11-05 21:06:48 +00:00
|
|
|
return -1;
|
|
|
|
|
2013-11-09 22:31:12 +00:00
|
|
|
gl->VDPAUInitNV(BRAINDEATH(p->ctx->vdp_device), p->ctx->get_proc_address);
|
2013-11-05 21:06:48 +00:00
|
|
|
|
2016-02-08 12:39:33 +00:00
|
|
|
p->vdpgl_initialized = true;
|
|
|
|
|
vo_gpu: hwdec_vdpau: remove direct_mode
As we are less and less interested in vpdpau, with nvdec and vaapi
being better choices in general on nvidia and AMD respectively, we
might consider removing direct_mode, where we bypass the vdpau
mixer and work directly with yuv textures. Normally, working with
yuv textures would be great, but vdpau built in an assumption that
all frames are delivered as separate fields, causing us to have
to re-interleave them.
nvidia then introduces a new OpenGL extension that can return the
yuv frames as frames, but we can't just unconditionally switch to
that as we'd want to keep supporting older hardware where the drivers
are no longer getting new features. The end result is that we
wouldn't be able to get rid of the old code paths.
Removing direct_mode means we always use the mixer, and work with
rgba frame textures. There are some theoretical limitations to
this, but in practice they probably don't matter much - unsupported
colourspaces don't matter because without 10bit decoding support,
we can't use them anyway, and apparently we're not doing separate
chroma scaling these days, so scaling the rbga doesn't really lose
anything (and the vdpau hq scaling option remains available).
2019-11-16 23:10:13 +00:00
|
|
|
gl->GenTextures(1, &p->gl_texture);
|
|
|
|
|
|
|
|
gl->BindTexture(GL_TEXTURE_2D, p->gl_texture);
|
|
|
|
gl->TexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
|
|
|
|
gl->TexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
|
|
|
|
gl->TexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
|
|
|
|
gl->TexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
|
|
|
|
gl->BindTexture(GL_TEXTURE_2D, 0);
|
|
|
|
|
|
|
|
vdp_st = vdp->output_surface_create(p->ctx->vdp_device,
|
|
|
|
VDP_RGBA_FORMAT_B8G8R8A8,
|
|
|
|
mapper->src_params.w,
|
|
|
|
mapper->src_params.h,
|
|
|
|
&p->vdp_surface);
|
|
|
|
CHECK_VDP_ERROR(mapper, "Error when calling vdp_output_surface_create");
|
|
|
|
|
|
|
|
p->vdpgl_surface = gl->VDPAURegisterOutputSurfaceNV(BRAINDEATH(p->vdp_surface),
|
|
|
|
GL_TEXTURE_2D,
|
|
|
|
1, &p->gl_texture);
|
|
|
|
if (!p->vdpgl_surface)
|
|
|
|
return -1;
|
2013-11-05 21:06:48 +00:00
|
|
|
|
vo_gpu: hwdec_vdpau: remove direct_mode
As we are less and less interested in vpdpau, with nvdec and vaapi
being better choices in general on nvidia and AMD respectively, we
might consider removing direct_mode, where we bypass the vdpau
mixer and work directly with yuv textures. Normally, working with
yuv textures would be great, but vdpau built in an assumption that
all frames are delivered as separate fields, causing us to have
to re-interleave them.
nvidia then introduces a new OpenGL extension that can return the
yuv frames as frames, but we can't just unconditionally switch to
that as we'd want to keep supporting older hardware where the drivers
are no longer getting new features. The end result is that we
wouldn't be able to get rid of the old code paths.
Removing direct_mode means we always use the mixer, and work with
rgba frame textures. There are some theoretical limitations to
this, but in practice they probably don't matter much - unsupported
colourspaces don't matter because without 10bit decoding support,
we can't use them anyway, and apparently we're not doing separate
chroma scaling these days, so scaling the rbga doesn't really lose
anything (and the vdpau hq scaling option remains available).
2019-11-16 23:10:13 +00:00
|
|
|
gl->VDPAUSurfaceAccessNV(p->vdpgl_surface, GL_READ_ONLY);
|
2013-11-05 21:06:48 +00:00
|
|
|
|
vo_gpu: hwdec_vdpau: remove direct_mode
As we are less and less interested in vpdpau, with nvdec and vaapi
being better choices in general on nvidia and AMD respectively, we
might consider removing direct_mode, where we bypass the vdpau
mixer and work directly with yuv textures. Normally, working with
yuv textures would be great, but vdpau built in an assumption that
all frames are delivered as separate fields, causing us to have
to re-interleave them.
nvidia then introduces a new OpenGL extension that can return the
yuv frames as frames, but we can't just unconditionally switch to
that as we'd want to keep supporting older hardware where the drivers
are no longer getting new features. The end result is that we
wouldn't be able to get rid of the old code paths.
Removing direct_mode means we always use the mixer, and work with
rgba frame textures. There are some theoretical limitations to
this, but in practice they probably don't matter much - unsupported
colourspaces don't matter because without 10bit decoding support,
we can't use them anyway, and apparently we're not doing separate
chroma scaling these days, so scaling the rbga doesn't really lose
anything (and the vdpau hq scaling option remains available).
2019-11-16 23:10:13 +00:00
|
|
|
mapper->dst_params.imgfmt = IMGFMT_RGB0;
|
|
|
|
mapper->dst_params.hw_subfmt = 0;
|
2013-11-05 21:06:48 +00:00
|
|
|
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
gl_check_error(gl, mapper->log, "After initializing vdpau OpenGL interop");
|
vo_opengl: refactor how hwdec interop exports textures
Rename gl_hwdec_driver.map_image to map_frame, and let it fill out a
struct gl_hwdec_frame describing the exact texture layout. This gives
more flexibility to what the hwdec interop can export. In particular, it
can export strange component orders/permutations and textures with
padded size. (The latter originating from cropped video.)
The way gl_hwdec_frame works is in the spirit of the rest of the
vo_opengl video processing code, which tends to put as much information
in immediate state (as part of the dataflow), instead of declaring it
globally. To some degree this duplicates the texplane and img_tex
structs, but until we somehow unify those, it's better to give the hwdec
state its own struct. The fact that changing the hwdec struct would
require changes and testing on at least 4 platform/GPU combinations
makes duplicating it almost a requirement to avoid pain later.
Make gl_hwdec_driver.reinit set the new image format and remove the
gl_hwdec.converted_imgfmt field.
Likewise, gl_hwdec.gl_texture_target is replaced with
gl_hwdec_plane.gl_target.
Split out a init_image_desc function from init_format. The latter is not
called in the hwdec case at all anymore. Setting up most of struct
texplane is also completely separate in the hwdec and normal cases.
video.c does not check whether the hwdec "mapped" image format is
supported. This should not really happen anyway, and if it does, the
hwdec interop backend must fail at creation time, so this is not an
issue.
2016-05-10 16:29:10 +00:00
|
|
|
|
2013-11-05 21:06:48 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
static int mapper_map(struct ra_hwdec_mapper *mapper)
|
2013-11-05 21:06:48 +00:00
|
|
|
{
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
struct priv *p = mapper->priv;
|
|
|
|
GL *gl = p->gl;
|
2013-11-05 21:06:48 +00:00
|
|
|
|
2014-05-09 19:49:32 +00:00
|
|
|
int pe = mp_vdpau_handle_preemption(p->ctx, &p->preemption_counter);
|
|
|
|
if (pe < 1) {
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
mark_vdpau_objects_uninitialized(mapper);
|
2014-05-09 19:49:32 +00:00
|
|
|
if (pe < 0)
|
|
|
|
return -1;
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
mapper_uninit(mapper);
|
|
|
|
if (mapper_init(mapper) < 0)
|
2014-05-09 19:49:32 +00:00
|
|
|
return -1;
|
|
|
|
}
|
2013-11-05 21:06:48 +00:00
|
|
|
|
vo_gpu: hwdec_vdpau: remove direct_mode
As we are less and less interested in vpdpau, with nvdec and vaapi
being better choices in general on nvidia and AMD respectively, we
might consider removing direct_mode, where we bypass the vdpau
mixer and work directly with yuv textures. Normally, working with
yuv textures would be great, but vdpau built in an assumption that
all frames are delivered as separate fields, causing us to have
to re-interleave them.
nvidia then introduces a new OpenGL extension that can return the
yuv frames as frames, but we can't just unconditionally switch to
that as we'd want to keep supporting older hardware where the drivers
are no longer getting new features. The end result is that we
wouldn't be able to get rid of the old code paths.
Removing direct_mode means we always use the mixer, and work with
rgba frame textures. There are some theoretical limitations to
this, but in practice they probably don't matter much - unsupported
colourspaces don't matter because without 10bit decoding support,
we can't use them anyway, and apparently we're not doing separate
chroma scaling these days, so scaling the rbga doesn't really lose
anything (and the vdpau hq scaling option remains available).
2019-11-16 23:10:13 +00:00
|
|
|
if (!p->vdpgl_surface)
|
|
|
|
return -1;
|
2016-06-19 17:58:40 +00:00
|
|
|
|
vo_gpu: hwdec_vdpau: remove direct_mode
As we are less and less interested in vpdpau, with nvdec and vaapi
being better choices in general on nvidia and AMD respectively, we
might consider removing direct_mode, where we bypass the vdpau
mixer and work directly with yuv textures. Normally, working with
yuv textures would be great, but vdpau built in an assumption that
all frames are delivered as separate fields, causing us to have
to re-interleave them.
nvidia then introduces a new OpenGL extension that can return the
yuv frames as frames, but we can't just unconditionally switch to
that as we'd want to keep supporting older hardware where the drivers
are no longer getting new features. The end result is that we
wouldn't be able to get rid of the old code paths.
Removing direct_mode means we always use the mixer, and work with
rgba frame textures. There are some theoretical limitations to
this, but in practice they probably don't matter much - unsupported
colourspaces don't matter because without 10bit decoding support,
we can't use them anyway, and apparently we're not doing separate
chroma scaling these days, so scaling the rbga doesn't really lose
anything (and the vdpau hq scaling option remains available).
2019-11-16 23:10:13 +00:00
|
|
|
mp_vdpau_mixer_render(p->mixer, NULL, p->vdp_surface, NULL, mapper->src,
|
|
|
|
NULL);
|
2016-06-19 17:58:40 +00:00
|
|
|
|
vo_gpu: hwdec_vdpau: remove direct_mode
As we are less and less interested in vpdpau, with nvdec and vaapi
being better choices in general on nvidia and AMD respectively, we
might consider removing direct_mode, where we bypass the vdpau
mixer and work directly with yuv textures. Normally, working with
yuv textures would be great, but vdpau built in an assumption that
all frames are delivered as separate fields, causing us to have
to re-interleave them.
nvidia then introduces a new OpenGL extension that can return the
yuv frames as frames, but we can't just unconditionally switch to
that as we'd want to keep supporting older hardware where the drivers
are no longer getting new features. The end result is that we
wouldn't be able to get rid of the old code paths.
Removing direct_mode means we always use the mixer, and work with
rgba frame textures. There are some theoretical limitations to
this, but in practice they probably don't matter much - unsupported
colourspaces don't matter because without 10bit decoding support,
we can't use them anyway, and apparently we're not doing separate
chroma scaling these days, so scaling the rbga doesn't really lose
anything (and the vdpau hq scaling option remains available).
2019-11-16 23:10:13 +00:00
|
|
|
gl->VDPAUMapSurfacesNV(1, &p->vdpgl_surface);
|
2016-06-19 17:58:40 +00:00
|
|
|
|
vo_gpu: hwdec_vdpau: remove direct_mode
As we are less and less interested in vpdpau, with nvdec and vaapi
being better choices in general on nvidia and AMD respectively, we
might consider removing direct_mode, where we bypass the vdpau
mixer and work directly with yuv textures. Normally, working with
yuv textures would be great, but vdpau built in an assumption that
all frames are delivered as separate fields, causing us to have
to re-interleave them.
nvidia then introduces a new OpenGL extension that can return the
yuv frames as frames, but we can't just unconditionally switch to
that as we'd want to keep supporting older hardware where the drivers
are no longer getting new features. The end result is that we
wouldn't be able to get rid of the old code paths.
Removing direct_mode means we always use the mixer, and work with
rgba frame textures. There are some theoretical limitations to
this, but in practice they probably don't matter much - unsupported
colourspaces don't matter because without 10bit decoding support,
we can't use them anyway, and apparently we're not doing separate
chroma scaling these days, so scaling the rbga doesn't really lose
anything (and the vdpau hq scaling option remains available).
2019-11-16 23:10:13 +00:00
|
|
|
p->mapped = true;
|
2017-08-14 10:17:34 +00:00
|
|
|
|
vo_gpu: hwdec_vdpau: remove direct_mode
As we are less and less interested in vpdpau, with nvdec and vaapi
being better choices in general on nvidia and AMD respectively, we
might consider removing direct_mode, where we bypass the vdpau
mixer and work directly with yuv textures. Normally, working with
yuv textures would be great, but vdpau built in an assumption that
all frames are delivered as separate fields, causing us to have
to re-interleave them.
nvidia then introduces a new OpenGL extension that can return the
yuv frames as frames, but we can't just unconditionally switch to
that as we'd want to keep supporting older hardware where the drivers
are no longer getting new features. The end result is that we
wouldn't be able to get rid of the old code paths.
Removing direct_mode means we always use the mixer, and work with
rgba frame textures. There are some theoretical limitations to
this, but in practice they probably don't matter much - unsupported
colourspaces don't matter because without 10bit decoding support,
we can't use them anyway, and apparently we're not doing separate
chroma scaling these days, so scaling the rbga doesn't really lose
anything (and the vdpau hq scaling option remains available).
2019-11-16 23:10:13 +00:00
|
|
|
struct ra_tex_params params = {
|
|
|
|
.dimensions = 2,
|
|
|
|
.w = mapper->src_params.w,
|
|
|
|
.h = mapper->src_params.h,
|
|
|
|
.d = 1,
|
|
|
|
.format = ra_find_unorm_format(mapper->ra, 1, 4),
|
|
|
|
.render_src = true,
|
|
|
|
.src_linear = true,
|
|
|
|
};
|
2017-08-14 10:17:34 +00:00
|
|
|
|
vo_gpu: hwdec_vdpau: remove direct_mode
As we are less and less interested in vpdpau, with nvdec and vaapi
being better choices in general on nvidia and AMD respectively, we
might consider removing direct_mode, where we bypass the vdpau
mixer and work directly with yuv textures. Normally, working with
yuv textures would be great, but vdpau built in an assumption that
all frames are delivered as separate fields, causing us to have
to re-interleave them.
nvidia then introduces a new OpenGL extension that can return the
yuv frames as frames, but we can't just unconditionally switch to
that as we'd want to keep supporting older hardware where the drivers
are no longer getting new features. The end result is that we
wouldn't be able to get rid of the old code paths.
Removing direct_mode means we always use the mixer, and work with
rgba frame textures. There are some theoretical limitations to
this, but in practice they probably don't matter much - unsupported
colourspaces don't matter because without 10bit decoding support,
we can't use them anyway, and apparently we're not doing separate
chroma scaling these days, so scaling the rbga doesn't really lose
anything (and the vdpau hq scaling option remains available).
2019-11-16 23:10:13 +00:00
|
|
|
if (!params.format)
|
|
|
|
return -1;
|
2017-08-14 10:17:34 +00:00
|
|
|
|
vo_gpu: hwdec_vdpau: remove direct_mode
As we are less and less interested in vpdpau, with nvdec and vaapi
being better choices in general on nvidia and AMD respectively, we
might consider removing direct_mode, where we bypass the vdpau
mixer and work directly with yuv textures. Normally, working with
yuv textures would be great, but vdpau built in an assumption that
all frames are delivered as separate fields, causing us to have
to re-interleave them.
nvidia then introduces a new OpenGL extension that can return the
yuv frames as frames, but we can't just unconditionally switch to
that as we'd want to keep supporting older hardware where the drivers
are no longer getting new features. The end result is that we
wouldn't be able to get rid of the old code paths.
Removing direct_mode means we always use the mixer, and work with
rgba frame textures. There are some theoretical limitations to
this, but in practice they probably don't matter much - unsupported
colourspaces don't matter because without 10bit decoding support,
we can't use them anyway, and apparently we're not doing separate
chroma scaling these days, so scaling the rbga doesn't really lose
anything (and the vdpau hq scaling option remains available).
2019-11-16 23:10:13 +00:00
|
|
|
mapper->tex[0] =
|
|
|
|
ra_create_wrapped_tex(mapper->ra, ¶ms, p->gl_texture);
|
|
|
|
if (!mapper->tex[0])
|
|
|
|
return -1;
|
2016-06-19 17:58:40 +00:00
|
|
|
|
2013-11-05 21:06:48 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
const struct ra_hwdec_driver ra_hwdec_vdpau = {
|
2019-11-07 21:32:58 +00:00
|
|
|
.name = "vdpau-gl",
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
.priv_size = sizeof(struct priv_owner),
|
|
|
|
.imgfmts = {IMGFMT_VDPAU, 0},
|
2024-08-06 03:24:37 +00:00
|
|
|
.device_type = AV_HWDEVICE_TYPE_VDPAU,
|
vo_opengl: separate hwdec context and mapping, port it to use ra
This does two separate rather intrusive things:
1. Make the hwdec context (which does initialization, provides the
device to the decoder, and other basic state) and frame mapping
(getting textures from a mp_image) separate. This is more
flexible, and you could map multiple images at once. It will
help removing some hwdec special-casing from video.c.
2. Switch all hwdec API use to ra. Of course all code is still
GL specific, but in theory it would be possible to support other
backends. The most important change is that the hwdec interop
returns ra objects, instead of anything GL specific. This removes
the last dependency on GL-specific header files from video.c.
I'm mixing these separate changes because both requires essentially
rewriting all the glue code, so better do them at once. For the same
reason, this change isn't done incrementally.
hwdec_ios.m is untested, since I can't test it. Apart from superficial
mistakes, this also requires dealing with Apple's texture format
fuckups: they force you to use GL_LUMINANCE[_ALPHA] instead of GL_RED
and GL_RG. We also need to report the correct format via ra_tex to
the renderer, which is done by find_la_variant(). It's unknown whether
this works correctly.
hwdec_rpi.c as well as vo_rpi.c are still broken. (I need to pull my
RPI out of a dusty pile of devices and cables, so, later.)
2017-08-10 15:48:33 +00:00
|
|
|
.init = init,
|
|
|
|
.uninit = uninit,
|
|
|
|
.mapper = &(const struct ra_hwdec_mapper_driver){
|
|
|
|
.priv_size = sizeof(struct priv),
|
|
|
|
.init = mapper_init,
|
|
|
|
.uninit = mapper_uninit,
|
|
|
|
.map = mapper_map,
|
|
|
|
.unmap = mapper_unmap,
|
|
|
|
},
|
2013-11-05 21:06:48 +00:00
|
|
|
};
|