AuroraMiddleware/gtk2
1
0
Fork 0
forked from AuroraMiddleware/gtk
Code Issues Pull Requests Projects Releases Wiki Activity
3c2484142d
gtk2/gdk/gdkgl.c

428 lines
14 KiB
C
Raw Normal View History

gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
/* GDK - The GIMP Drawing Kit
Correct copyright year
2014-10-12 03:04:40 +00:00
* Copyright (C) 2014 Red Hat, Inc.
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
*
* This library 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 of the License, or (at your option) any later version.
*
* This library 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 this library. If not, see <http://www.gnu.org/licenses/>.
*/
#include "config.h"
#include "gdkcairo.h"
#include "gdkglcontextprivate.h"
#include "gdkinternals.h"
#include <epoxy/gl.h>
#include <math.h>
gdkgl: Texture quads in one giant draw call This requires us to use GL_TRIANGLES and six verts per quad instead of four, which makes me think it might not be worth it on well-optimized GL drivers. However, from talking to some driver developers about it, the GL_TRIANGLES should be faster, since this means that there's one giant contiguous buffer instead of many small buffers. If we were really rendering a lot of quads, I'd use an element buffer and GL_PRIMITIVE_RESTART, but we're really not ever rendering that many quads, and the setup cost for that would just be too annoying.
2014-11-22 18:17:22 +00:00
#include <string.h>
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
static const char *
gdkgl: Rename function This returns the name of the shader type.
2020-10-01 04:24:20 +00:00
get_shader_type_name (int type)
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
{
switch (type)
{
case GL_VERTEX_SHADER:
return "vertex";
case GL_GEOMETRY_SHADER:
return "geometry";
case GL_FRAGMENT_SHADER:
return "fragment";
build: Enable -Wswitch-enum and -Wswitch-default This patch makes that work using 1 of 2 options: 1. Add all missing enums to the switch statement or 2. Cast the switch argument to a uint to avoid having to do that (mostly for GdkEventType). I even found a bug while doing that: clearing a GtkImage with a surface did not notify thae surface property. The reason for enabling this flag even though it is tedious at times is that it is very useful when adding values to an enum, because it makes GTK immediately warn about all the switch statements where this enum is relevant. And I expect changes to enums to be frequent during the GTK4 development cycle.
2017-10-06 19:19:42 +00:00
default:
return "unknown";
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
}
}
gdkgl: Make some helper functions static
2014-11-06 21:47:59 +00:00
static guint
Fix some clang warnings It complains about double const, and it is right.
2015-03-22 18:40:57 +00:00
create_shader (int type,
const char *code)
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
{
guint shader;
int status;
shader = glCreateShader (type);
glShaderSource (shader, 1, &code, NULL);
glCompileShader (shader);
glGetShaderiv (shader, GL_COMPILE_STATUS, &status);
if (status == GL_FALSE)
{
int log_len;
char *buffer;
glGetShaderiv (shader, GL_INFO_LOG_LENGTH, &log_len);
buffer = g_malloc (log_len + 1);
glGetShaderInfoLog (shader, log_len, NULL, buffer);
gdkgl: Rename function This returns the name of the shader type.
2020-10-01 04:24:20 +00:00
g_warning ("Compile failure in %s shader:\n%s", get_shader_type_name (type), buffer);
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
g_free (buffer);
glDeleteShader (shader);
return 0;
}
return shader;
}
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
static void
make_program (GdkGLContextProgram *program,
gdk: Move GLSL shaders into GResource Having the shaders inlined as C strings makes them harder to edit and maintain.
2016-04-22 11:34:33 +00:00
const char *vertex_shader_path,
const char *fragment_shader_path)
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
{
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
guint vertex_shader, fragment_shader;
gdk: Move GLSL shaders into GResource Having the shaders inlined as C strings makes them harder to edit and maintain.
2016-04-22 11:34:33 +00:00
GBytes *source;
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
int status;
gdk: Move GLSL shaders into GResource Having the shaders inlined as C strings makes them harder to edit and maintain.
2016-04-22 11:34:33 +00:00
source = g_resources_lookup_data (vertex_shader_path, 0, NULL);
g_assert (source != NULL);
vertex_shader = create_shader (GL_VERTEX_SHADER, g_bytes_get_data (source, NULL));
g_bytes_unref (source);
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
if (vertex_shader == 0)
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
return;
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
gdk: Move GLSL shaders into GResource Having the shaders inlined as C strings makes them harder to edit and maintain.
2016-04-22 11:34:33 +00:00
source = g_resources_lookup_data (fragment_shader_path, 0, NULL);
g_assert (source != NULL);
fragment_shader = create_shader (GL_FRAGMENT_SHADER, g_bytes_get_data (source, NULL));
g_bytes_unref (source);
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
if (fragment_shader == 0)
{
glDeleteShader (vertex_shader);
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
return;
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
}
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
program->program = glCreateProgram ();
glAttachShader (program->program, vertex_shader);
glAttachShader (program->program, fragment_shader);
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
glLinkProgram (program->program);
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
glDeleteShader (vertex_shader);
glDeleteShader (fragment_shader);
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
glGetProgramiv (program->program, GL_LINK_STATUS, &status);
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
if (status == GL_FALSE)
{
int log_len;
char *buffer;
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
glGetProgramiv (program->program, GL_INFO_LOG_LENGTH, &log_len);
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
buffer = g_malloc (log_len + 1);
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
glGetProgramInfoLog (program->program, log_len, NULL, buffer);
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
g_warning ("Linker failure: %s\n", buffer);
g_free (buffer);
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
glDeleteProgram (program->program);
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
}
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
program->position_location = glGetAttribLocation (program->program, "position");
program->uv_location = glGetAttribLocation (program->program, "uv");
program->map_location = glGetUniformLocation (program->program, "map");
gl: Use a uniform to flip R and B colors on GLES This allows us to decide when the R and B color channels should be flipped with a much better granularity. For instance, when using GLX_EXT_texture_from_pixmap to create a GL texture from a surface we don't need to swap the R and B channels, as the internal representation of the texture data will already have the appropriate colors. We also don't need to flip color channels when blitting from a texture.
2016-04-25 12:38:22 +00:00
program->flip_location = glGetUniformLocation (program->program, "flipColors");
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
}
static void
bind_vao (GdkGLContextPaintData *paint_data)
{
if (paint_data->vertex_array_object == 0)
{
glGenVertexArrays (1, &paint_data->vertex_array_object);
/* ATM we only use one VAO, so always bind it */
glBindVertexArray (paint_data->vertex_array_object);
}
}
gdk: Add GLES shaders We cannot use GL shaders with GLES.
2016-04-22 12:52:49 +00:00
static void
use_texture_gles_program (GdkGLContextPaintData *paint_data)
{
if (paint_data->texture_2d_quad_program.program == 0)
make_program (&paint_data->texture_2d_quad_program,
"/org/gtk/libgdk/glsl/gles2-texture.vs.glsl",
"/org/gtk/libgdk/glsl/gles2-texture.fs.glsl");
if (paint_data->current_program != &paint_data->texture_2d_quad_program)
{
paint_data->current_program = &paint_data->texture_2d_quad_program;
glUseProgram (paint_data->current_program->program);
}
}
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
static void
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
use_texture_2d_program (GdkGLContextPaintData *paint_data)
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
{
gdk: Move GLSL shaders into GResource Having the shaders inlined as C strings makes them harder to edit and maintain.
2016-04-22 11:34:33 +00:00
const char *vertex_shader_path = paint_data->is_legacy
? "/org/gtk/libgdk/glsl/gl2-texture-2d.vs.glsl"
: "/org/gtk/libgdk/glsl/gl3-texture-2d.vs.glsl";
const char *fragment_shader_path = paint_data->is_legacy
? "/org/gtk/libgdk/glsl/gl2-texture-2d.fs.glsl"
: "/org/gtk/libgdk/glsl/gl3-texture-2d.fs.glsl";
gl: Use older GLSL shaders with legacy contexts If we're using modern GLSL, then we should stop using deprecated modifiers, like 'varying' and 'attribute', as well as deprecated global variables, like 'gl_FragColor'. On the other hand, with legacy contexts we should be using older GLSL shaders, to maximize compatibility. https://bugzilla.gnome.org/show_bug.cgi?id=756142
2015-10-06 18:55:43 +00:00
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
if (paint_data->texture_2d_quad_program.program == 0)
gdk: Move GLSL shaders into GResource Having the shaders inlined as C strings makes them harder to edit and maintain.
2016-04-22 11:34:33 +00:00
make_program (&paint_data->texture_2d_quad_program, vertex_shader_path, fragment_shader_path);
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
if (paint_data->current_program != &paint_data->texture_2d_quad_program)
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
{
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
paint_data->current_program = &paint_data->texture_2d_quad_program;
glUseProgram (paint_data->current_program->program);
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
}
}
GdkGL: Fix up GL_ARB_texture_rectangle support This broke when gdk_gl_texture_quad moved to shaders. We need a specialized shader for the rectangle case.
2014-11-06 10:31:01 +00:00
static void
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
use_texture_rect_program (GdkGLContextPaintData *paint_data)
GdkGL: Fix up GL_ARB_texture_rectangle support This broke when gdk_gl_texture_quad moved to shaders. We need a specialized shader for the rectangle case.
2014-11-06 10:31:01 +00:00
{
gdk: Move GLSL shaders into GResource Having the shaders inlined as C strings makes them harder to edit and maintain.
2016-04-22 11:34:33 +00:00
const char *vertex_shader_path = paint_data->is_legacy
? "/org/gtk/libgdk/glsl/gl2-texture-rect.vs.glsl"
: "/org/gtk/libgdk/glsl/gl3-texture-rect.vs.glsl";
const char *fragment_shader_path = paint_data->is_legacy
? "/org/gtk/libgdk/glsl/gl2-texture-rect.fs.glsl"
: "/org/gtk/libgdk/glsl/gl3-texture-rect.vs.glsl";
GdkGL: Fix up GL_ARB_texture_rectangle support This broke when gdk_gl_texture_quad moved to shaders. We need a specialized shader for the rectangle case.
2014-11-06 10:31:01 +00:00
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
if (paint_data->texture_rect_quad_program.program == 0)
gdk: Move GLSL shaders into GResource Having the shaders inlined as C strings makes them harder to edit and maintain.
2016-04-22 11:34:33 +00:00
make_program (&paint_data->texture_rect_quad_program, vertex_shader_path, fragment_shader_path);
GdkGL: Fix up GL_ARB_texture_rectangle support This broke when gdk_gl_texture_quad moved to shaders. We need a specialized shader for the rectangle case.
2014-11-06 10:31:01 +00:00
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
if (paint_data->current_program != &paint_data->texture_rect_quad_program)
GdkGL: Fix up GL_ARB_texture_rectangle support This broke when gdk_gl_texture_quad moved to shaders. We need a specialized shader for the rectangle case.
2014-11-06 10:31:01 +00:00
{
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
paint_data->current_program = &paint_data->texture_rect_quad_program;
glUseProgram (paint_data->current_program->program);
GdkGL: Fix up GL_ARB_texture_rectangle support This broke when gdk_gl_texture_quad moved to shaders. We need a specialized shader for the rectangle case.
2014-11-06 10:31:01 +00:00
}
}
Add gdk_gl_texture_quad() helper Right now this just centralizes the glBegin/glEnd code, but this will be replaced with buffer objects later.
2014-11-05 14:42:06 +00:00
void
gdkgl: Texture many quads at once for performance reasons This isn't fully performant yet. To be fully performant, we'd need to do everything in one giant buffer.
2014-11-22 17:47:35 +00:00
gdk_gl_texture_quads (GdkGLContext *paint_context,
guint texture_target,
int n_quads,
gl: Use a uniform to flip R and B colors on GLES This allows us to decide when the R and B color channels should be flipped with a much better granularity. For instance, when using GLX_EXT_texture_from_pixmap to create a GL texture from a surface we don't need to swap the R and B channels, as the internal representation of the texture data will already have the appropriate colors. We also don't need to flip color channels when blitting from a texture.
2016-04-25 12:38:22 +00:00
GdkTexturedQuad *quads,
gboolean flip_colors)
Add gdk_gl_texture_quad() helper Right now this just centralizes the glBegin/glEnd code, but this will be replaced with buffer objects later.
2014-11-05 14:42:06 +00:00
{
gdk/gdkgl.c: Fix Typo The extra ';' actually broke builds on C89 compilers...
2014-11-14 09:26:29 +00:00
GdkGLContextPaintData *paint_data = gdk_gl_context_get_paint_data (paint_context);
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
GdkGLContextProgram *program;
GdkSurface: Rename lots of stuff from window->surface Mostly these are internal things, but the major public change is that event.window is now event.surface.
2018-03-20 14:14:10 +00:00
GdkSurface *surface = gdk_gl_context_get_surface (paint_context);
int surface_scale = gdk_surface_get_scale_factor (surface);
float w = gdk_surface_get_width (surface) * surface_scale;
float h = gdk_surface_get_height (surface) * surface_scale;
gdkgl: Texture many quads at once for performance reasons This isn't fully performant yet. To be fully performant, we'd need to do everything in one giant buffer.
2014-11-22 17:47:35 +00:00
int i;
gdkgl: Texture quads in one giant draw call This requires us to use GL_TRIANGLES and six verts per quad instead of four, which makes me think it might not be worth it on well-optimized GL drivers. However, from talking to some driver developers about it, the GL_TRIANGLES should be faster, since this means that there's one giant contiguous buffer instead of many small buffers. If we were really rendering a lot of quads, I'd use an element buffer and GL_PRIMITIVE_RESTART, but we're really not ever rendering that many quads, and the setup cost for that would just be too annoying.
2014-11-22 18:17:22 +00:00
float *vertex_buffer_data;
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
bind_vao (paint_data);
if (paint_data->tmp_vertex_buffer == 0)
glGenBuffers(1, &paint_data->tmp_vertex_buffer);
gdk: Add GLES shaders We cannot use GL shaders with GLES.
2016-04-22 12:52:49 +00:00
if (paint_data->use_es)
use_texture_gles_program (paint_data);
GdkGL: Fix up GL_ARB_texture_rectangle support This broke when gdk_gl_texture_quad moved to shaders. We need a specialized shader for the rectangle case.
2014-11-06 10:31:01 +00:00
else
gdk: Add GLES shaders We cannot use GL shaders with GLES.
2016-04-22 12:52:49 +00:00
{
if (texture_target == GL_TEXTURE_RECTANGLE_ARB)
use_texture_rect_program (paint_data);
else
use_texture_2d_program (paint_data);
}
gdkgl: Clean up program-manipulating code Abstract the program ID and locations into a "program struct" which we then use.
2014-11-22 17:29:27 +00:00
program = paint_data->current_program;
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
gl: Use a uniform to flip R and B colors on GLES This allows us to decide when the R and B color channels should be flipped with a much better granularity. For instance, when using GLX_EXT_texture_from_pixmap to create a GL texture from a surface we don't need to swap the R and B channels, as the internal representation of the texture data will already have the appropriate colors. We also don't need to flip color channels when blitting from a texture.
2016-04-25 12:38:22 +00:00
/* Use texture unit 0 */
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
glActiveTexture (GL_TEXTURE0);
gl: Use a uniform to flip R and B colors on GLES This allows us to decide when the R and B color channels should be flipped with a much better granularity. For instance, when using GLX_EXT_texture_from_pixmap to create a GL texture from a surface we don't need to swap the R and B channels, as the internal representation of the texture data will already have the appropriate colors. We also don't need to flip color channels when blitting from a texture.
2016-04-25 12:38:22 +00:00
glUniform1i(program->map_location, 0);
/* Flip 'R' and 'B' colors on GLES, if necessary */
if (gdk_gl_context_get_use_es (paint_context))
glUniform1i (program->flip_location, flip_colors ? 1 : 0);
GdkGL: Fix up GL_ARB_texture_rectangle support This broke when gdk_gl_texture_quad moved to shaders. We need a specialized shader for the rectangle case.
2014-11-06 10:31:01 +00:00
Fix core context creation in GdkGLContext - Specifically request GL version when creating context. Just specifying core profile bit results in the requested version defaulting to 1.0 which causes the core profile bit to be ignored and an arbitrary compatability context to be returned. - Fix GL painting by removing GL calls that have been depricated by the 3.2 core profile. - Additionally remove glInvalidateFramebuffer() call, it is not supported by 3.2 core. https://bugzilla.gnome.org/show_bug.cgi?id=742953
2015-01-15 01:30:28 +00:00
glEnableVertexAttribArray (program->position_location);
glEnableVertexAttribArray (program->uv_location);
gdk_gl_texture_quad: Use shaders to texture things This is the modern way OpenGL works, and using it will let us switch to a core context for the paint context, and work on OpenGL ES 2.0.
2014-11-06 08:13:36 +00:00
glBindBuffer (GL_ARRAY_BUFFER, paint_data->tmp_vertex_buffer);
gdkgl: Use one VBO for both position and uv attributes
2014-11-22 18:12:58 +00:00
glVertexAttribPointer (program->position_location, 2, GL_FLOAT, GL_FALSE, sizeof(float) * 4, NULL);
gdkgl.c: Avoid GCCism in Pointer Arithmetic NULL-plus-something could be seen by the compiler to attempt to do arithmetic with void *, which is a GCCism. Instead, do the math normally and cast the results as a void *. https://bugzilla.gnome.org/show_bug.cgi?id=740605
2014-11-24 03:05:18 +00:00
glVertexAttribPointer (program->uv_location, 2, GL_FLOAT, GL_FALSE, sizeof(float) * 4, (void *) (sizeof(float) * 2));
gdkgl: Texture many quads at once for performance reasons This isn't fully performant yet. To be fully performant, we'd need to do everything in one giant buffer.
2014-11-22 17:47:35 +00:00
gdkgl: Texture quads in one giant draw call This requires us to use GL_TRIANGLES and six verts per quad instead of four, which makes me think it might not be worth it on well-optimized GL drivers. However, from talking to some driver developers about it, the GL_TRIANGLES should be faster, since this means that there's one giant contiguous buffer instead of many small buffers. If we were really rendering a lot of quads, I'd use an element buffer and GL_PRIMITIVE_RESTART, but we're really not ever rendering that many quads, and the setup cost for that would just be too annoying.
2014-11-22 18:17:22 +00:00
#define VERTEX_SIZE 4
#define QUAD_N_VERTICES 6
#define QUAD_SIZE (VERTEX_SIZE * QUAD_N_VERTICES)
vertex_buffer_data = g_new (float, n_quads * QUAD_SIZE);
gdkgl: Texture many quads at once for performance reasons This isn't fully performant yet. To be fully performant, we'd need to do everything in one giant buffer.
2014-11-22 17:47:35 +00:00
for (i = 0; i < n_quads; i++)
{
GdkTexturedQuad *quad = &quads[i];
gdkgl: Texture quads in one giant draw call This requires us to use GL_TRIANGLES and six verts per quad instead of four, which makes me think it might not be worth it on well-optimized GL drivers. However, from talking to some driver developers about it, the GL_TRIANGLES should be faster, since this means that there's one giant contiguous buffer instead of many small buffers. If we were really rendering a lot of quads, I'd use an element buffer and GL_PRIMITIVE_RESTART, but we're really not ever rendering that many quads, and the setup cost for that would just be too annoying.
2014-11-22 18:17:22 +00:00
float vertex_data[] = {
(quad->x1 * 2) / w - 1, (quad->y1 * 2) / h - 1, quad->u1, quad->v1,
(quad->x1 * 2) / w - 1, (quad->y2 * 2) / h - 1, quad->u1, quad->v2,
gdkgl: Use one VBO for both position and uv attributes
2014-11-22 18:12:58 +00:00
(quad->x2 * 2) / w - 1, (quad->y1 * 2) / h - 1, quad->u2, quad->v1,
gdkgl: Texture quads in one giant draw call This requires us to use GL_TRIANGLES and six verts per quad instead of four, which makes me think it might not be worth it on well-optimized GL drivers. However, from talking to some driver developers about it, the GL_TRIANGLES should be faster, since this means that there's one giant contiguous buffer instead of many small buffers. If we were really rendering a lot of quads, I'd use an element buffer and GL_PRIMITIVE_RESTART, but we're really not ever rendering that many quads, and the setup cost for that would just be too annoying.
2014-11-22 18:17:22 +00:00
gdkgl: Use one VBO for both position and uv attributes
2014-11-22 18:12:58 +00:00
(quad->x2 * 2) / w - 1, (quad->y2 * 2) / h - 1, quad->u2, quad->v2,
(quad->x1 * 2) / w - 1, (quad->y2 * 2) / h - 1, quad->u1, quad->v2,
gdkgl: Texture quads in one giant draw call This requires us to use GL_TRIANGLES and six verts per quad instead of four, which makes me think it might not be worth it on well-optimized GL drivers. However, from talking to some driver developers about it, the GL_TRIANGLES should be faster, since this means that there's one giant contiguous buffer instead of many small buffers. If we were really rendering a lot of quads, I'd use an element buffer and GL_PRIMITIVE_RESTART, but we're really not ever rendering that many quads, and the setup cost for that would just be too annoying.
2014-11-22 18:17:22 +00:00
(quad->x2 * 2) / w - 1, (quad->y1 * 2) / h - 1, quad->u2, quad->v1,
gdkgl: Texture many quads at once for performance reasons This isn't fully performant yet. To be fully performant, we'd need to do everything in one giant buffer.
2014-11-22 17:47:35 +00:00
};
gdkgl: Texture quads in one giant draw call This requires us to use GL_TRIANGLES and six verts per quad instead of four, which makes me think it might not be worth it on well-optimized GL drivers. However, from talking to some driver developers about it, the GL_TRIANGLES should be faster, since this means that there's one giant contiguous buffer instead of many small buffers. If we were really rendering a lot of quads, I'd use an element buffer and GL_PRIMITIVE_RESTART, but we're really not ever rendering that many quads, and the setup cost for that would just be too annoying.
2014-11-22 18:17:22 +00:00
float *vertex = &vertex_buffer_data[i * QUAD_SIZE];
memcpy (vertex, vertex_data, sizeof(vertex_data));
gdkgl: Texture many quads at once for performance reasons This isn't fully performant yet. To be fully performant, we'd need to do everything in one giant buffer.
2014-11-22 17:47:35 +00:00
}
gdkgl: Texture quads in one giant draw call This requires us to use GL_TRIANGLES and six verts per quad instead of four, which makes me think it might not be worth it on well-optimized GL drivers. However, from talking to some driver developers about it, the GL_TRIANGLES should be faster, since this means that there's one giant contiguous buffer instead of many small buffers. If we were really rendering a lot of quads, I'd use an element buffer and GL_PRIMITIVE_RESTART, but we're really not ever rendering that many quads, and the setup cost for that would just be too annoying.
2014-11-22 18:17:22 +00:00
glBufferData (GL_ARRAY_BUFFER, sizeof(float) * n_quads * QUAD_SIZE, vertex_buffer_data, GL_STREAM_DRAW);
glDrawArrays (GL_TRIANGLES, 0, n_quads * QUAD_N_VERTICES);
g_free (vertex_buffer_data);
Fix core context creation in GdkGLContext - Specifically request GL version when creating context. Just specifying core profile bit results in the requested version defaulting to 1.0 which causes the core profile bit to be ignored and an arbitrary compatability context to be returned. - Fix GL painting by removing GL calls that have been depricated by the 3.2 core profile. - Additionally remove glInvalidateFramebuffer() call, it is not supported by 3.2 core. https://bugzilla.gnome.org/show_bug.cgi?id=742953
2015-01-15 01:30:28 +00:00
glDisableVertexAttribArray (program->position_location);
glDisableVertexAttribArray (program->uv_location);
Add gdk_gl_texture_quad() helper Right now this just centralizes the glBegin/glEnd code, but this will be replaced with buffer objects later.
2014-11-05 14:42:06 +00:00
}
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
/* x,y,width,height describes a rectangle in the gl render buffer
coordinate space, and its top left corner is drawn at the current
position according to the cairo translation. */
Trivial formatting fixes
2014-10-12 03:17:34 +00:00
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
/**
Trivial formatting fixes
2014-10-12 03:17:34 +00:00
* gdk_cairo_draw_from_gl:
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
* @cr: a cairo context
GdkSurface: Rename lots of stuff from window->surface Mostly these are internal things, but the major public change is that event.window is now event.surface.
2018-03-20 14:14:10 +00:00
* @surface: The surface we're rendering for (not necessarily into)
Trivial formatting fixes
2014-10-12 03:17:34 +00:00
* @source: The GL ID of the source buffer
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
* @source_type: The type of the @source
* @buffer_scale: The scale-factor that the @source buffer is allocated for
* @x: The source x position in @source to start copying from in GL coordinates
* @y: The source y position in @source to start copying from in GL coordinates
* @width: The width of the region to draw
* @height: The height of the region to draw
*
docs: Its just GTK now Another round of + removal.
2019-02-24 18:32:24 +00:00
* This is the main way to draw GL content in GTK. It takes a render buffer ID
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
* (@source_type == #GL_RENDERBUFFER) or a texture id (@source_type == #GL_TEXTURE)
* and draws it onto @cr with an OVER operation, respecting the current clip.
Clarify docs for gdk_cairo_draw_from_gl
2014-10-30 11:22:08 +00:00
* The top left corner of the rectangle specified by @x, @y, @width and @height
* will be drawn at the current (0,0) position of the cairo_t.
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
*
GdkSurface: Rename lots of stuff from window->surface Mostly these are internal things, but the major public change is that event.window is now event.surface.
2018-03-20 14:14:10 +00:00
* This will work for *all* cairo_t, as long as @surface is realized, but the
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
* fallback implementation that reads back the pixels from the buffer may be
GdkSurface: Rename lots of stuff from window->surface Mostly these are internal things, but the major public change is that event.window is now event.surface.
2018-03-20 14:14:10 +00:00
* used in the general case. In the case of direct drawing to a surface with
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
* no special effects applied to @cr it will however use a more efficient
* approach.
*
* For #GL_RENDERBUFFER the code will always fall back to software for buffers
* with alpha components, so make sure you use #GL_TEXTURE if using alpha.
Trivial formatting fixes
2014-10-12 03:17:34 +00:00
*
gdk_cairo_draw_from_gl: Document that this changes the current context
2014-11-06 08:11:17 +00:00
* Calling this may change the current GL context.
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
*/
void
gdk_cairo_draw_from_gl (cairo_t *cr,
GdkSurface: Rename lots of stuff from window->surface Mostly these are internal things, but the major public change is that event.window is now event.surface.
2018-03-20 14:14:10 +00:00
GdkSurface *surface,
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
int source,
int source_type,
int buffer_scale,
int x,
int y,
int width,
int height)
{
GdkGL: Rename "context" to "paint_context" to clarify things a bit
2014-11-06 08:12:40 +00:00
GdkGLContext *paint_context;
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
cairo_surface_t *image;
gdkgl: Don't constantly re-create the tmp framebuffer
2014-11-05 14:19:00 +00:00
guint framebuffer;
int alpha_size = 0;
GdkGLContextPaintData *paint_data;
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
int major, minor, version;
OpenGL/ES: Fix 'R' and 'B' bits inverted on Windows We need to use GL_BGRA instead of GL_RGBA when doing glReadPixels() on EGL on Windows (ANGLE) so that the red and blue bits won't be displayed inverted. Also fix the logic where we determine whether to bit blit or redraw everything.
2018-07-31 10:18:59 +00:00
gboolean es_use_bgra = FALSE;
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
GdkSurface: Rename lots of stuff from window->surface Mostly these are internal things, but the major public change is that event.window is now event.surface.
2018-03-20 14:14:10 +00:00
paint_context = gdk_surface_get_paint_gl_context (surface, NULL);
GdkGL: Rename "context" to "paint_context" to clarify things a bit
2014-11-06 08:12:40 +00:00
if (paint_context == NULL)
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
{
g_warning ("gdk_cairo_draw_gl_render_buffer failed - no paint context");
return;
}
OpenGL/ES: Fix 'R' and 'B' bits inverted on Windows We need to use GL_BGRA instead of GL_RGBA when doing glReadPixels() on EGL on Windows (ANGLE) so that the red and blue bits won't be displayed inverted. Also fix the logic where we determine whether to bit blit or redraw everything.
2018-07-31 10:18:59 +00:00
es_use_bgra = gdk_gl_context_use_es_bgra (paint_context);
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
GdkGL: Rename "context" to "paint_context" to clarify things a bit
2014-11-06 08:12:40 +00:00
gdk_gl_context_make_current (paint_context);
paint_data = gdk_gl_context_get_paint_data (paint_context);
gdkgl: Don't constantly re-create the tmp framebuffer
2014-11-05 14:19:00 +00:00
if (paint_data->tmp_framebuffer == 0)
Switch to GL_ARB_framebuffer_object GTK 4.0 was currently using GL_EXT_framebuffer_object, which is deprecated as the ARB version has been merged into OpenGL 3.0 as well as OpenGL ES 2.0, and provides laxer requirements. This is a port of !2076 for 4.x.
2020-06-12 13:08:44 +00:00
glGenFramebuffers (1, &paint_data->tmp_framebuffer);
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
if (source_type == GL_RENDERBUFFER)
{
gl: Do not use the extension API for core GL Since we are using a Core GL profile, we need to drop the extension-based API. https://bugzilla.gnome.org/show_bug.cgi?id=741946
2015-01-28 19:06:37 +00:00
glBindRenderbuffer (GL_RENDERBUFFER, source);
GdkGL: Fix and clean up gdk_cairo_draw_from_gl() This was unnecessarily creating a framebuffer in the texture case, and it was not properly setting up a framebuffer with the texture as source in the software fallback w/ texture source case.
2014-11-05 08:52:51 +00:00
glGetRenderbufferParameteriv (GL_RENDERBUFFER, GL_RENDERBUFFER_ALPHA_SIZE, &alpha_size);
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
}
else if (source_type == GL_TEXTURE)
{
opengl: Use GL_TEXTURE_2D rather than GL_TEXTURE_RECTANGLE_ARB when possible This is more standard, and most driver support non-power-of-2 TEXTURE_2D these days. We fall back for ancient drivers.
2014-10-27 20:15:37 +00:00
glBindTexture (GL_TEXTURE_2D, source);
gl: Add more OpenGL ES checks Check for the appropriate extensions depending on which type of API we're using.
2016-04-22 16:27:04 +00:00
if (gdk_gl_context_get_use_es (paint_context))
alpha_size = 1;
else
glGetTexLevelParameteriv (GL_TEXTURE_2D, 0, GL_TEXTURE_ALPHA_SIZE, &alpha_size);
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
}
else
{
g_warning ("Unsupported gl source type %d\n", source_type);
return;
}
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
gdk_gl_context_get_version (paint_context, &major, &minor);
version = major * 100 + minor;
GdkGL: Fix and clean up gdk_cairo_draw_from_gl() This was unnecessarily creating a framebuffer in the texture case, and it was not properly setting up a framebuffer with the texture as source in the software fallback w/ texture source case.
2014-11-05 08:52:51 +00:00
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
/* TODO: Use glTexSubImage2D() and do a row-by-row copy to replace
* the GL_UNPACK_ROW_LENGTH support
*/
if (gdk_gl_context_get_use_es (paint_context) &&
!(version >= 300 || gdk_gl_context_has_unpack_subimage (paint_context)))
gdkgl: Stop getting the clip region from cairo Unused
2020-10-01 07:47:34 +00:00
return;
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
/* TODO: avoid reading back non-required data due to dest clip */
image = cairo_surface_create_similar_image (cairo_get_target (cr),
(alpha_size == 0) ? CAIRO_FORMAT_RGB24 : CAIRO_FORMAT_ARGB32,
width, height);
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
cairo_surface_set_device_scale (image, buffer_scale, buffer_scale);
gl: Add more OpenGL ES checks Check for the appropriate extensions depending on which type of API we're using.
2016-04-22 16:27:04 +00:00
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
framebuffer = paint_data->tmp_framebuffer;
Switch to GL_ARB_framebuffer_object GTK 4.0 was currently using GL_EXT_framebuffer_object, which is deprecated as the ARB version has been merged into OpenGL 3.0 as well as OpenGL ES 2.0, and provides laxer requirements. This is a port of !2076 for 4.x.
2020-06-12 13:08:44 +00:00
glBindFramebuffer (GL_FRAMEBUFFER, framebuffer);
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
if (source_type == GL_RENDERBUFFER)
{
/* Create a framebuffer with the source renderbuffer and
make it the current target for reads */
Switch to GL_ARB_framebuffer_object GTK 4.0 was currently using GL_EXT_framebuffer_object, which is deprecated as the ARB version has been merged into OpenGL 3.0 as well as OpenGL ES 2.0, and provides laxer requirements. This is a port of !2076 for 4.x.
2020-06-12 13:08:44 +00:00
glFramebufferRenderbuffer (GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0,
GL_RENDERBUFFER, source);
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
}
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
else
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
{
Switch to GL_ARB_framebuffer_object GTK 4.0 was currently using GL_EXT_framebuffer_object, which is deprecated as the ARB version has been merged into OpenGL 3.0 as well as OpenGL ES 2.0, and provides laxer requirements. This is a port of !2076 for 4.x.
2020-06-12 13:08:44 +00:00
glFramebufferTexture2D (GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0,
GL_TEXTURE_2D, source, 0);
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
}
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
glPixelStorei (GL_PACK_ALIGNMENT, 4);
glPixelStorei (GL_PACK_ROW_LENGTH, cairo_image_surface_get_stride (image) / 4);
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
/* The implicit format conversion is going to make this path slower */
if (!gdk_gl_context_get_use_es (paint_context))
glReadPixels (x, y, width, height, GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV,
cairo_image_surface_get_data (image));
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
else
OpenGL/ES: Fix 'R' and 'B' bits inverted on Windows We need to use GL_BGRA instead of GL_RGBA when doing glReadPixels() on EGL on Windows (ANGLE) so that the red and blue bits won't be displayed inverted. Also fix the logic where we determine whether to bit blit or redraw everything.
2018-07-31 10:18:59 +00:00
glReadPixels (x, y, width, height, es_use_bgra ? GL_BGRA : GL_RGBA, GL_UNSIGNED_BYTE,
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
cairo_image_surface_get_data (image));
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
glPixelStorei (GL_PACK_ROW_LENGTH, 0);
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
Switch to GL_ARB_framebuffer_object GTK 4.0 was currently using GL_EXT_framebuffer_object, which is deprecated as the ARB version has been merged into OpenGL 3.0 as well as OpenGL ES 2.0, and provides laxer requirements. This is a port of !2076 for 4.x.
2020-06-12 13:08:44 +00:00
glBindFramebuffer (GL_FRAMEBUFFER, 0);
GdkGL: Fix and clean up gdk_cairo_draw_from_gl() This was unnecessarily creating a framebuffer in the texture case, and it was not properly setting up a framebuffer with the texture as source in the software fallback w/ texture source case.
2014-11-05 08:52:51 +00:00
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
cairo_surface_mark_dirty (image);
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
cairo_set_source_surface (cr, image, 0, 0);
cairo_set_operator (cr, CAIRO_OPERATOR_OVER);
cairo_paint (cr);
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
gdk: Remove all code that only existed because of use_gl Now that we don't use GL anymore, this code is unnecessary.
2016-11-22 19:03:14 +00:00
cairo_surface_destroy (image);
gdk: Add support for OpenGL This adds the new type GdkGLContext that wraps an OpenGL context for a particular native window. It also adds support for the gdk paint machinery to use OpenGL to draw everything. As soon as anyone creates a GL context for a native window we create a "paint context" for that GdkWindow and switch to using GL for painting it. This commit contains only an implementation for X11 (using GLX). The way painting works is that all client gl contexts draw into offscreen buffers rather than directly to the back buffer, and the way something gets onto the window is by using gdk_cairo_draw_from_gl() to draw part of that buffer onto the draw cairo context. As a fallback (if we're doing redirected drawing or some effect like a cairo_push_group()) we read back the gl buffer into memory and composite using cairo. This means that GL rendering works in all cases, including rendering to a PDF. However, this is not particularly fast. In the *typical* case, where we're drawing directly to the window in the regular paint loop we hit the fast path. The fast path uses opengl to draw the buffer to the window back buffer, either by blitting or texturing. Then we track the region that was drawn, and when the draw ends we paint the normal cairo surface to the window (using texture-from-pixmap in the X11 case, or texture from cairo image otherwise) in the regions where there is no gl painted. There are some complexities wrt layering of gl and cairo areas though: * We track via gdk_window_mark_paint_from_clip() whenever gtk is painting over a region we previously rendered with opengl (flushed_region). This area (needs_blend_region) is blended rather than copied at the end of the frame. * If we're drawing a gl texture with alpha we first copy the current cairo_surface inside the target region to the back buffer before we blend over it. These two operations allow us full stacking of transparent gl and cairo regions.
2014-10-09 08:45:44 +00:00
}
Reference in New Issue Copy Permalink