AuroraMiddleware/gtk2
1
0
Fork 0
forked from AuroraMiddleware/gtk
Code Issues Pull Requests Projects Releases Wiki Activity
331302005c
gtk2/gsk/gskgldriver.c

852 lines
21 KiB
C
Raw Normal View History

gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
#include "config.h"
#include "gskgldriverprivate.h"
#include "gskdebugprivate.h"
gsktexture: Allow attaching render data to textures This allows renderers (or anyone really) to attach "render data" to textures. Only the first render data sticks. You can gsk_texture_set_render_data() with the key you will use to look the data up again, and if no data has been set yet, yours will be set. You can retrieve this data via gsk_texture_get_render_data() later on. If your data has been cleared, NULL will be returned. When gsk_texture_clear_render_data() is called (which the texture will call when it is finalized), your destory notify will be called and you have to release your render data. The GL driver uses this to attach texture ids to GskTextures.
2016-11-16 04:37:20 +00:00
#include "gsktextureprivate.h"
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
#include <gdk/gdk.h>
#include <epoxy/gl.h>
typedef struct {
GLuint texture_id;
int width;
int height;
gsk: Split surface upload from texture creation We're going to need a method to create textures without a surface.
2016-07-12 10:15:45 +00:00
GLuint min_filter;
GLuint mag_filter;
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
GArray *fbos;
gsktexture: Allow attaching render data to textures This allows renderers (or anyone really) to attach "render data" to textures. Only the first render data sticks. You can gsk_texture_set_render_data() with the key you will use to look the data up again, and if no data has been set yet, yours will be set. You can retrieve this data via gsk_texture_get_render_data() later on. If your data has been cleared, NULL will be returned. When gsk_texture_clear_render_data() is called (which the texture will call when it is finalized), your destory notify will be called and you have to release your render data. The GL driver uses this to attach texture ids to GskTextures.
2016-11-16 04:37:20 +00:00
GskTexture *user;
gsk: Recycle textures across frames We keep the textures used inside a frame around until the end of the following frame; whenever we need a texture with the same size, and it's not marked in use, then we just reuse the existing texture.
2016-08-04 11:37:54 +00:00
gboolean in_use : 1;
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
} Texture;
typedef struct {
GLuint vao_id;
GLuint buffer_id;
GLuint position_id;
GLuint uv_id;
gsk: Reuse VAOs with identical buffers Just like we reuse texture ids with the same size we can, at the expense of a little memory, reuse vertex buffers if they reference the same attributes and contain the same data. Each VAO is marked as free at the end of the frame, and if it's not reused in the following frame, it gets dropped.
2016-08-09 10:24:05 +00:00
GskQuadVertex *quads;
int n_quads;
gboolean in_use : 1;
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
} Vao;
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
typedef struct {
GLuint fbo_id;
GLuint depth_stencil_id;
} Fbo;
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
struct _GskGLDriver
{
GObject parent_instance;
GdkGLContext *gl_context;
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
Fbo default_fbo;
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
gsk: Turn texture cache into a hash table We want to add the list of FBOs tied to a texture; this means we cannot trivally copy the Texture structure when adding it to a GArray. We're also going to have more textures than VAOs, so it makes more sense to use a O(1) access data structure for them.
2016-07-12 13:59:37 +00:00
GHashTable *textures;
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
GHashTable *vaos;
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
Texture *bound_source_texture;
Texture *bound_mask_texture;
Vao *bound_vao;
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
Fbo *bound_fbo;
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
gsk: Add check on GL_MAX_TEXTURE_SIZE We should check (once) what's the maximum GL texture size and then use that size to check when we create any GL texture, to avoid errors.
2016-10-26 14:47:29 +00:00
int max_texture_size;
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
gboolean in_frame : 1;
};
enum
{
PROP_GL_CONTEXT = 1,
N_PROPS
};
static GParamSpec *gsk_gl_driver_properties[N_PROPS];
G_DEFINE_TYPE (GskGLDriver, gsk_gl_driver, G_TYPE_OBJECT)
gsk: Turn texture cache into a hash table We want to add the list of FBOs tied to a texture; this means we cannot trivally copy the Texture structure when adding it to a GArray. We're also going to have more textures than VAOs, so it makes more sense to use a O(1) access data structure for them.
2016-07-12 13:59:37 +00:00
static Texture *
texture_new (void)
{
return g_slice_new0 (Texture);
}
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
static void
gsk: Turn texture cache into a hash table We want to add the list of FBOs tied to a texture; this means we cannot trivally copy the Texture structure when adding it to a GArray. We're also going to have more textures than VAOs, so it makes more sense to use a O(1) access data structure for them.
2016-07-12 13:59:37 +00:00
texture_free (gpointer data)
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
{
Texture *t = data;
gsktexture: Allow attaching render data to textures This allows renderers (or anyone really) to attach "render data" to textures. Only the first render data sticks. You can gsk_texture_set_render_data() with the key you will use to look the data up again, and if no data has been set yet, yours will be set. You can retrieve this data via gsk_texture_get_render_data() later on. If your data has been cleared, NULL will be returned. When gsk_texture_clear_render_data() is called (which the texture will call when it is finalized), your destory notify will be called and you have to release your render data. The GL driver uses this to attach texture ids to GskTextures.
2016-11-16 04:37:20 +00:00
if (t->user)
gsk_texture_clear_render_data (t->user);
gsk: Add GskTexture
2016-11-07 16:59:38 +00:00
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
g_clear_pointer (&t->fbos, g_array_unref);
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
glDeleteTextures (1, &t->texture_id);
gsk: Turn texture cache into a hash table We want to add the list of FBOs tied to a texture; this means we cannot trivally copy the Texture structure when adding it to a GArray. We're also going to have more textures than VAOs, so it makes more sense to use a O(1) access data structure for them.
2016-07-12 13:59:37 +00:00
g_slice_free (Texture, t);
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
}
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
static void
fbo_clear (gpointer data)
{
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
Fbo *f = data;
if (f->depth_stencil_id != 0)
glDeleteRenderbuffers (1, &f->depth_stencil_id);
glDeleteFramebuffers (1, &f->fbo_id);
}
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
static Vao *
vao_new (void)
{
return g_slice_new0 (Vao);
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
}
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
static void
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
vao_free (gpointer data)
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
{
Vao *v = data;
gsk: Reuse VAOs with identical buffers Just like we reuse texture ids with the same size we can, at the expense of a little memory, reuse vertex buffers if they reference the same attributes and contain the same data. Each VAO is marked as free at the end of the frame, and if it's not reused in the following frame, it gets dropped.
2016-08-09 10:24:05 +00:00
g_free (v->quads);
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
glDeleteBuffers (1, &v->buffer_id);
glDeleteVertexArrays (1, &v->vao_id);
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
g_slice_free (Vao, v);
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
}
static void
gsk_gl_driver_finalize (GObject *gobject)
{
GskGLDriver *self = GSK_GL_DRIVER (gobject);
gsk: Recycle textures across frames We keep the textures used inside a frame around until the end of the following frame; whenever we need a texture with the same size, and it's not marked in use, then we just reuse the existing texture.
2016-08-04 11:37:54 +00:00
gdk_gl_context_make_current (self->gl_context);
gsk: Turn texture cache into a hash table We want to add the list of FBOs tied to a texture; this means we cannot trivally copy the Texture structure when adding it to a GArray. We're also going to have more textures than VAOs, so it makes more sense to use a O(1) access data structure for them.
2016-07-12 13:59:37 +00:00
g_clear_pointer (&self->textures, g_hash_table_unref);
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
g_clear_pointer (&self->vaos, g_hash_table_unref);
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
gsk: Recycle textures across frames We keep the textures used inside a frame around until the end of the following frame; whenever we need a texture with the same size, and it's not marked in use, then we just reuse the existing texture.
2016-08-04 11:37:54 +00:00
if (self->gl_context == gdk_gl_context_get_current ())
gdk_gl_context_clear_current ();
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
g_clear_object (&self->gl_context);
G_OBJECT_CLASS (gsk_gl_driver_parent_class)->finalize (gobject);
}
static void
gsk_gl_driver_set_property (GObject *gobject,
guint prop_id,
const GValue *value,
GParamSpec *pspec)
{
GskGLDriver *self = GSK_GL_DRIVER (gobject);
switch (prop_id)
{
case PROP_GL_CONTEXT:
self->gl_context = g_value_dup_object (value);
break;
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (gobject, prop_id, pspec);
}
}
static void
gsk_gl_driver_get_property (GObject *gobject,
guint prop_id,
GValue *value,
GParamSpec *pspec)
{
GskGLDriver *self = GSK_GL_DRIVER (gobject);
switch (prop_id)
{
case PROP_GL_CONTEXT:
g_value_set_object (value, self->gl_context);
break;
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (gobject, prop_id, pspec);
}
}
static void
gsk_gl_driver_class_init (GskGLDriverClass *klass)
{
GObjectClass *gobject_class = G_OBJECT_CLASS (klass);
gobject_class->set_property = gsk_gl_driver_set_property;
gobject_class->get_property = gsk_gl_driver_get_property;
gobject_class->finalize = gsk_gl_driver_finalize;
gsk_gl_driver_properties[PROP_GL_CONTEXT] =
g_param_spec_object ("gl-context", "GL Context", "The GL context used by the driver",
GDK_TYPE_GL_CONTEXT,
G_PARAM_READWRITE | G_PARAM_CONSTRUCT_ONLY | G_PARAM_STATIC_STRINGS);
g_object_class_install_properties (gobject_class, N_PROPS, gsk_gl_driver_properties);
}
static void
gsk_gl_driver_init (GskGLDriver *self)
{
gsk: Turn texture cache into a hash table We want to add the list of FBOs tied to a texture; this means we cannot trivally copy the Texture structure when adding it to a GArray. We're also going to have more textures than VAOs, so it makes more sense to use a O(1) access data structure for them.
2016-07-12 13:59:37 +00:00
self->textures = g_hash_table_new_full (NULL, NULL, NULL, texture_free);
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
self->vaos = g_hash_table_new_full (NULL, NULL, NULL, vao_free);
gsk: Add check on GL_MAX_TEXTURE_SIZE We should check (once) what's the maximum GL texture size and then use that size to check when we create any GL texture, to avoid errors.
2016-10-26 14:47:29 +00:00
self->max_texture_size = -1;
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
}
GskGLDriver *
gsk_gl_driver_new (GdkGLContext *context)
{
g_return_val_if_fail (GDK_IS_GL_CONTEXT (context), NULL);
return g_object_new (GSK_TYPE_GL_DRIVER,
"gl-context", context,
NULL);
}
void
gsk_gl_driver_begin_frame (GskGLDriver *driver)
{
g_return_if_fail (GSK_IS_GL_DRIVER (driver));
g_return_if_fail (!driver->in_frame);
driver->in_frame = TRUE;
gsk: Add check on GL_MAX_TEXTURE_SIZE We should check (once) what's the maximum GL texture size and then use that size to check when we create any GL texture, to avoid errors.
2016-10-26 14:47:29 +00:00
if (driver->max_texture_size < 0)
{
glGetIntegerv (GL_MAX_TEXTURE_SIZE, (GLint *) &driver->max_texture_size);
GSK_NOTE (OPENGL, g_print ("GL max texture size: %d\n", driver->max_texture_size));
}
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
glGetIntegerv (GL_FRAMEBUFFER_BINDING, (GLint *) &(driver->default_fbo.fbo_id));
driver->bound_fbo = &driver->default_fbo;
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
glActiveTexture (GL_TEXTURE0);
glBindTexture (GL_TEXTURE_2D, 0);
glActiveTexture (GL_TEXTURE0 + 1);
glBindTexture (GL_TEXTURE_2D, 0);
glBindVertexArray (0);
glUseProgram (0);
glActiveTexture (GL_TEXTURE0);
}
void
gsk_gl_driver_end_frame (GskGLDriver *driver)
{
g_return_if_fail (GSK_IS_GL_DRIVER (driver));
g_return_if_fail (driver->in_frame);
glBindTexture (GL_TEXTURE_2D, 0);
glUseProgram (0);
glBindVertexArray (0);
driver->bound_source_texture = NULL;
driver->bound_mask_texture = NULL;
driver->bound_vao = NULL;
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
driver->bound_fbo = NULL;
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
GSK_NOTE (OPENGL,
g_print ("*** Frame end: textures=%d, vaos=%d\n",
g_hash_table_size (driver->textures),
g_hash_table_size (driver->vaos)));
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
driver->in_frame = FALSE;
}
gsk: Reuse VAOs with identical buffers Just like we reuse texture ids with the same size we can, at the expense of a little memory, reuse vertex buffers if they reference the same attributes and contain the same data. Each VAO is marked as free at the end of the frame, and if it's not reused in the following frame, it gets dropped.
2016-08-09 10:24:05 +00:00
int
gsk: Recycle textures across frames We keep the textures used inside a frame around until the end of the following frame; whenever we need a texture with the same size, and it's not marked in use, then we just reuse the existing texture.
2016-08-04 11:37:54 +00:00
gsk_gl_driver_collect_textures (GskGLDriver *driver)
{
GHashTableIter iter;
gpointer value_p = NULL;
gsk: Reuse VAOs with identical buffers Just like we reuse texture ids with the same size we can, at the expense of a little memory, reuse vertex buffers if they reference the same attributes and contain the same data. Each VAO is marked as free at the end of the frame, and if it's not reused in the following frame, it gets dropped.
2016-08-09 10:24:05 +00:00
int old_size;
gsk: Recycle textures across frames We keep the textures used inside a frame around until the end of the following frame; whenever we need a texture with the same size, and it's not marked in use, then we just reuse the existing texture.
2016-08-04 11:37:54 +00:00
gsk: Reuse VAOs with identical buffers Just like we reuse texture ids with the same size we can, at the expense of a little memory, reuse vertex buffers if they reference the same attributes and contain the same data. Each VAO is marked as free at the end of the frame, and if it's not reused in the following frame, it gets dropped.
2016-08-09 10:24:05 +00:00
g_return_val_if_fail (GSK_IS_GL_DRIVER (driver), 0);
g_return_val_if_fail (!driver->in_frame, 0);
old_size = g_hash_table_size (driver->textures);
gsk: Recycle textures across frames We keep the textures used inside a frame around until the end of the following frame; whenever we need a texture with the same size, and it's not marked in use, then we just reuse the existing texture.
2016-08-04 11:37:54 +00:00
g_hash_table_iter_init (&iter, driver->textures);
while (g_hash_table_iter_next (&iter, NULL, &value_p))
{
Texture *t = value_p;
gsktexture: Allow attaching render data to textures This allows renderers (or anyone really) to attach "render data" to textures. Only the first render data sticks. You can gsk_texture_set_render_data() with the key you will use to look the data up again, and if no data has been set yet, yours will be set. You can retrieve this data via gsk_texture_get_render_data() later on. If your data has been cleared, NULL will be returned. When gsk_texture_clear_render_data() is called (which the texture will call when it is finalized), your destory notify will be called and you have to release your render data. The GL driver uses this to attach texture ids to GskTextures.
2016-11-16 04:37:20 +00:00
if (t->user)
gsk: Add GskTexture
2016-11-07 16:59:38 +00:00
continue;
gsk: Recycle textures across frames We keep the textures used inside a frame around until the end of the following frame; whenever we need a texture with the same size, and it's not marked in use, then we just reuse the existing texture.
2016-08-04 11:37:54 +00:00
if (t->in_use)
gsk: Reuse VAOs with identical buffers Just like we reuse texture ids with the same size we can, at the expense of a little memory, reuse vertex buffers if they reference the same attributes and contain the same data. Each VAO is marked as free at the end of the frame, and if it's not reused in the following frame, it gets dropped.
2016-08-09 10:24:05 +00:00
{
t->in_use = FALSE;
g_clear_pointer (&t->fbos, g_array_unref);
}
gsk: Recycle textures across frames We keep the textures used inside a frame around until the end of the following frame; whenever we need a texture with the same size, and it's not marked in use, then we just reuse the existing texture.
2016-08-04 11:37:54 +00:00
else
g_hash_table_iter_remove (&iter);
}
gsk: Reuse VAOs with identical buffers Just like we reuse texture ids with the same size we can, at the expense of a little memory, reuse vertex buffers if they reference the same attributes and contain the same data. Each VAO is marked as free at the end of the frame, and if it's not reused in the following frame, it gets dropped.
2016-08-09 10:24:05 +00:00
return old_size - g_hash_table_size (driver->textures);
}
int
gsk_gl_driver_collect_vaos (GskGLDriver *driver)
{
GHashTableIter iter;
gpointer value_p = NULL;
int old_size;
g_return_val_if_fail (GSK_IS_GL_DRIVER (driver), 0);
g_return_val_if_fail (!driver->in_frame, 0);
old_size = g_hash_table_size (driver->vaos);
g_hash_table_iter_init (&iter, driver->vaos);
while (g_hash_table_iter_next (&iter, NULL, &value_p))
{
Vao *v = value_p;
if (v->in_use)
v->in_use = FALSE;
else
g_hash_table_iter_remove (&iter);
}
return old_size - g_hash_table_size (driver->vaos);
gsk: Recycle textures across frames We keep the textures used inside a frame around until the end of the following frame; whenever we need a texture with the same size, and it's not marked in use, then we just reuse the existing texture.
2016-08-04 11:37:54 +00:00
}
gsk: Add check on GL_MAX_TEXTURE_SIZE We should check (once) what's the maximum GL texture size and then use that size to check when we create any GL texture, to avoid errors.
2016-10-26 14:47:29 +00:00
int
gsk_gl_driver_get_max_texture_size (GskGLDriver *driver)
{
if (driver->max_texture_size < 0)
{
if (gdk_gl_context_get_use_es (driver->gl_context))
return 2048;
return 1024;
}
return driver->max_texture_size;
}
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
static Texture *
gsk_gl_driver_get_texture (GskGLDriver *driver,
int texture_id)
{
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
Texture *t;
if (g_hash_table_lookup_extended (driver->textures, GINT_TO_POINTER (texture_id), NULL, (gpointer *) &t))
return t;
return NULL;
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
}
static Vao *
gsk_gl_driver_get_vao (GskGLDriver *driver,
int vao_id)
{
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
Vao *v;
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
if (g_hash_table_lookup_extended (driver->vaos, GINT_TO_POINTER (vao_id), NULL, (gpointer *) &v))
return v;
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
return NULL;
}
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
static Fbo *
gsk_gl_driver_get_fbo (GskGLDriver *driver,
int texture_id)
{
Texture *t = gsk_gl_driver_get_texture (driver, texture_id);
if (t->fbos == NULL)
return &driver->default_fbo;
return &g_array_index (t->fbos, Fbo, 0);
}
gsk: Recycle textures across frames We keep the textures used inside a frame around until the end of the following frame; whenever we need a texture with the same size, and it's not marked in use, then we just reuse the existing texture.
2016-08-04 11:37:54 +00:00
static Texture *
find_texture_by_size (GHashTable *textures,
int width,
int height)
{
GHashTableIter iter;
gpointer value_p = NULL;
g_hash_table_iter_init (&iter, textures);
while (g_hash_table_iter_next (&iter, NULL, &value_p))
{
Texture *t = value_p;
if (t->width == width && t->height == height)
return t;
}
return NULL;
}
gsktexture: Allow attaching render data to textures This allows renderers (or anyone really) to attach "render data" to textures. Only the first render data sticks. You can gsk_texture_set_render_data() with the key you will use to look the data up again, and if no data has been set yet, yours will be set. You can retrieve this data via gsk_texture_get_render_data() later on. If your data has been cleared, NULL will be returned. When gsk_texture_clear_render_data() is called (which the texture will call when it is finalized), your destory notify will be called and you have to release your render data. The GL driver uses this to attach texture ids to GskTextures.
2016-11-16 04:37:20 +00:00
static Texture *
create_texture (GskGLDriver *driver,
int width,
int height)
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
{
guint texture_id;
gsk: Turn texture cache into a hash table We want to add the list of FBOs tied to a texture; this means we cannot trivally copy the Texture structure when adding it to a GArray. We're also going to have more textures than VAOs, so it makes more sense to use a O(1) access data structure for them.
2016-07-12 13:59:37 +00:00
Texture *t;
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
gsk: Add check on GL_MAX_TEXTURE_SIZE We should check (once) what's the maximum GL texture size and then use that size to check when we create any GL texture, to avoid errors.
2016-10-26 14:47:29 +00:00
if (width >= driver->max_texture_size ||
height >= driver->max_texture_size)
{
g_critical ("Texture %d x %d is bigger than supported texture limit of %d; clipping...",
width, height,
driver->max_texture_size);
width = MIN (width, driver->max_texture_size);
height = MIN (height, driver->max_texture_size);
}
gsk: Recycle textures across frames We keep the textures used inside a frame around until the end of the following frame; whenever we need a texture with the same size, and it's not marked in use, then we just reuse the existing texture.
2016-08-04 11:37:54 +00:00
t = find_texture_by_size (driver->textures, width, height);
gsktexture: Allow attaching render data to textures This allows renderers (or anyone really) to attach "render data" to textures. Only the first render data sticks. You can gsk_texture_set_render_data() with the key you will use to look the data up again, and if no data has been set yet, yours will be set. You can retrieve this data via gsk_texture_get_render_data() later on. If your data has been cleared, NULL will be returned. When gsk_texture_clear_render_data() is called (which the texture will call when it is finalized), your destory notify will be called and you have to release your render data. The GL driver uses this to attach texture ids to GskTextures.
2016-11-16 04:37:20 +00:00
if (t != NULL && !t->in_use && t->user == NULL)
gsk: Recycle textures across frames We keep the textures used inside a frame around until the end of the following frame; whenever we need a texture with the same size, and it's not marked in use, then we just reuse the existing texture.
2016-08-04 11:37:54 +00:00
{
gsk: Reuse VAOs with identical buffers Just like we reuse texture ids with the same size we can, at the expense of a little memory, reuse vertex buffers if they reference the same attributes and contain the same data. Each VAO is marked as free at the end of the frame, and if it's not reused in the following frame, it gets dropped.
2016-08-09 10:24:05 +00:00
GSK_NOTE (OPENGL, g_print ("Reusing Texture(%d) for size %dx%d\n",
t->texture_id, t->width, t->height));
gsk: Recycle textures across frames We keep the textures used inside a frame around until the end of the following frame; whenever we need a texture with the same size, and it's not marked in use, then we just reuse the existing texture.
2016-08-04 11:37:54 +00:00
t->in_use = TRUE;
gsktexture: Allow attaching render data to textures This allows renderers (or anyone really) to attach "render data" to textures. Only the first render data sticks. You can gsk_texture_set_render_data() with the key you will use to look the data up again, and if no data has been set yet, yours will be set. You can retrieve this data via gsk_texture_get_render_data() later on. If your data has been cleared, NULL will be returned. When gsk_texture_clear_render_data() is called (which the texture will call when it is finalized), your destory notify will be called and you have to release your render data. The GL driver uses this to attach texture ids to GskTextures.
2016-11-16 04:37:20 +00:00
return t;
gsk: Recycle textures across frames We keep the textures used inside a frame around until the end of the following frame; whenever we need a texture with the same size, and it's not marked in use, then we just reuse the existing texture.
2016-08-04 11:37:54 +00:00
}
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
glGenTextures (1, &texture_id);
gsk: Turn texture cache into a hash table We want to add the list of FBOs tied to a texture; this means we cannot trivally copy the Texture structure when adding it to a GArray. We're also going to have more textures than VAOs, so it makes more sense to use a O(1) access data structure for them.
2016-07-12 13:59:37 +00:00
t = texture_new ();
t->texture_id = texture_id;
t->width = width;
t->height = height;
gsk: Rework surface -> texture API Drop the texture parameters handling from the texture creation, and associate them with the contents upload. Also, rename the function to something more in line with what it does.
2016-07-19 13:32:17 +00:00
t->min_filter = GL_NEAREST;
t->mag_filter = GL_NEAREST;
gsk: Recycle textures across frames We keep the textures used inside a frame around until the end of the following frame; whenever we need a texture with the same size, and it's not marked in use, then we just reuse the existing texture.
2016-08-04 11:37:54 +00:00
t->in_use = TRUE;
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
g_hash_table_insert (driver->textures, GINT_TO_POINTER (texture_id), t);
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
gsktexture: Allow attaching render data to textures This allows renderers (or anyone really) to attach "render data" to textures. Only the first render data sticks. You can gsk_texture_set_render_data() with the key you will use to look the data up again, and if no data has been set yet, yours will be set. You can retrieve this data via gsk_texture_get_render_data() later on. If your data has been cleared, NULL will be returned. When gsk_texture_clear_render_data() is called (which the texture will call when it is finalized), your destory notify will be called and you have to release your render data. The GL driver uses this to attach texture ids to GskTextures.
2016-11-16 04:37:20 +00:00
return t;
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
}
gsktexture: Allow attaching render data to textures This allows renderers (or anyone really) to attach "render data" to textures. Only the first render data sticks. You can gsk_texture_set_render_data() with the key you will use to look the data up again, and if no data has been set yet, yours will be set. You can retrieve this data via gsk_texture_get_render_data() later on. If your data has been cleared, NULL will be returned. When gsk_texture_clear_render_data() is called (which the texture will call when it is finalized), your destory notify will be called and you have to release your render data. The GL driver uses this to attach texture ids to GskTextures.
2016-11-16 04:37:20 +00:00
static void
gsk_gl_driver_release_texture (gpointer data)
{
Texture *t = data;
t->user = NULL;
}
int
gsk_gl_driver_get_texture_for_texture (GskGLDriver *driver,
GskTexture *texture,
int min_filter,
int mag_filter)
gsk: Add GskTexture
2016-11-07 16:59:38 +00:00
{
Texture *t;
gsktexture: Allow attaching render data to textures This allows renderers (or anyone really) to attach "render data" to textures. Only the first render data sticks. You can gsk_texture_set_render_data() with the key you will use to look the data up again, and if no data has been set yet, yours will be set. You can retrieve this data via gsk_texture_get_render_data() later on. If your data has been cleared, NULL will be returned. When gsk_texture_clear_render_data() is called (which the texture will call when it is finalized), your destory notify will be called and you have to release your render data. The GL driver uses this to attach texture ids to GskTextures.
2016-11-16 04:37:20 +00:00
cairo_surface_t *surface;
g_return_val_if_fail (GSK_IS_GL_DRIVER (driver), -1);
g_return_val_if_fail (GSK_IS_TEXTURE (texture), -1);
t = gsk_texture_get_render_data (texture, driver);
if (t)
{
if (t->min_filter == min_filter && t->mag_filter == mag_filter)
return t->texture_id;
}
gsk: Add GskTexture
2016-11-07 16:59:38 +00:00
gsktexture: Allow attaching render data to textures This allows renderers (or anyone really) to attach "render data" to textures. Only the first render data sticks. You can gsk_texture_set_render_data() with the key you will use to look the data up again, and if no data has been set yet, yours will be set. You can retrieve this data via gsk_texture_get_render_data() later on. If your data has been cleared, NULL will be returned. When gsk_texture_clear_render_data() is called (which the texture will call when it is finalized), your destory notify will be called and you have to release your render data. The GL driver uses this to attach texture ids to GskTextures.
2016-11-16 04:37:20 +00:00
t = create_texture (driver, gsk_texture_get_width (texture), gsk_texture_get_height (texture));
if (gsk_texture_set_render_data (texture, driver, t, gsk_gl_driver_release_texture))
t->user = texture;
surface = gsk_texture_download (texture);
gsk_gl_driver_bind_source_texture (driver, t->texture_id);
gsk_gl_driver_init_texture_with_surface (driver,
t->texture_id,
surface,
min_filter,
mag_filter);
gsk: Don't leak texture surfaces gsk_texture_download returns a reference.
2016-11-18 01:29:07 +00:00
cairo_surface_destroy (surface);
gsktexture: Allow attaching render data to textures This allows renderers (or anyone really) to attach "render data" to textures. Only the first render data sticks. You can gsk_texture_set_render_data() with the key you will use to look the data up again, and if no data has been set yet, yours will be set. You can retrieve this data via gsk_texture_get_render_data() later on. If your data has been cleared, NULL will be returned. When gsk_texture_clear_render_data() is called (which the texture will call when it is finalized), your destory notify will be called and you have to release your render data. The GL driver uses this to attach texture ids to GskTextures.
2016-11-16 04:37:20 +00:00
return t->texture_id;
}
int
gsk_gl_driver_create_texture (GskGLDriver *driver,
int width,
int height)
{
Texture *t;
g_return_val_if_fail (GSK_IS_GL_DRIVER (driver), -1);
gsk: Add GskTexture
2016-11-07 16:59:38 +00:00
gsktexture: Allow attaching render data to textures This allows renderers (or anyone really) to attach "render data" to textures. Only the first render data sticks. You can gsk_texture_set_render_data() with the key you will use to look the data up again, and if no data has been set yet, yours will be set. You can retrieve this data via gsk_texture_get_render_data() later on. If your data has been cleared, NULL will be returned. When gsk_texture_clear_render_data() is called (which the texture will call when it is finalized), your destory notify will be called and you have to release your render data. The GL driver uses this to attach texture ids to GskTextures.
2016-11-16 04:37:20 +00:00
t = create_texture (driver, width, height);
return t->texture_id;
gsk: Add GskTexture
2016-11-07 16:59:38 +00:00
}
gsk: Reuse VAOs with identical buffers Just like we reuse texture ids with the same size we can, at the expense of a little memory, reuse vertex buffers if they reference the same attributes and contain the same data. Each VAO is marked as free at the end of the frame, and if it's not reused in the following frame, it gets dropped.
2016-08-09 10:24:05 +00:00
static Vao *
find_vao (GHashTable *vaos,
int position_id,
int uv_id,
int n_quads,
GskQuadVertex *quads)
{
GHashTableIter iter;
gpointer value_p = NULL;
g_hash_table_iter_init (&iter, vaos);
while (g_hash_table_iter_next (&iter, NULL, &value_p))
{
Vao *v = value_p;
if (v->position_id != position_id || v->uv_id != uv_id)
continue;
if (v->n_quads != n_quads)
continue;
if (memcmp (v->quads, quads, sizeof (GskQuadVertex) * n_quads) == 0)
return v;
}
return NULL;
}
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
int
gsk_gl_driver_create_vao_for_quad (GskGLDriver *driver,
int position_id,
int uv_id,
int n_quads,
GskQuadVertex *quads)
{
GLuint vao_id, buffer_id;
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
Vao *v;
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
g_return_val_if_fail (GSK_IS_GL_DRIVER (driver), -1);
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
g_return_val_if_fail (driver->in_frame, -1);
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
gsk: Reuse VAOs with identical buffers Just like we reuse texture ids with the same size we can, at the expense of a little memory, reuse vertex buffers if they reference the same attributes and contain the same data. Each VAO is marked as free at the end of the frame, and if it's not reused in the following frame, it gets dropped.
2016-08-09 10:24:05 +00:00
v = find_vao (driver->vaos, position_id, uv_id, n_quads, quads);
if (v != NULL && !v->in_use)
{
GSK_NOTE (OPENGL, g_print ("Reusing VAO(%d)\n", v->vao_id));
v->in_use = TRUE;
return v->vao_id;
}
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
glGenVertexArrays (1, &vao_id);
glBindVertexArray (vao_id);
glGenBuffers (1, &buffer_id);
glBindBuffer (GL_ARRAY_BUFFER, buffer_id);
gsk: Pass the appropriate value for the n_quads argument Instead of passing the size of the buffer, we should pass the number of quads; we know what the size of a single quad structure is, so we can do the multiplication internally when creating the VAO. This allows us to print the quads for debugging purposes.
2016-08-04 18:14:51 +00:00
glBufferData (GL_ARRAY_BUFFER, sizeof (GskQuadVertex) * n_quads, quads, GL_STATIC_DRAW);
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
glEnableVertexAttribArray (position_id);
glVertexAttribPointer (position_id, 2, GL_FLOAT, GL_FALSE,
sizeof (GskQuadVertex),
(void *) G_STRUCT_OFFSET (GskQuadVertex, position));
glEnableVertexAttribArray (uv_id);
glVertexAttribPointer (uv_id, 2, GL_FLOAT, GL_FALSE,
sizeof (GskQuadVertex),
(void *) G_STRUCT_OFFSET (GskQuadVertex, uv));
glBindBuffer (GL_ARRAY_BUFFER, 0);
glBindVertexArray (0);
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
v = vao_new ();
v->vao_id = vao_id;
v->buffer_id = buffer_id;
v->position_id = position_id;
v->uv_id = uv_id;
gsk: Reuse VAOs with identical buffers Just like we reuse texture ids with the same size we can, at the expense of a little memory, reuse vertex buffers if they reference the same attributes and contain the same data. Each VAO is marked as free at the end of the frame, and if it's not reused in the following frame, it gets dropped.
2016-08-09 10:24:05 +00:00
v->n_quads = n_quads;
v->quads = g_memdup (quads, sizeof (GskQuadVertex) * n_quads);
v->in_use = TRUE;
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
g_hash_table_insert (driver->vaos, GINT_TO_POINTER (vao_id), v);
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
gsk: Pass the appropriate value for the n_quads argument Instead of passing the size of the buffer, we should pass the number of quads; we know what the size of a single quad structure is, so we can do the multiplication internally when creating the VAO. This allows us to print the quads for debugging purposes.
2016-08-04 18:14:51 +00:00
#ifdef G_ENABLE_DEBUG
if (GSK_DEBUG_CHECK (OPENGL))
{
int i;
g_print ("New VAO(%d) for quad[%d] : {\n", v->vao_id, n_quads);
for (i = 0; i < n_quads; i++)
{
g_print (" { x:%.2f, y:%.2f } { u:%.2f, v:%.2f }\n",
quads[i].position[0], quads[i].position[1],
quads[i].uv[0], quads[i].uv[1]);
}
g_print ("}\n");
}
#endif
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
return vao_id;
}
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
int
gsk_gl_driver_create_render_target (GskGLDriver *driver,
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
int texture_id,
gboolean add_depth_buffer,
gboolean add_stencil_buffer)
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
{
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
GLuint fbo_id, depth_stencil_buffer_id;
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
Texture *t;
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
Fbo f;
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
g_return_val_if_fail (GSK_IS_GL_DRIVER (driver), -1);
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
g_return_val_if_fail (driver->in_frame, -1);
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
t = gsk_gl_driver_get_texture (driver, texture_id);
if (t == NULL)
return -1;
if (t->fbos == NULL)
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
{
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
t->fbos = g_array_new (FALSE, FALSE, sizeof (Fbo));
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
g_array_set_clear_func (t->fbos, fbo_clear);
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
}
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
glGenFramebuffers (1, &fbo_id);
glBindFramebuffer (GL_FRAMEBUFFER, fbo_id);
glFramebufferTexture2D (GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, t->texture_id, 0);
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
if (add_depth_buffer || add_stencil_buffer)
glGenRenderbuffersEXT (1, &depth_stencil_buffer_id);
else
depth_stencil_buffer_id = 0;
glBindRenderbuffer (GL_RENDERBUFFER, depth_stencil_buffer_id);
if (add_depth_buffer || add_stencil_buffer)
{
if (add_stencil_buffer)
glRenderbufferStorage (GL_RENDERBUFFER, GL_DEPTH24_STENCIL8, t->width, t->height);
else
glRenderbufferStorage (GL_RENDERBUFFER, GL_DEPTH_COMPONENT24, t->width, t->height);
if (add_depth_buffer)
glFramebufferRenderbuffer (GL_FRAMEBUFFER, GL_DEPTH_ATTACHMENT,
GL_RENDERBUFFER, depth_stencil_buffer_id);
if (add_stencil_buffer)
glFramebufferRenderbufferEXT (GL_FRAMEBUFFER, GL_STENCIL_ATTACHMENT,
GL_RENDERBUFFER, depth_stencil_buffer_id);
}
f.fbo_id = fbo_id;
f.depth_stencil_id = depth_stencil_buffer_id;
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
g_array_append_val (t->fbos, f);
glBindFramebuffer (GL_FRAMEBUFFER, driver->default_fbo.fbo_id);
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
return fbo_id;
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
}
void
gsk_gl_driver_bind_source_texture (GskGLDriver *driver,
int texture_id)
{
Texture *t;
g_return_if_fail (GSK_IS_GL_DRIVER (driver));
g_return_if_fail (driver->in_frame);
t = gsk_gl_driver_get_texture (driver, texture_id);
if (t == NULL)
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
{
g_critical ("No texture %d found.", texture_id);
return;
}
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
if (driver->bound_source_texture != t)
{
glActiveTexture (GL_TEXTURE0);
glBindTexture (GL_TEXTURE_2D, t->texture_id);
driver->bound_source_texture = t;
}
}
void
gsk_gl_driver_bind_mask_texture (GskGLDriver *driver,
int texture_id)
{
Texture *t;
g_return_if_fail (GSK_IS_GL_DRIVER (driver));
g_return_if_fail (driver->in_frame);
t = gsk_gl_driver_get_texture (driver, texture_id);
if (t == NULL)
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
{
g_critical ("No texture %d found.", texture_id);
return;
}
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
if (driver->bound_mask_texture != t)
{
glActiveTexture (GL_TEXTURE0 + 1);
glBindTexture (GL_TEXTURE_2D, t->texture_id);
glActiveTexture (GL_TEXTURE0);
driver->bound_mask_texture = t;
}
}
void
gsk_gl_driver_bind_vao (GskGLDriver *driver,
int vao_id)
{
Vao *v;
g_return_if_fail (GSK_IS_GL_DRIVER (driver));
g_return_if_fail (driver->in_frame);
v = gsk_gl_driver_get_vao (driver, vao_id);
if (v == NULL)
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
{
g_critical ("No VAO %d found.", vao_id);
return;
}
gsk: Add GskGLDriver We can move the caching and management of textures, VAOs, and state of the GL machinery into a single object, GskGLDriver.
2016-07-08 15:21:13 +00:00
if (driver->bound_vao != v)
{
glBindVertexArray (v->vao_id);
glBindBuffer (GL_ARRAY_BUFFER, v->buffer_id);
glEnableVertexAttribArray (v->position_id);
glEnableVertexAttribArray (v->uv_id);
driver->bound_vao = v;
}
}
gsk: Split surface upload from texture creation We're going to need a method to create textures without a surface.
2016-07-12 10:15:45 +00:00
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
gboolean
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
gsk_gl_driver_bind_render_target (GskGLDriver *driver,
int texture_id)
{
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
int status;
Fbo *f;
g_return_val_if_fail (GSK_IS_GL_DRIVER (driver), FALSE);
g_return_val_if_fail (driver->in_frame, FALSE);
f = gsk_gl_driver_get_fbo (driver, texture_id);
if (f == NULL)
{
g_critical ("No render target associated to texture %d found.", texture_id);
return FALSE;
}
if (f != driver->bound_fbo)
{
glBindFramebuffer (GL_FRAMEBUFFER, f->fbo_id);
driver->bound_fbo = f;
}
status = glCheckFramebufferStatus (GL_FRAMEBUFFER);
return status == GL_FRAMEBUFFER_COMPLETE;
}
void
gsk_gl_driver_destroy_texture (GskGLDriver *driver,
int texture_id)
{
g_return_if_fail (GSK_IS_GL_DRIVER (driver));
g_hash_table_remove (driver->textures, GINT_TO_POINTER (texture_id));
}
void
gsk_gl_driver_destroy_vao (GskGLDriver *driver,
int vao_id)
{
g_return_if_fail (GSK_IS_GL_DRIVER (driver));
g_hash_table_remove (driver->vaos, GINT_TO_POINTER (vao_id));
}
static void
gsk_gl_driver_set_texture_parameters (GskGLDriver *driver,
int min_filter,
int mag_filter)
{
glTexParameteri (GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, min_filter);
glTexParameteri (GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, mag_filter);
glTexParameteri (GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
glTexParameteri (GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
}
void
gsk_gl_driver_init_texture_empty (GskGLDriver *driver,
int texture_id)
{
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
Texture *t;
g_return_if_fail (GSK_IS_GL_DRIVER (driver));
t = gsk_gl_driver_get_texture (driver, texture_id);
if (t == NULL)
{
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
g_critical ("No texture %d found.", texture_id);
return;
}
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
if (!(driver->bound_source_texture == t || driver->bound_mask_texture == t))
{
g_critical ("You must bind the texture before initializing it.");
return;
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
}
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
gsk_gl_driver_set_texture_parameters (driver, t->min_filter, t->mag_filter);
if (gdk_gl_context_get_use_es (driver->gl_context))
glTexImage2D (GL_TEXTURE_2D, 0, GL_RGBA8, t->width, t->height, 0, GL_RGBA, GL_UNSIGNED_BYTE, NULL);
else
glTexImage2D (GL_TEXTURE_2D, 0, GL_RGBA8, t->width, t->height, 0, GL_BGRA, GL_UNSIGNED_BYTE, NULL);
glBindTexture (GL_TEXTURE_2D, 0);
gsk: Add GL driver API to create render targets It's going to be used by the GL renderer to safely create and bind FBOs when needed.
2016-07-12 14:41:31 +00:00
}
gsk: Split surface upload from texture creation We're going to need a method to create textures without a surface.
2016-07-12 10:15:45 +00:00
void
gsk: Rework surface -> texture API Drop the texture parameters handling from the texture creation, and associate them with the contents upload. Also, rename the function to something more in line with what it does.
2016-07-19 13:32:17 +00:00
gsk_gl_driver_init_texture_with_surface (GskGLDriver *driver,
int texture_id,
gsk: Split surface upload from texture creation We're going to need a method to create textures without a surface.
2016-07-12 10:15:45 +00:00
cairo_surface_t *surface,
gsk: Rework surface -> texture API Drop the texture parameters handling from the texture creation, and associate them with the contents upload. Also, rename the function to something more in line with what it does.
2016-07-19 13:32:17 +00:00
int min_filter,
int mag_filter)
gsk: Split surface upload from texture creation We're going to need a method to create textures without a surface.
2016-07-12 10:15:45 +00:00
{
Texture *t;
g_return_if_fail (GSK_IS_GL_DRIVER (driver));
t = gsk_gl_driver_get_texture (driver, texture_id);
if (t == NULL)
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
{
g_critical ("No texture %d found.", texture_id);
return;
}
gsk: Split surface upload from texture creation We're going to need a method to create textures without a surface.
2016-07-12 10:15:45 +00:00
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
if (!(driver->bound_source_texture == t || driver->bound_mask_texture == t))
{
g_critical ("You must bind the texture before initializing it.");
return;
}
gsk: Rework surface -> texture API Drop the texture parameters handling from the texture creation, and associate them with the contents upload. Also, rename the function to something more in line with what it does.
2016-07-19 13:32:17 +00:00
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
gsk_gl_driver_set_texture_parameters (driver, min_filter, mag_filter);
gsk: Rework surface -> texture API Drop the texture parameters handling from the texture creation, and associate them with the contents upload. Also, rename the function to something more in line with what it does.
2016-07-19 13:32:17 +00:00
gsk: Split surface upload from texture creation We're going to need a method to create textures without a surface.
2016-07-12 10:15:45 +00:00
gdk_cairo_surface_upload_to_gl (surface, GL_TEXTURE_2D, t->width, t->height, NULL);
gsk: Rework surface -> texture API Drop the texture parameters handling from the texture creation, and associate them with the contents upload. Also, rename the function to something more in line with what it does.
2016-07-19 13:32:17 +00:00
t->min_filter = min_filter;
t->mag_filter = mag_filter;
gsk: Split surface upload from texture creation We're going to need a method to create textures without a surface.
2016-07-12 10:15:45 +00:00
if (t->min_filter != GL_NEAREST)
glGenerateMipmap (GL_TEXTURE_2D);
gsk: Move resource handling inside GskGLDriver We want the GL driver to cache as many resources as possible, so we can always ensure that we're in a consistent state, and we can handle state transitions more appropriately.
2016-07-20 16:55:35 +00:00
glBindTexture (GL_TEXTURE_2D, 0);
gsk: Split surface upload from texture creation We're going to need a method to create textures without a surface.
2016-07-12 10:15:45 +00:00
}
Reference in New Issue Copy Permalink