RepoMirrors
/
mpv
mirror of https://github.com/mpv-player/mpv
1
0
Fork 0
Code Issues Releases Wiki Activity
mpv/video/out/opengl/context_vdpau.c

418 lines
12 KiB
C
Raw Normal View History

vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
/*
* This file is part of mpv.
*
* 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.
*
* 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
* GNU Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with mpv. If not, see <http://www.gnu.org/licenses/>.
*/
#include <X11/Xlib.h>
#include <GL/glx.h>
#include "video/vdpau.h"
#include "video/out/x11_common.h"
#include "context.h"
// 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))
struct surface {
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
int w, h;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
VdpOutputSurface surface;
// This nested shitshow of handles to the same object piss me off.
GLvdpauSurfaceNV registered;
vo_opengl: context_vdpau: minor fixes Don't assume 0 is an invalid object handle. vdpau with its weird API design makes all objects indexes, with 0 being a perfectly valid and common value. You need to use VDP_INVALID_HANDLE, which is not 0. Don't crash if init fails at vdpau initialization. It's because mp_vdpau_destroy(NULL) crashes. Simplify it. Destroy output surface backed FBO before output surface. Also, strictly bookkeep the map/unmap calls (and unmap surfaces before destroying the FBO/texture). I can't see a change in the weird errors when resizing the window, but I guess it's slightly more correct. Add the GL_WRITE_DISCARD_NV symbol to header_fixes.h, because we might fail compilation with headers that do not contain the vdpau extension (well, probably doesn't matter).
2017-03-19 06:41:53 +00:00
bool mapped;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
GLuint texture;
GLuint fbo;
};
struct priv {
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
GL gl;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
GLXContext context;
struct mp_vdpau_ctx *vdp;
VdpPresentationQueueTarget vdp_target;
VdpPresentationQueue vdp_queue;
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
struct surface *surfaces;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
int num_surfaces;
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
int idx_surfaces;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
};
typedef GLXContext (*glXCreateContextAttribsARBProc)
(Display*, GLXFBConfig, GLXContext, Bool, const int*);
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
static bool create_context_x11(struct ra_ctx *ctx)
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
{
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
struct priv *p = ctx->priv;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
struct vo *vo = ctx->vo;
int glx_major, glx_minor;
if (!glXQueryVersion(vo->x11->display, &glx_major, &glx_minor)) {
MP_ERR(vo, "GLX not found.\n");
return false;
}
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
if (!ra_gl_ctx_test_version(ctx, MPGL_VER(glx_major, glx_minor), false))
return false;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
int glx_attribs[] = {
GLX_X_RENDERABLE, True,
GLX_X_VISUAL_TYPE, GLX_TRUE_COLOR,
GLX_RED_SIZE, 1,
GLX_GREEN_SIZE, 1,
GLX_BLUE_SIZE, 1,
GLX_DOUBLEBUFFER, True,
None
};
int fbcount;
GLXFBConfig *fbcs = glXChooseFBConfig(vo->x11->display, vo->x11->screen,
glx_attribs, &fbcount);
if (!fbcs)
return false;
// The list in fbc is sorted (so that the first element is the best).
GLXFBConfig fbc = fbcount > 0 ? fbcs[0] : NULL;
XFree(fbcs);
if (!fbc) {
MP_ERR(vo, "no GLX support present\n");
return false;
}
glXCreateContextAttribsARBProc glXCreateContextAttribsARB =
(glXCreateContextAttribsARBProc)
glXGetProcAddressARB((const GLubyte *)"glXCreateContextAttribsARB");
const char *glxstr =
glXQueryExtensionsString(vo->x11->display, vo->x11->screen);
bool have_ctx_ext = glxstr && !!strstr(glxstr, "GLX_ARB_create_context");
if (!(have_ctx_ext && glXCreateContextAttribsARB)) {
return false;
}
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
int ctx_flags = ctx->opts.debug ? GLX_CONTEXT_DEBUG_BIT_ARB : 0;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
int context_attribs[] = {
GLX_CONTEXT_MAJOR_VERSION_ARB, 4,
GLX_CONTEXT_MINOR_VERSION_ARB, 0,
GLX_CONTEXT_PROFILE_MASK_ARB, GLX_CONTEXT_CORE_PROFILE_BIT_ARB,
GLX_CONTEXT_FLAGS_ARB, ctx_flags,
None
};
GLXContext context = glXCreateContextAttribsARB(vo->x11->display, fbc, 0,
True, context_attribs);
if (!context)
return false;
// Pass 0 as drawable for offscreen use. This is probably (?) not valid in
// standard GLX, but the nVidia drivers accept it.
if (!glXMakeCurrent(vo->x11->display, 0, context)) {
MP_FATAL(vo, "Could not set GLX context!\n");
glXDestroyContext(vo->x11->display, context);
return false;
}
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
p->context = context;
mpgl_load_functions(&p->gl, (void *)glXGetProcAddressARB, glxstr, vo->log);
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
return true;
}
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
static int create_vdpau_objects(struct ra_ctx *ctx)
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
{
struct priv *p = ctx->priv;
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
struct GL *gl = &p->gl;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
VdpDevice dev = p->vdp->vdp_device;
struct vdp_functions *vdp = &p->vdp->vdp;
VdpStatus vdp_st;
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
gl->VDPAUInitNV(BRAINDEATH(dev), p->vdp->get_proc_address);
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
vdp_st = vdp->presentation_queue_target_create_x11(dev, ctx->vo->x11->window,
&p->vdp_target);
CHECK_VDP_ERROR(ctx, "creating vdp target");
vdp_st = vdp->presentation_queue_create(dev, p->vdp_target, &p->vdp_queue);
CHECK_VDP_ERROR(ctx, "creating vdp presentation queue");
return 0;
}
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
static void destroy_vdpau_surface(struct ra_ctx *ctx,
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
struct surface *surface)
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
{
struct priv *p = ctx->priv;
struct vdp_functions *vdp = &p->vdp->vdp;
VdpStatus vdp_st;
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
GL *gl = &p->gl;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
if (surface->mapped)
gl->VDPAUUnmapSurfacesNV(1, &surface->registered);
vo_opengl: context_vdpau: minor fixes Don't assume 0 is an invalid object handle. vdpau with its weird API design makes all objects indexes, with 0 being a perfectly valid and common value. You need to use VDP_INVALID_HANDLE, which is not 0. Don't crash if init fails at vdpau initialization. It's because mp_vdpau_destroy(NULL) crashes. Simplify it. Destroy output surface backed FBO before output surface. Also, strictly bookkeep the map/unmap calls (and unmap surfaces before destroying the FBO/texture). I can't see a change in the weird errors when resizing the window, but I guess it's slightly more correct. Add the GL_WRITE_DISCARD_NV symbol to header_fixes.h, because we might fail compilation with headers that do not contain the vdpau extension (well, probably doesn't matter).
2017-03-19 06:41:53 +00:00
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
gl->DeleteFramebuffers(1, &surface->fbo);
gl->DeleteTextures(1, &surface->texture);
vo_opengl: context_vdpau: minor fixes Don't assume 0 is an invalid object handle. vdpau with its weird API design makes all objects indexes, with 0 being a perfectly valid and common value. You need to use VDP_INVALID_HANDLE, which is not 0. Don't crash if init fails at vdpau initialization. It's because mp_vdpau_destroy(NULL) crashes. Simplify it. Destroy output surface backed FBO before output surface. Also, strictly bookkeep the map/unmap calls (and unmap surfaces before destroying the FBO/texture). I can't see a change in the weird errors when resizing the window, but I guess it's slightly more correct. Add the GL_WRITE_DISCARD_NV symbol to header_fixes.h, because we might fail compilation with headers that do not contain the vdpau extension (well, probably doesn't matter).
2017-03-19 06:41:53 +00:00
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
if (surface->registered)
gl->VDPAUUnregisterSurfaceNV(surface->registered);
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
if (surface->surface != VDP_INVALID_HANDLE) {
vdp_st = vdp->output_surface_destroy(surface->surface);
CHECK_VDP_WARNING(ctx, "destroying vdpau surface");
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
}
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
*surface = (struct surface){
.surface = VDP_INVALID_HANDLE,
};
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
}
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
static bool recreate_vdpau_surface(struct ra_ctx *ctx,
struct surface *surface)
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
{
struct priv *p = ctx->priv;
VdpDevice dev = p->vdp->vdp_device;
struct vdp_functions *vdp = &p->vdp->vdp;
VdpStatus vdp_st;
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
GL *gl = &p->gl;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
destroy_vdpau_surface(ctx, surface);
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
surface->w = ctx->vo->dwidth;
surface->h = ctx->vo->dheight;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
vdp_st = vdp->output_surface_create(dev, VDP_RGBA_FORMAT_B8G8R8A8,
surface->w, surface->h,
&surface->surface);
CHECK_VDP_ERROR_NORETURN(ctx, "creating vdp output surface");
if (vdp_st != VDP_STATUS_OK)
goto error;
gl->GenTextures(1, &surface->texture);
surface->registered =
gl->VDPAURegisterOutputSurfaceNV(BRAINDEATH(surface->surface),
GL_TEXTURE_2D,
1, &surface->texture);
if (!surface->registered) {
MP_ERR(ctx, "could not register vdpau surface with GL\n");
goto error;
}
gl->VDPAUSurfaceAccessNV(surface->registered, GL_WRITE_DISCARD_NV);
gl->VDPAUMapSurfacesNV(1, &surface->registered);
surface->mapped = true;
gl->GenFramebuffers(1, &surface->fbo);
gl->BindFramebuffer(GL_FRAMEBUFFER, surface->fbo);
gl->FramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0,
GL_TEXTURE_2D, surface->texture, 0);
GLenum err = gl->CheckFramebufferStatus(GL_FRAMEBUFFER);
if (err != GL_FRAMEBUFFER_COMPLETE) {
MP_ERR(ctx, "Framebuffer completeness check failed (error=%d).\n",
(int)err);
goto error;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
}
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
gl->BindFramebuffer(GL_FRAMEBUFFER, 0);
gl->VDPAUUnmapSurfacesNV(1, &surface->registered);
surface->mapped = false;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
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
return true;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
error:
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
destroy_vdpau_surface(ctx, surface);
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
return false;
}
static void vdpau_swap_buffers(struct ra_ctx *ctx)
{
struct priv *p = ctx->priv;
struct vdp_functions *vdp = &p->vdp->vdp;
VdpStatus vdp_st;
// This is the *next* surface we will be rendering to. By delaying the
// block_until_idle, we're essentially allowing p->num_surfaces - 1
// in-flight surfaces, plus the one currently visible surface.
struct surface *surf = &p->surfaces[p->idx_surfaces];
if (surf->surface == VDP_INVALID_HANDLE)
return;
VdpTime prev_vsync_time;
vdp_st = vdp->presentation_queue_block_until_surface_idle(p->vdp_queue,
surf->surface,
&prev_vsync_time);
CHECK_VDP_WARNING(ctx, "waiting for surface failed");
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
}
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
static void vdpau_uninit(struct ra_ctx *ctx)
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
{
struct priv *p = ctx->priv;
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
ra_gl_ctx_uninit(ctx);
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
vo_opengl: context_vdpau: minor fixes Don't assume 0 is an invalid object handle. vdpau with its weird API design makes all objects indexes, with 0 being a perfectly valid and common value. You need to use VDP_INVALID_HANDLE, which is not 0. Don't crash if init fails at vdpau initialization. It's because mp_vdpau_destroy(NULL) crashes. Simplify it. Destroy output surface backed FBO before output surface. Also, strictly bookkeep the map/unmap calls (and unmap surfaces before destroying the FBO/texture). I can't see a change in the weird errors when resizing the window, but I guess it's slightly more correct. Add the GL_WRITE_DISCARD_NV symbol to header_fixes.h, because we might fail compilation with headers that do not contain the vdpau extension (well, probably doesn't matter).
2017-03-19 06:41:53 +00:00
if (p->vdp) {
struct vdp_functions *vdp = &p->vdp->vdp;
VdpStatus vdp_st;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
for (int n = 0; n < p->num_surfaces; n++)
destroy_vdpau_surface(ctx, &p->surfaces[n]);
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
vo_opengl: context_vdpau: minor fixes Don't assume 0 is an invalid object handle. vdpau with its weird API design makes all objects indexes, with 0 being a perfectly valid and common value. You need to use VDP_INVALID_HANDLE, which is not 0. Don't crash if init fails at vdpau initialization. It's because mp_vdpau_destroy(NULL) crashes. Simplify it. Destroy output surface backed FBO before output surface. Also, strictly bookkeep the map/unmap calls (and unmap surfaces before destroying the FBO/texture). I can't see a change in the weird errors when resizing the window, but I guess it's slightly more correct. Add the GL_WRITE_DISCARD_NV symbol to header_fixes.h, because we might fail compilation with headers that do not contain the vdpau extension (well, probably doesn't matter).
2017-03-19 06:41:53 +00:00
if (p->vdp_queue != VDP_INVALID_HANDLE) {
vdp_st = vdp->presentation_queue_destroy(p->vdp_queue);
CHECK_VDP_WARNING(ctx, "destroying presentation queue");
}
if (p->vdp_target != VDP_INVALID_HANDLE) {
vdp_st = vdp->presentation_queue_target_destroy(p->vdp_target);
CHECK_VDP_WARNING(ctx, "destroying presentation target");
}
mp_vdpau_destroy(p->vdp);
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
}
if (p->context) {
Display *display = ctx->vo->x11->display;
glXMakeCurrent(display, None, NULL);
glXDestroyContext(display, p->context);
}
vo_x11_uninit(ctx->vo);
}
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
static const struct ra_swapchain_fns vdpau_swapchain;
static bool vdpau_init(struct ra_ctx *ctx)
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
{
struct vo *vo = ctx->vo;
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
struct priv *p = ctx->priv = talloc_zero(ctx, struct priv);
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
vo_opengl: context_vdpau: minor fixes Don't assume 0 is an invalid object handle. vdpau with its weird API design makes all objects indexes, with 0 being a perfectly valid and common value. You need to use VDP_INVALID_HANDLE, which is not 0. Don't crash if init fails at vdpau initialization. It's because mp_vdpau_destroy(NULL) crashes. Simplify it. Destroy output surface backed FBO before output surface. Also, strictly bookkeep the map/unmap calls (and unmap surfaces before destroying the FBO/texture). I can't see a change in the weird errors when resizing the window, but I guess it's slightly more correct. Add the GL_WRITE_DISCARD_NV symbol to header_fixes.h, because we might fail compilation with headers that do not contain the vdpau extension (well, probably doesn't matter).
2017-03-19 06:41:53 +00:00
p->vdp_queue = VDP_INVALID_HANDLE;
p->vdp_target = VDP_INVALID_HANDLE;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
if (ctx->vo->probing)
goto uninit;
if (!vo_x11_init(ctx->vo))
goto uninit;
p->vdp = mp_vdpau_create_device_x11(ctx->log, ctx->vo->x11->display, false);
if (!p->vdp)
goto uninit;
if (!vo_x11_create_vo_window(vo, NULL, "vdpauglx"))
goto uninit;
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
if (!create_context_x11(ctx))
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
goto uninit;
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
if (!(p->gl.mpgl_caps & MPGL_CAP_VDPAU))
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
goto uninit;
if (create_vdpau_objects(ctx) < 0)
goto uninit;
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
p->num_surfaces = ctx->opts.swapchain_depth + 1; // +1 for the visible image
p->surfaces = talloc_zero_array(p, struct surface, p->num_surfaces);
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
for (int n = 0; n < p->num_surfaces; n++)
p->surfaces[n].surface = VDP_INVALID_HANDLE;
vo_opengl: add a backend start_frame callback for context_vdpau Might be useful for other backends too. For context_vdpau, resize handling, presentation, and handling the mapping state becomes somewhat less awkward.
2017-03-20 12:37:47 +00:00
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
struct ra_gl_ctx_params params = {
.swap_buffers = vdpau_swap_buffers,
.external_swapchain = &vdpau_swapchain,
.flipped = true,
};
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
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
if (!ra_gl_ctx_init(ctx, &p->gl, params))
goto uninit;
return true;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
uninit:
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
vdpau_uninit(ctx);
return false;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
}
vo_gpu: simplify structs / names Due to the plethora of historical baggage from different eras getting confusing, I decided to simplify and unify the struct organization and naming scheme. Structs that got renamed: 1. fbodst -> ra_fbo (and moved to gpu/context.h) 2. fbotex -> removed (redundant after 2af2fa7a) 3. fbosurface -> surface 4. img_tex -> image In addition to these structs being renamed, all of the names have been made consistent. The new scheme is as follows: struct image img; struct ra_tex *tex; struct ra_fbo fbo; This also affects derived names, e.g. indirect_fbo -> indirect_tex. Notably also, finish_pass_fbo -> finish_pass_tex and finish_pass_direct -> finish_pass_fbo. The new equivalent of fbotex_change() is called ra_tex_resize(). This commit (should) contain no logic changes, just renaming a bunch of crap.
2017-09-20 08:45:33 +00:00
static bool vdpau_start_frame(struct ra_swapchain *sw, struct ra_fbo *out_fbo)
vo_opengl: add a backend start_frame callback for context_vdpau Might be useful for other backends too. For context_vdpau, resize handling, presentation, and handling the mapping state becomes somewhat less awkward.
2017-03-20 12:37:47 +00:00
{
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
struct priv *p = sw->ctx->priv;
struct vo *vo = sw->ctx->vo;
GL *gl = &p->gl;
struct surface *surf = &p->surfaces[p->idx_surfaces];
if (surf->w != vo->dwidth || surf->h != vo->dheight ||
surf->surface == VDP_INVALID_HANDLE)
{
if (!recreate_vdpau_surface(sw->ctx, surf))
return NULL;
}
vo_opengl: add a backend start_frame callback for context_vdpau Might be useful for other backends too. For context_vdpau, resize handling, presentation, and handling the mapping state becomes somewhat less awkward.
2017-03-20 12:37:47 +00:00
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
assert(!surf->mapped);
gl->VDPAUMapSurfacesNV(1, &surf->registered);
surf->mapped = true;
ra_gl_ctx_resize(sw, surf->w, surf->h, surf->fbo);
vo_gpu: simplify structs / names Due to the plethora of historical baggage from different eras getting confusing, I decided to simplify and unify the struct organization and naming scheme. Structs that got renamed: 1. fbodst -> ra_fbo (and moved to gpu/context.h) 2. fbotex -> removed (redundant after 2af2fa7a) 3. fbosurface -> surface 4. img_tex -> image In addition to these structs being renamed, all of the names have been made consistent. The new scheme is as follows: struct image img; struct ra_tex *tex; struct ra_fbo fbo; This also affects derived names, e.g. indirect_fbo -> indirect_tex. Notably also, finish_pass_fbo -> finish_pass_tex and finish_pass_direct -> finish_pass_fbo. The new equivalent of fbotex_change() is called ra_tex_resize(). This commit (should) contain no logic changes, just renaming a bunch of crap.
2017-09-20 08:45:33 +00:00
return ra_gl_ctx_start_frame(sw, out_fbo);
vo_opengl: add a backend start_frame callback for context_vdpau Might be useful for other backends too. For context_vdpau, resize handling, presentation, and handling the mapping state becomes somewhat less awkward.
2017-03-20 12:37:47 +00:00
}
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
static bool vdpau_submit_frame(struct ra_swapchain *sw,
const struct vo_frame *frame)
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
{
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
struct priv *p = sw->ctx->priv;
GL *gl = &p->gl;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
struct vdp_functions *vdp = &p->vdp->vdp;
VdpStatus vdp_st;
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
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
struct surface *surf = &p->surfaces[p->idx_surfaces];
assert(surf->surface != VDP_INVALID_HANDLE);
assert(surf->mapped);
gl->VDPAUUnmapSurfacesNV(1, &surf->registered);
surf->mapped = false;
vo_opengl: context_vdpau: resize output surfaces lazily This actually fixes the dreaded errors during resizing. It works pretty much like before, except each surface is reallocated before it's used. It implies surfaces with the old size remain in the presentation queue and will be displayed.
2017-03-19 08:06:46 +00:00
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
vdp_st = vdp->presentation_queue_display(p->vdp_queue, surf->surface, 0, 0, 0);
CHECK_VDP_WARNING(sw->ctx, "trying to present vdp surface");
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
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
p->idx_surfaces = (p->idx_surfaces + 1) % p->num_surfaces;
return ra_gl_ctx_submit_frame(sw, frame) && vdp_st == VDP_STATUS_OK;
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
}
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
static bool vdpau_reconfig(struct ra_ctx *ctx)
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
{
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
vo_x11_config_vo_window(ctx->vo);
return true;
}
vo_opengl: add a backend start_frame callback for context_vdpau Might be useful for other backends too. For context_vdpau, resize handling, presentation, and handling the mapping state becomes somewhat less awkward.
2017-03-20 12:37:47 +00:00
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
static int vdpau_control(struct ra_ctx *ctx, int *events, int request, void *arg)
{
return vo_x11_control(ctx->vo, events, request, arg);
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
}
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
static void vdpau_wakeup(struct ra_ctx *ctx)
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
{
vo_x11_wakeup(ctx->vo);
}
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
static void vdpau_wait_events(struct ra_ctx *ctx, int64_t until_time_us)
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
{
vo_x11_wait_events(ctx->vo, until_time_us);
}
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
static const struct ra_swapchain_fns vdpau_swapchain = {
.start_frame = vdpau_start_frame,
.submit_frame = vdpau_submit_frame,
};
const struct ra_ctx_fns ra_ctx_vdpauglx = {
.type = "opengl",
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
.name = "vdpauglx",
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
.reconfig = vdpau_reconfig,
.control = vdpau_control,
.wakeup = vdpau_wakeup,
.wait_events = vdpau_wait_events,
.init = vdpau_init,
.uninit = vdpau_uninit,
vo_opengl: add experimental vdpauglx backend As the manpage says, this has no value other than adding bugs. It uses code based on context_x11.c, and basically does very stripped down context creation (no alpha support etc.). It uses vdpau for display, and maps vdpau output surfaces as FBOs to render into them. This might be good to experiment with asynchronous presentation. For now, it presents synchronously, with a 4 frame delay (which should whack off A/V sync). The forced 4 frame delay is probably also why interaction feels slower. There are some weird vdpau errors on resizing and uninit. No idea what causes them.
2017-03-18 16:43:57 +00:00
};