mutter/clutter/x11/clutter-x11-texture-pixmap.c

1511 lines
43 KiB
C
Raw Normal View History

/*
* Clutter.
*
* An OpenGL based 'interactive canvas' library.
*
* Authored By Johan Bilien <johan.bilien@nokia.com>
*
* Copyright (C) 2007 OpenedHand
*
* 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/>.
*
*
*/
/**
* SECTION:clutter-x11-texture-pixmap
* @short_description: A texture which displays the content of an X Pixmap.
*
* #ClutterX11TexturePixmap is a class for displaying the content of an
* X Pixmap as a ClutterActor. Used together with the X Composite extension,
* it allows to display the content of X Windows inside Clutter.
*
* The class uses the GLX_EXT_texture_from_pixmap OpenGL extension
* (http://people.freedesktop.org/~davidr/GLX_EXT_texture_from_pixmap.txt)
* if available
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "../clutter-marshal.h"
#include "clutter-x11-texture-pixmap.h"
#include "clutter-x11.h"
#include "clutter-backend-x11.h"
Adds initial clipped redraw support to Clutter A new (internal only currently) API, _clutter_actor_queue_clipped_redraw can be used to queue a redraw along with a clip rectangle in actor coordinates. This clip rectangle propagates up to the stage and clutter backend which may optionally use the information to optimize stage redraws. The GLX backend in particular may scissor the next redraw to the clip rectangle and use GLX_MESA_copy_sub_buffer to present the stage subregion. The intention is that any actors that can naturally determine the bounds of updates should queue clipped redraws to reduce the cost of updating small regions of the screen. Notes: » If GLX_MESA_copy_sub_buffer isn't available then the GLX backend ignores any clip rectangles. » queuing multiple clipped redraws will result in the bounding box of each clip rectangle being used. » If a clipped redraw has a height > 300 pixels then it's promoted into a full stage redraw, so that the GPU doesn't end up blocking too long waiting for the vsync to reach the optimal position to avoid tearing. » Note: no empirical data was used to come up with this threshold so we may need to tune this. » Currently only ClutterX11TexturePixmap makes use of this new API. This is done via a new "queue-damage-redraw" signal that is emitted when the pixmap is updated. The default handler queues a clipped redraw with the assumption that the pixmap is being painted as a rectangle covering the actors transformed allocation. If you subclass ClutterX11TexturePixmap and change how it's painted you now also need to override the signal handler and queue your own redraw. Technically this is a semantic break, but it's assumed that no one is currently doing this. This still leaves a few unsolved issues with regards to optimizing sub stage redraws that need to be addressed in further work so this can only be considered a stepping stone a this point: » Because we have no reliable way to determine if the painting of any given actor is being modified any optimizations implemented using _clutter_actor_queue_redraw_with_clip must be overridable by a subclass, and technically must be opt-in for existing classes to avoid a change in semantics. E.g. consider that a user connects to the paint signal for ClutterTexture and paints a circle instead of a rectangle. In this case any original logic to queue clipped redraws would be incorrect. » Currently only the implementation of an actor has enough information with which to queue clipped redraws. E.g. It is not possible for generic code in clutter-actor.c to queue a clipped redraw when hiding an actor because actors have no way to report a "paint box". (remember actors can draw outside their allocation and actors with depth may also be projected outside of their allocation) » The current plan is to add a actor_class->get_paint_cuboid() virtual so actors can report a bounding cube for everything they would draw in their current state and use that to queue clipped redraws against the stage by projecting the paint cube into stage coordinates. » Our heuristics for promoting clipped redraws into full redraws to avoid blocking the GPU while we wait for the vsync need improving: » vsync issues aren't relevant for redirected/composited applications so they should use different heuristics. In this case we instead need to trade off the cost of blitting when using glXCopySubBuffer vs promoting to a full redraw and flipping instead.
2009-11-30 17:47:55 +00:00
#include "clutter-private.h"
#include "cogl/cogl.h"
#include <X11/extensions/Xdamage.h>
#include <X11/extensions/Xcomposite.h>
#include <sys/ipc.h>
#include <sys/shm.h>
#include <sys/time.h>
#include <X11/extensions/XShm.h>
enum
{
PROP_PIXMAP = 1,
PROP_PIXMAP_WIDTH,
PROP_PIXMAP_HEIGHT,
PROP_DEPTH,
PROP_AUTO,
PROP_WINDOW,
PROP_WINDOW_REDIRECT_AUTOMATIC,
PROP_WINDOW_MAPPED,
PROP_DESTROYED,
PROP_WINDOW_X,
PROP_WINDOW_Y,
PROP_WINDOW_OVERRIDE_REDIRECT
};
enum
{
UPDATE_AREA,
Adds initial clipped redraw support to Clutter A new (internal only currently) API, _clutter_actor_queue_clipped_redraw can be used to queue a redraw along with a clip rectangle in actor coordinates. This clip rectangle propagates up to the stage and clutter backend which may optionally use the information to optimize stage redraws. The GLX backend in particular may scissor the next redraw to the clip rectangle and use GLX_MESA_copy_sub_buffer to present the stage subregion. The intention is that any actors that can naturally determine the bounds of updates should queue clipped redraws to reduce the cost of updating small regions of the screen. Notes: » If GLX_MESA_copy_sub_buffer isn't available then the GLX backend ignores any clip rectangles. » queuing multiple clipped redraws will result in the bounding box of each clip rectangle being used. » If a clipped redraw has a height > 300 pixels then it's promoted into a full stage redraw, so that the GPU doesn't end up blocking too long waiting for the vsync to reach the optimal position to avoid tearing. » Note: no empirical data was used to come up with this threshold so we may need to tune this. » Currently only ClutterX11TexturePixmap makes use of this new API. This is done via a new "queue-damage-redraw" signal that is emitted when the pixmap is updated. The default handler queues a clipped redraw with the assumption that the pixmap is being painted as a rectangle covering the actors transformed allocation. If you subclass ClutterX11TexturePixmap and change how it's painted you now also need to override the signal handler and queue your own redraw. Technically this is a semantic break, but it's assumed that no one is currently doing this. This still leaves a few unsolved issues with regards to optimizing sub stage redraws that need to be addressed in further work so this can only be considered a stepping stone a this point: » Because we have no reliable way to determine if the painting of any given actor is being modified any optimizations implemented using _clutter_actor_queue_redraw_with_clip must be overridable by a subclass, and technically must be opt-in for existing classes to avoid a change in semantics. E.g. consider that a user connects to the paint signal for ClutterTexture and paints a circle instead of a rectangle. In this case any original logic to queue clipped redraws would be incorrect. » Currently only the implementation of an actor has enough information with which to queue clipped redraws. E.g. It is not possible for generic code in clutter-actor.c to queue a clipped redraw when hiding an actor because actors have no way to report a "paint box". (remember actors can draw outside their allocation and actors with depth may also be projected outside of their allocation) » The current plan is to add a actor_class->get_paint_cuboid() virtual so actors can report a bounding cube for everything they would draw in their current state and use that to queue clipped redraws against the stage by projecting the paint cube into stage coordinates. » Our heuristics for promoting clipped redraws into full redraws to avoid blocking the GPU while we wait for the vsync need improving: » vsync issues aren't relevant for redirected/composited applications so they should use different heuristics. In this case we instead need to trade off the cost of blitting when using glXCopySubBuffer vs promoting to a full redraw and flipping instead.
2009-11-30 17:47:55 +00:00
QUEUE_DAMAGE_REDRAW,
/* FIXME: Pixmap lost signal? */
LAST_SIGNAL
};
static ClutterX11FilterReturn
on_x_event_filter (XEvent *xev, ClutterEvent *cev, gpointer data);
static void
clutter_x11_texture_pixmap_update_area_real (ClutterX11TexturePixmap *texture,
gint x,
gint y,
gint width,
gint height);
static void
clutter_x11_texture_pixmap_set_mapped (ClutterX11TexturePixmap *texture, gboolean mapped);
static void
clutter_x11_texture_pixmap_destroyed (ClutterX11TexturePixmap *texture);
static guint signals[LAST_SIGNAL] = { 0, };
struct _ClutterX11TexturePixmapPrivate
{
Window window;
Pixmap pixmap;
guint pixmap_width, pixmap_height;
guint depth;
XImage *image;
XShmSegmentInfo shminfo;
gboolean automatic_updates;
Damage damage;
Drawable damage_drawable;
/* FIXME: lots of gbooleans. coalesce into bitfields */
gboolean have_shm;
gboolean window_redirect_automatic;
gboolean window_mapped;
gboolean destroyed;
gboolean owns_pixmap;
gboolean override_redirect;
gint window_x, window_y;
};
static int _damage_event_base = 0;
/* FIXME: Ultimatly with current cogl we should subclass clutter actor */
G_DEFINE_TYPE (ClutterX11TexturePixmap, \
clutter_x11_texture_pixmap, \
CLUTTER_TYPE_TEXTURE);
static gboolean
check_extensions (ClutterX11TexturePixmap *texture)
{
int damage_error;
ClutterX11TexturePixmapPrivate *priv;
Display *dpy;
priv = texture->priv;
if (_damage_event_base)
return TRUE;
dpy = clutter_x11_get_default_display();
if (!XDamageQueryExtension (dpy,
&_damage_event_base, &damage_error))
{
g_warning ("No Damage extension");
return FALSE;
}
return TRUE;
}
static void
free_shm_resources (ClutterX11TexturePixmap *texture)
{
ClutterX11TexturePixmapPrivate *priv;
priv = texture->priv;
if (priv->shminfo.shmid != -1)
{
XShmDetach(clutter_x11_get_default_display(),
&priv->shminfo);
shmdt(priv->shminfo.shmaddr);
shmctl(priv->shminfo.shmid, IPC_RMID, 0);
priv->shminfo.shmid = -1;
}
}
/* Tries to allocate enough shared mem to handle a full size
* update size of the X Pixmap. */
static gboolean
try_alloc_shm (ClutterX11TexturePixmap *texture)
{
ClutterX11TexturePixmapPrivate *priv;
XImage *dummy_image;
Display *dpy;
priv = texture->priv;
dpy = clutter_x11_get_default_display();
g_return_val_if_fail (priv->pixmap, FALSE);
if (!XShmQueryExtension(dpy) || g_getenv("CLUTTER_X11_NO_SHM"))
{
priv->have_shm = FALSE;
return FALSE;
}
clutter_x11_trap_x_errors ();
/* We are creating a dummy_image so we can have Xlib calculate
* image->bytes_per_line - including any magic padding it may
* want - for the largest possible ximage we might need to use
* when handling updates to the texture.
*
* Note: we pass a NULL shminfo here, but that has no bearing
* on the setup of the XImage, except that ximage->obdata will
* == NULL.
*/
dummy_image =
XShmCreateImage(dpy,
DefaultVisual(dpy,
clutter_x11_get_default_screen()),
priv->depth,
ZPixmap,
NULL,
NULL, /* shminfo, */
priv->pixmap_width,
priv->pixmap_height);
if (!dummy_image)
goto failed_image_create;
priv->shminfo.shmid = shmget (IPC_PRIVATE,
dummy_image->bytes_per_line
* dummy_image->height,
IPC_CREAT | 0777);
if (priv->shminfo.shmid == -1)
goto failed_shmget;
priv->shminfo.shmaddr = shmat (priv->shminfo.shmid, 0, 0);
if (priv->shminfo.shmaddr == (void *) -1)
goto failed_shmat;
priv->shminfo.readOnly = False;
if (XShmAttach (dpy, &priv->shminfo) == 0)
goto failed_xshmattach;
if (clutter_x11_untrap_x_errors ())
g_warning ("X Error: Failed to setup XShm");
XDestroyImage (dummy_image);
priv->have_shm = TRUE;
return TRUE;
failed_xshmattach:
g_warning ("XShmAttach failed");
shmdt (priv->shminfo.shmaddr);
failed_shmat:
g_warning ("shmat failed");
shmctl (priv->shminfo.shmid, IPC_RMID, 0);
failed_shmget:
g_warning ("shmget failed");
XDestroyImage (dummy_image);
failed_image_create:
if (clutter_x11_untrap_x_errors ())
g_warning ("X Error: Failed to setup XShm");
priv->have_shm = FALSE;
return FALSE;
}
Adds initial clipped redraw support to Clutter A new (internal only currently) API, _clutter_actor_queue_clipped_redraw can be used to queue a redraw along with a clip rectangle in actor coordinates. This clip rectangle propagates up to the stage and clutter backend which may optionally use the information to optimize stage redraws. The GLX backend in particular may scissor the next redraw to the clip rectangle and use GLX_MESA_copy_sub_buffer to present the stage subregion. The intention is that any actors that can naturally determine the bounds of updates should queue clipped redraws to reduce the cost of updating small regions of the screen. Notes: » If GLX_MESA_copy_sub_buffer isn't available then the GLX backend ignores any clip rectangles. » queuing multiple clipped redraws will result in the bounding box of each clip rectangle being used. » If a clipped redraw has a height > 300 pixels then it's promoted into a full stage redraw, so that the GPU doesn't end up blocking too long waiting for the vsync to reach the optimal position to avoid tearing. » Note: no empirical data was used to come up with this threshold so we may need to tune this. » Currently only ClutterX11TexturePixmap makes use of this new API. This is done via a new "queue-damage-redraw" signal that is emitted when the pixmap is updated. The default handler queues a clipped redraw with the assumption that the pixmap is being painted as a rectangle covering the actors transformed allocation. If you subclass ClutterX11TexturePixmap and change how it's painted you now also need to override the signal handler and queue your own redraw. Technically this is a semantic break, but it's assumed that no one is currently doing this. This still leaves a few unsolved issues with regards to optimizing sub stage redraws that need to be addressed in further work so this can only be considered a stepping stone a this point: » Because we have no reliable way to determine if the painting of any given actor is being modified any optimizations implemented using _clutter_actor_queue_redraw_with_clip must be overridable by a subclass, and technically must be opt-in for existing classes to avoid a change in semantics. E.g. consider that a user connects to the paint signal for ClutterTexture and paints a circle instead of a rectangle. In this case any original logic to queue clipped redraws would be incorrect. » Currently only the implementation of an actor has enough information with which to queue clipped redraws. E.g. It is not possible for generic code in clutter-actor.c to queue a clipped redraw when hiding an actor because actors have no way to report a "paint box". (remember actors can draw outside their allocation and actors with depth may also be projected outside of their allocation) » The current plan is to add a actor_class->get_paint_cuboid() virtual so actors can report a bounding cube for everything they would draw in their current state and use that to queue clipped redraws against the stage by projecting the paint cube into stage coordinates. » Our heuristics for promoting clipped redraws into full redraws to avoid blocking the GPU while we wait for the vsync need improving: » vsync issues aren't relevant for redirected/composited applications so they should use different heuristics. In this case we instead need to trade off the cost of blitting when using glXCopySubBuffer vs promoting to a full redraw and flipping instead.
2009-11-30 17:47:55 +00:00
static void
check_for_pixmap_damage (ClutterX11TexturePixmap *texture)
{
ClutterX11TexturePixmapPrivate *priv = texture->priv;
Display *dpy;
XserverRegion parts;
int i, r_count;
XRectangle *r_damage;
XRectangle r_bounds;
clutter_x11_trap_x_errors ();
/*
* Retrieve the damaged region and break it down into individual
* rectangles so we do not have to update the whole shebang.
*/
dpy = clutter_x11_get_default_display();
parts = XFixesCreateRegion (dpy, 0, 0);
XDamageSubtract (dpy, priv->damage, None, parts);
r_damage = XFixesFetchRegionAndBounds (dpy,
parts,
&r_count,
&r_bounds);
clutter_x11_untrap_x_errors ();
if (r_damage)
{
for (i = 0; i < r_count; ++i)
clutter_x11_texture_pixmap_update_area (texture,
r_damage[i].x,
r_damage[i].y,
r_damage[i].width,
r_damage[i].height);
XFree (r_damage);
}
XFixesDestroyRegion (dpy, parts);
}
static ClutterX11FilterReturn
on_x_event_filter (XEvent *xev, ClutterEvent *cev, gpointer data)
{
ClutterX11TexturePixmap *texture;
ClutterX11TexturePixmapPrivate *priv;
texture = CLUTTER_X11_TEXTURE_PIXMAP (data);
g_return_val_if_fail (CLUTTER_X11_IS_TEXTURE_PIXMAP (texture), \
CLUTTER_X11_FILTER_CONTINUE);
priv = texture->priv;
if (xev->type == _damage_event_base + XDamageNotify)
{
XDamageNotifyEvent *dev = (XDamageNotifyEvent*)xev;
if (dev->drawable != priv->damage_drawable)
return CLUTTER_X11_FILTER_CONTINUE;
Adds initial clipped redraw support to Clutter A new (internal only currently) API, _clutter_actor_queue_clipped_redraw can be used to queue a redraw along with a clip rectangle in actor coordinates. This clip rectangle propagates up to the stage and clutter backend which may optionally use the information to optimize stage redraws. The GLX backend in particular may scissor the next redraw to the clip rectangle and use GLX_MESA_copy_sub_buffer to present the stage subregion. The intention is that any actors that can naturally determine the bounds of updates should queue clipped redraws to reduce the cost of updating small regions of the screen. Notes: » If GLX_MESA_copy_sub_buffer isn't available then the GLX backend ignores any clip rectangles. » queuing multiple clipped redraws will result in the bounding box of each clip rectangle being used. » If a clipped redraw has a height > 300 pixels then it's promoted into a full stage redraw, so that the GPU doesn't end up blocking too long waiting for the vsync to reach the optimal position to avoid tearing. » Note: no empirical data was used to come up with this threshold so we may need to tune this. » Currently only ClutterX11TexturePixmap makes use of this new API. This is done via a new "queue-damage-redraw" signal that is emitted when the pixmap is updated. The default handler queues a clipped redraw with the assumption that the pixmap is being painted as a rectangle covering the actors transformed allocation. If you subclass ClutterX11TexturePixmap and change how it's painted you now also need to override the signal handler and queue your own redraw. Technically this is a semantic break, but it's assumed that no one is currently doing this. This still leaves a few unsolved issues with regards to optimizing sub stage redraws that need to be addressed in further work so this can only be considered a stepping stone a this point: » Because we have no reliable way to determine if the painting of any given actor is being modified any optimizations implemented using _clutter_actor_queue_redraw_with_clip must be overridable by a subclass, and technically must be opt-in for existing classes to avoid a change in semantics. E.g. consider that a user connects to the paint signal for ClutterTexture and paints a circle instead of a rectangle. In this case any original logic to queue clipped redraws would be incorrect. » Currently only the implementation of an actor has enough information with which to queue clipped redraws. E.g. It is not possible for generic code in clutter-actor.c to queue a clipped redraw when hiding an actor because actors have no way to report a "paint box". (remember actors can draw outside their allocation and actors with depth may also be projected outside of their allocation) » The current plan is to add a actor_class->get_paint_cuboid() virtual so actors can report a bounding cube for everything they would draw in their current state and use that to queue clipped redraws against the stage by projecting the paint cube into stage coordinates. » Our heuristics for promoting clipped redraws into full redraws to avoid blocking the GPU while we wait for the vsync need improving: » vsync issues aren't relevant for redirected/composited applications so they should use different heuristics. In this case we instead need to trade off the cost of blitting when using glXCopySubBuffer vs promoting to a full redraw and flipping instead.
2009-11-30 17:47:55 +00:00
check_for_pixmap_damage (texture);
}
return CLUTTER_X11_FILTER_CONTINUE;
}
static ClutterX11FilterReturn
on_x_event_filter_too (XEvent *xev, ClutterEvent *cev, gpointer data)
{
ClutterX11TexturePixmap *texture;
ClutterX11TexturePixmapPrivate *priv;
texture = CLUTTER_X11_TEXTURE_PIXMAP (data);
g_return_val_if_fail (CLUTTER_X11_IS_TEXTURE_PIXMAP (texture), \
CLUTTER_X11_FILTER_CONTINUE);
priv = texture->priv;
if (xev->xany.window != priv->window)
return CLUTTER_X11_FILTER_CONTINUE;
switch (xev->type) {
case MapNotify:
case ConfigureNotify:
clutter_x11_texture_pixmap_sync_window (texture);
break;
case UnmapNotify:
clutter_x11_texture_pixmap_set_mapped (texture, FALSE);
break;
case DestroyNotify:
clutter_x11_texture_pixmap_destroyed (texture);
break;
default:
break;
}
return CLUTTER_X11_FILTER_CONTINUE;
}
static void
create_damage_resources (ClutterX11TexturePixmap *texture)
{
ClutterX11TexturePixmapPrivate *priv;
Display *dpy;
priv = texture->priv;
dpy = clutter_x11_get_default_display();
if (!priv->window && !priv->pixmap)
return;
clutter_x11_trap_x_errors ();
if (priv->window)
priv->damage_drawable = priv->window;
else
priv->damage_drawable = priv->pixmap;
priv->damage = XDamageCreate (dpy,
priv->damage_drawable,
XDamageReportNonEmpty);
/* Errors here might occur if the window is already destroyed, we
* simply skip processing damage and assume that the texture pixmap
* will be cleaned up by the app when it gets a DestroyNotify.
*/
XSync (dpy, FALSE);
clutter_x11_untrap_x_errors ();
if (priv->damage)
clutter_x11_add_filter (on_x_event_filter, (gpointer)texture);
}
static void
free_damage_resources (ClutterX11TexturePixmap *texture)
{
ClutterX11TexturePixmapPrivate *priv;
Display *dpy;
priv = texture->priv;
dpy = clutter_x11_get_default_display();
if (priv->damage)
{
clutter_x11_trap_x_errors ();
XDamageDestroy (dpy, priv->damage);
XSync (dpy, FALSE);
clutter_x11_untrap_x_errors ();
priv->damage = None;
priv->damage_drawable = None;
clutter_x11_remove_filter (on_x_event_filter, (gpointer)texture);
}
}
Adds initial clipped redraw support to Clutter A new (internal only currently) API, _clutter_actor_queue_clipped_redraw can be used to queue a redraw along with a clip rectangle in actor coordinates. This clip rectangle propagates up to the stage and clutter backend which may optionally use the information to optimize stage redraws. The GLX backend in particular may scissor the next redraw to the clip rectangle and use GLX_MESA_copy_sub_buffer to present the stage subregion. The intention is that any actors that can naturally determine the bounds of updates should queue clipped redraws to reduce the cost of updating small regions of the screen. Notes: » If GLX_MESA_copy_sub_buffer isn't available then the GLX backend ignores any clip rectangles. » queuing multiple clipped redraws will result in the bounding box of each clip rectangle being used. » If a clipped redraw has a height > 300 pixels then it's promoted into a full stage redraw, so that the GPU doesn't end up blocking too long waiting for the vsync to reach the optimal position to avoid tearing. » Note: no empirical data was used to come up with this threshold so we may need to tune this. » Currently only ClutterX11TexturePixmap makes use of this new API. This is done via a new "queue-damage-redraw" signal that is emitted when the pixmap is updated. The default handler queues a clipped redraw with the assumption that the pixmap is being painted as a rectangle covering the actors transformed allocation. If you subclass ClutterX11TexturePixmap and change how it's painted you now also need to override the signal handler and queue your own redraw. Technically this is a semantic break, but it's assumed that no one is currently doing this. This still leaves a few unsolved issues with regards to optimizing sub stage redraws that need to be addressed in further work so this can only be considered a stepping stone a this point: » Because we have no reliable way to determine if the painting of any given actor is being modified any optimizations implemented using _clutter_actor_queue_redraw_with_clip must be overridable by a subclass, and technically must be opt-in for existing classes to avoid a change in semantics. E.g. consider that a user connects to the paint signal for ClutterTexture and paints a circle instead of a rectangle. In this case any original logic to queue clipped redraws would be incorrect. » Currently only the implementation of an actor has enough information with which to queue clipped redraws. E.g. It is not possible for generic code in clutter-actor.c to queue a clipped redraw when hiding an actor because actors have no way to report a "paint box". (remember actors can draw outside their allocation and actors with depth may also be projected outside of their allocation) » The current plan is to add a actor_class->get_paint_cuboid() virtual so actors can report a bounding cube for everything they would draw in their current state and use that to queue clipped redraws against the stage by projecting the paint cube into stage coordinates. » Our heuristics for promoting clipped redraws into full redraws to avoid blocking the GPU while we wait for the vsync need improving: » vsync issues aren't relevant for redirected/composited applications so they should use different heuristics. In this case we instead need to trade off the cost of blitting when using glXCopySubBuffer vs promoting to a full redraw and flipping instead.
2009-11-30 17:47:55 +00:00
static void
clutter_x11_texture_pixmap_real_queue_damage_redraw (
ClutterX11TexturePixmap *texture,
gint x,
gint y,
gint width,
gint height)
{
ClutterActor *self = CLUTTER_ACTOR (texture);
ClutterActorBox allocation;
guint pixmap_width = 0;
guint pixmap_height = 0;
float scale_x;
float scale_y;
ClutterActorBox clip;
/* NB: clutter_actor_queue_clipped_redraw expects a box in the actor's
* coordinate space so we need to convert from pixmap coordinates to
* actor coordinates...
*/
/* XXX: we don't care if we get an out of date allocation here because
* clutter_actor_queue_clipped_redraw knows to ignore the clip if the
* actor's allocation is invalid.
*
* This is noted because clutter_actor_get_allocation_box does some
* unnecessary work to support buggy code with a comment suggesting that
* it could be changed later which would be good for this use case!
*/
clutter_actor_get_allocation_box (self, &allocation);
g_object_get (self,
"pixmap-width", &pixmap_width,
"pixmap-height", &pixmap_height,
NULL);
scale_x = (allocation.x2 - allocation.x1) / pixmap_width;
scale_y = (allocation.y2 - allocation.y1) / pixmap_height;
clip.x1 = x * scale_x;
clip.y1 = y * scale_y;
clip.x2 = clip.x1 + width * scale_x;
clip.y2 = clip.y1 + height * scale_y;
_clutter_actor_queue_redraw_with_clip (self,
CLUTTER_REDRAW_CLIPPED_TO_BOX,
&clip);
}
static void
clutter_x11_texture_pixmap_init (ClutterX11TexturePixmap *self)
{
self->priv =
G_TYPE_INSTANCE_GET_PRIVATE (self,
CLUTTER_X11_TYPE_TEXTURE_PIXMAP,
ClutterX11TexturePixmapPrivate);
if (!check_extensions (self))
{
/* FIMXE: means display lacks needed extensions for at least auto.
* - a _can_autoupdate() method ?
*/
}
self->priv->image = NULL;
self->priv->automatic_updates = FALSE;
self->priv->damage = None;
self->priv->damage_drawable = None;
self->priv->window = None;
self->priv->pixmap = None;
self->priv->pixmap_height = 0;
self->priv->pixmap_width = 0;
self->priv->shminfo.shmid = -1;
self->priv->window_redirect_automatic = TRUE;
self->priv->window_mapped = FALSE;
self->priv->destroyed = FALSE;
self->priv->override_redirect = FALSE;
self->priv->window_x = 0;
self->priv->window_y = 0;
}
static void
clutter_x11_texture_pixmap_dispose (GObject *object)
{
ClutterX11TexturePixmap *texture = CLUTTER_X11_TEXTURE_PIXMAP (object);
ClutterX11TexturePixmapPrivate *priv = texture->priv;
free_damage_resources (texture);
clutter_x11_remove_filter (on_x_event_filter_too, (gpointer)texture);
if (priv->owns_pixmap && priv->pixmap)
{
XFreePixmap (clutter_x11_get_default_display (), priv->pixmap);
priv->pixmap = None;
}
if (priv->image)
{
XDestroyImage (priv->image);
priv->image = NULL;
}
free_shm_resources (texture);
G_OBJECT_CLASS (clutter_x11_texture_pixmap_parent_class)->dispose (object);
}
static void
clutter_x11_texture_pixmap_set_property (GObject *object,
guint prop_id,
const GValue *value,
GParamSpec *pspec)
{
ClutterX11TexturePixmap *texture = CLUTTER_X11_TEXTURE_PIXMAP (object);
ClutterX11TexturePixmapPrivate *priv = texture->priv;
switch (prop_id)
{
case PROP_PIXMAP:
clutter_x11_texture_pixmap_set_pixmap (texture,
g_value_get_ulong (value));
break;
case PROP_AUTO:
clutter_x11_texture_pixmap_set_automatic (texture,
g_value_get_boolean (value));
break;
case PROP_WINDOW:
clutter_x11_texture_pixmap_set_window (texture,
g_value_get_ulong (value),
priv->window_redirect_automatic);
break;
case PROP_WINDOW_REDIRECT_AUTOMATIC:
{
gboolean new;
new = g_value_get_boolean (value);
/* Change the update mode.. */
if (new != priv->window_redirect_automatic && priv->window)
clutter_x11_texture_pixmap_set_window (texture, priv->window, new);
priv->window_redirect_automatic = new;
}
break;
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
break;
}
}
static void
clutter_x11_texture_pixmap_get_property (GObject *object,
guint prop_id,
GValue *value,
GParamSpec *pspec)
{
ClutterX11TexturePixmap *texture = CLUTTER_X11_TEXTURE_PIXMAP (object);
ClutterX11TexturePixmapPrivate *priv = texture->priv;
switch (prop_id)
{
case PROP_PIXMAP:
g_value_set_ulong (value, priv->pixmap);
break;
case PROP_PIXMAP_WIDTH:
g_value_set_uint (value, priv->pixmap_width);
break;
case PROP_PIXMAP_HEIGHT:
g_value_set_uint (value, priv->pixmap_height);
break;
case PROP_DEPTH:
g_value_set_uint (value, priv->depth);
break;
case PROP_AUTO:
g_value_set_boolean (value, priv->automatic_updates);
break;
case PROP_WINDOW:
g_value_set_ulong (value, priv->window);
break;
case PROP_WINDOW_REDIRECT_AUTOMATIC:
g_value_set_boolean (value, priv->window_redirect_automatic);
break;
case PROP_WINDOW_MAPPED:
g_value_set_boolean (value, priv->window_mapped);
break;
case PROP_DESTROYED:
g_value_set_boolean (value, priv->destroyed);
break;
case PROP_WINDOW_X:
g_value_set_int (value, priv->window_x);
break;
case PROP_WINDOW_Y:
g_value_set_int (value, priv->window_y);
break;
case PROP_WINDOW_OVERRIDE_REDIRECT:
g_value_set_boolean (value, priv->override_redirect);
break;
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
break;
}
}
static void
clutter_x11_texture_pixmap_realize (ClutterActor *actor)
{
ClutterX11TexturePixmap *texture = CLUTTER_X11_TEXTURE_PIXMAP (actor);
ClutterX11TexturePixmapPrivate *priv = texture->priv;
CLUTTER_ACTOR_CLASS (clutter_x11_texture_pixmap_parent_class)->
realize (actor);
clutter_x11_texture_pixmap_update_area_real (texture,
0, 0,
priv->pixmap_width,
priv->pixmap_height);
}
static void
clutter_x11_texture_pixmap_class_init (ClutterX11TexturePixmapClass *klass)
{
GObjectClass *object_class = G_OBJECT_CLASS (klass);
ClutterActorClass *actor_class = CLUTTER_ACTOR_CLASS (klass);
GParamSpec *pspec;
ClutterBackend *default_backend;
g_type_class_add_private (klass, sizeof (ClutterX11TexturePixmapPrivate));
object_class->dispose = clutter_x11_texture_pixmap_dispose;
object_class->set_property = clutter_x11_texture_pixmap_set_property;
object_class->get_property = clutter_x11_texture_pixmap_get_property;
actor_class->realize = clutter_x11_texture_pixmap_realize;
klass->update_area = clutter_x11_texture_pixmap_update_area_real;
pspec = g_param_spec_ulong ("pixmap",
"Pixmap",
"The X11 Pixmap to be bound",
0, G_MAXULONG,
None,
G_PARAM_READWRITE);
g_object_class_install_property (object_class, PROP_PIXMAP, pspec);
pspec = g_param_spec_uint ("pixmap-width",
"Pixmap width",
"The width of the "
"pixmap bound to this texture",
0, G_MAXUINT,
0,
G_PARAM_READABLE);
g_object_class_install_property (object_class, PROP_PIXMAP_WIDTH, pspec);
pspec = g_param_spec_uint ("pixmap-height",
"Pixmap height",
"The height of the "
"pixmap bound to this texture",
0, G_MAXUINT,
0,
G_PARAM_READABLE);
g_object_class_install_property (object_class, PROP_PIXMAP_HEIGHT, pspec);
pspec = g_param_spec_uint ("pixmap-depth",
"Pixmap Depth",
"The depth (in number of bits) of the "
"pixmap bound to this texture",
0, G_MAXUINT,
0,
G_PARAM_READABLE);
g_object_class_install_property (object_class, PROP_DEPTH, pspec);
pspec = g_param_spec_boolean ("automatic-updates",
"Automatic Updates",
"If the texture should be kept in "
"sync with any pixmap changes.",
FALSE,
G_PARAM_READWRITE);
g_object_class_install_property (object_class, PROP_AUTO, pspec);
pspec = g_param_spec_ulong ("window",
"Window",
"The X11 Window to be bound",
0, G_MAXULONG,
None,
G_PARAM_READWRITE);
g_object_class_install_property (object_class, PROP_WINDOW, pspec);
pspec = g_param_spec_boolean ("window-redirect-automatic",
"Window Redirect Automatic",
"If composite window redirects are set to "
"Automatic (or Manual if false)",
TRUE,
G_PARAM_READWRITE);
g_object_class_install_property (object_class,
PROP_WINDOW_REDIRECT_AUTOMATIC, pspec);
pspec = g_param_spec_boolean ("window-mapped",
"Window Mapped",
"If window is mapped",
FALSE,
G_PARAM_READABLE);
g_object_class_install_property (object_class,
PROP_WINDOW_MAPPED, pspec);
pspec = g_param_spec_boolean ("destroyed",
"Destroyed",
"If window has been destroyed",
FALSE,
G_PARAM_READABLE);
g_object_class_install_property (object_class,
PROP_DESTROYED, pspec);
pspec = g_param_spec_int ("window-x",
"Window X",
"X position of window on screen according to X11",
G_MININT, G_MAXINT, 0, G_PARAM_READABLE);
g_object_class_install_property (object_class,
PROP_WINDOW_X, pspec);
pspec = g_param_spec_int ("window-y",
"Window Y",
"Y position of window on screen according to X11",
G_MININT, G_MAXINT, 0, G_PARAM_READABLE);
g_object_class_install_property (object_class,
PROP_WINDOW_Y, pspec);
pspec = g_param_spec_boolean ("window-override-redirect",
"Window Override Redirect",
"If this is an override-redirect window",
FALSE,
G_PARAM_READABLE);
g_object_class_install_property (object_class,
PROP_WINDOW_OVERRIDE_REDIRECT, pspec);
/**
* ClutterX11TexturePixmap::update-area:
* @texture: the object which received the signal
*
* The ::hide signal is emitted to ask the texture to update its
* content from its source pixmap.
*
* Since: 0.8
*/
signals[UPDATE_AREA] =
g_signal_new ("update-area",
G_TYPE_FROM_CLASS (object_class),
G_SIGNAL_RUN_FIRST,
G_STRUCT_OFFSET (ClutterX11TexturePixmapClass, \
update_area),
NULL, NULL,
clutter_marshal_VOID__INT_INT_INT_INT,
G_TYPE_NONE, 4,
G_TYPE_INT,
G_TYPE_INT,
G_TYPE_INT,
G_TYPE_INT);
Adds initial clipped redraw support to Clutter A new (internal only currently) API, _clutter_actor_queue_clipped_redraw can be used to queue a redraw along with a clip rectangle in actor coordinates. This clip rectangle propagates up to the stage and clutter backend which may optionally use the information to optimize stage redraws. The GLX backend in particular may scissor the next redraw to the clip rectangle and use GLX_MESA_copy_sub_buffer to present the stage subregion. The intention is that any actors that can naturally determine the bounds of updates should queue clipped redraws to reduce the cost of updating small regions of the screen. Notes: » If GLX_MESA_copy_sub_buffer isn't available then the GLX backend ignores any clip rectangles. » queuing multiple clipped redraws will result in the bounding box of each clip rectangle being used. » If a clipped redraw has a height > 300 pixels then it's promoted into a full stage redraw, so that the GPU doesn't end up blocking too long waiting for the vsync to reach the optimal position to avoid tearing. » Note: no empirical data was used to come up with this threshold so we may need to tune this. » Currently only ClutterX11TexturePixmap makes use of this new API. This is done via a new "queue-damage-redraw" signal that is emitted when the pixmap is updated. The default handler queues a clipped redraw with the assumption that the pixmap is being painted as a rectangle covering the actors transformed allocation. If you subclass ClutterX11TexturePixmap and change how it's painted you now also need to override the signal handler and queue your own redraw. Technically this is a semantic break, but it's assumed that no one is currently doing this. This still leaves a few unsolved issues with regards to optimizing sub stage redraws that need to be addressed in further work so this can only be considered a stepping stone a this point: » Because we have no reliable way to determine if the painting of any given actor is being modified any optimizations implemented using _clutter_actor_queue_redraw_with_clip must be overridable by a subclass, and technically must be opt-in for existing classes to avoid a change in semantics. E.g. consider that a user connects to the paint signal for ClutterTexture and paints a circle instead of a rectangle. In this case any original logic to queue clipped redraws would be incorrect. » Currently only the implementation of an actor has enough information with which to queue clipped redraws. E.g. It is not possible for generic code in clutter-actor.c to queue a clipped redraw when hiding an actor because actors have no way to report a "paint box". (remember actors can draw outside their allocation and actors with depth may also be projected outside of their allocation) » The current plan is to add a actor_class->get_paint_cuboid() virtual so actors can report a bounding cube for everything they would draw in their current state and use that to queue clipped redraws against the stage by projecting the paint cube into stage coordinates. » Our heuristics for promoting clipped redraws into full redraws to avoid blocking the GPU while we wait for the vsync need improving: » vsync issues aren't relevant for redirected/composited applications so they should use different heuristics. In this case we instead need to trade off the cost of blitting when using glXCopySubBuffer vs promoting to a full redraw and flipping instead.
2009-11-30 17:47:55 +00:00
/**
* ClutterX11TexturePixmap::queue-damage-redraw
* @texture: the object which received the signal
* @x: The top left x position of the damage region
* @y: The top left y position of the damage region
* @width: The width of the damage region
* @height: The height of the damage region
*
* ::queue-damage-redraw is emitted to notify that some sub-region of the
* underlying pixmap has changed and you need to queue a
* corresponding redraw for the actor.
*
* The default handler will queue a clipped redraw in response to
* the damage, using the assumption that the pixmap is being painted
* to a rectangle covering the transformed allocation of the actor.
* If you sub-class and change the paint method so this isn't true
* then you must also provide your own damage signal handler to
* queue a redraw that blocks this default behaviour.
*
* Since: 1.2
*/
signals[QUEUE_DAMAGE_REDRAW] =
g_signal_new (g_intern_static_string ("queue-damage-redraw"),
G_TYPE_FROM_CLASS (object_class),
G_SIGNAL_RUN_FIRST,
0,
NULL, NULL,
clutter_marshal_VOID__INT_INT_INT_INT,
G_TYPE_NONE, 4,
G_TYPE_INT,
G_TYPE_INT,
G_TYPE_INT,
G_TYPE_INT);
g_signal_override_class_handler ("queue-damage-redraw",
CLUTTER_X11_TYPE_TEXTURE_PIXMAP,
G_CALLBACK (clutter_x11_texture_pixmap_real_queue_damage_redraw));
Adds initial clipped redraw support to Clutter A new (internal only currently) API, _clutter_actor_queue_clipped_redraw can be used to queue a redraw along with a clip rectangle in actor coordinates. This clip rectangle propagates up to the stage and clutter backend which may optionally use the information to optimize stage redraws. The GLX backend in particular may scissor the next redraw to the clip rectangle and use GLX_MESA_copy_sub_buffer to present the stage subregion. The intention is that any actors that can naturally determine the bounds of updates should queue clipped redraws to reduce the cost of updating small regions of the screen. Notes: » If GLX_MESA_copy_sub_buffer isn't available then the GLX backend ignores any clip rectangles. » queuing multiple clipped redraws will result in the bounding box of each clip rectangle being used. » If a clipped redraw has a height > 300 pixels then it's promoted into a full stage redraw, so that the GPU doesn't end up blocking too long waiting for the vsync to reach the optimal position to avoid tearing. » Note: no empirical data was used to come up with this threshold so we may need to tune this. » Currently only ClutterX11TexturePixmap makes use of this new API. This is done via a new "queue-damage-redraw" signal that is emitted when the pixmap is updated. The default handler queues a clipped redraw with the assumption that the pixmap is being painted as a rectangle covering the actors transformed allocation. If you subclass ClutterX11TexturePixmap and change how it's painted you now also need to override the signal handler and queue your own redraw. Technically this is a semantic break, but it's assumed that no one is currently doing this. This still leaves a few unsolved issues with regards to optimizing sub stage redraws that need to be addressed in further work so this can only be considered a stepping stone a this point: » Because we have no reliable way to determine if the painting of any given actor is being modified any optimizations implemented using _clutter_actor_queue_redraw_with_clip must be overridable by a subclass, and technically must be opt-in for existing classes to avoid a change in semantics. E.g. consider that a user connects to the paint signal for ClutterTexture and paints a circle instead of a rectangle. In this case any original logic to queue clipped redraws would be incorrect. » Currently only the implementation of an actor has enough information with which to queue clipped redraws. E.g. It is not possible for generic code in clutter-actor.c to queue a clipped redraw when hiding an actor because actors have no way to report a "paint box". (remember actors can draw outside their allocation and actors with depth may also be projected outside of their allocation) » The current plan is to add a actor_class->get_paint_cuboid() virtual so actors can report a bounding cube for everything they would draw in their current state and use that to queue clipped redraws against the stage by projecting the paint cube into stage coordinates. » Our heuristics for promoting clipped redraws into full redraws to avoid blocking the GPU while we wait for the vsync need improving: » vsync issues aren't relevant for redirected/composited applications so they should use different heuristics. In this case we instead need to trade off the cost of blitting when using glXCopySubBuffer vs promoting to a full redraw and flipping instead.
2009-11-30 17:47:55 +00:00
default_backend = clutter_get_default_backend ();
if (!CLUTTER_IS_BACKEND_X11 (default_backend))
{
g_critical ("ClutterX11TexturePixmap instanciated with a "
"non-X11 backend");
return;
}
}
static void
clutter_x11_texture_pixmap_update_area_real (ClutterX11TexturePixmap *texture,
gint x,
gint y,
gint width,
gint height)
{
ClutterX11TexturePixmapPrivate *priv;
Display *dpy;
XImage *image;
char *first_pixel;
GError *error = NULL;
guint bytes_per_line;
char *data;
gboolean data_allocated = FALSE;
int err_code;
if (!CLUTTER_ACTOR_IS_REALIZED (texture))
return;
priv = texture->priv;
dpy = clutter_x11_get_default_display();
if (!priv->pixmap)
return;
if (priv->shminfo.shmid == -1)
try_alloc_shm (texture);
clutter_x11_trap_x_errors ();
if (priv->have_shm)
{
image =
XShmCreateImage(dpy,
DefaultVisual(dpy,
clutter_x11_get_default_screen()),
priv->depth,
ZPixmap,
NULL,
&priv->shminfo,
width,
height);
image->data = priv->shminfo.shmaddr;
XShmGetImage (dpy, priv->pixmap, image, x, y, AllPlanes);
first_pixel = image->data;
}
else
{
if (!priv->image)
{
priv->image = XGetImage (dpy,
priv->pixmap,
0, 0,
priv->pixmap_width, priv->pixmap_height,
AllPlanes,
ZPixmap);
first_pixel = priv->image->data + priv->image->bytes_per_line * y
+ x * priv->image->bits_per_pixel/8;
}
else
{
XGetSubImage (dpy,
priv->pixmap,
x, y,
width, height,
AllPlanes,
ZPixmap,
priv->image,
x, y);
first_pixel = priv->image->data + priv->image->bytes_per_line * y
+ x * priv->image->bits_per_pixel/8;
}
image = priv->image;
}
XSync (dpy, FALSE);
if ((err_code = clutter_x11_untrap_x_errors ()))
{
g_warning ("Failed to get XImage of pixmap: %lx, removing",
priv->pixmap);
/* safe to assume pixmap has gone away? - therefor reset */
clutter_x11_texture_pixmap_set_pixmap (texture, None);
return;
}
if (priv->depth == 24)
{
guint xpos, ypos;
for (ypos=0; ypos<height; ypos++)
for (xpos=0; xpos<width; xpos++)
{
char *p = first_pixel + image->bytes_per_line*ypos
+ xpos * 4;
p[3] = 0xFF;
}
data = first_pixel;
bytes_per_line = image->bytes_per_line;
}
else if (priv->depth == 16)
{
guint xpos, ypos;
data = g_malloc (height * width * 4);
data_allocated = TRUE;
bytes_per_line = width * 4;
for (ypos=0; ypos<height; ypos++)
for (xpos=0; xpos<width; xpos++)
{
char *src_p = first_pixel + image->bytes_per_line * ypos
+ xpos * 2;
guint16 *src_pixel = (guint16 *)src_p;
char *dst_p = data + bytes_per_line * ypos + xpos * 4;
guint32 *dst_pixel = (guint32 *)dst_p;
*dst_pixel =
((((*src_pixel << 3) & 0xf8) | ((*src_pixel >> 2) & 0x7)) | \
(((*src_pixel << 5) & 0xfc00) | ((*src_pixel >> 1) & 0x300)) | \
(((*src_pixel << 8) & 0xf80000) | ((*src_pixel << 3) & 0x70000)))
| 0xff000000;
}
}
else if (priv->depth == 32)
{
bytes_per_line = image->bytes_per_line;
data = first_pixel;
}
else
return;
/* For debugging purposes, un comment to simply generate dummy
* pixmap data. (A Green background and Blue cross) */
#if 0
{
guint xpos, ypos;
if (data_allocated)
g_free (data);
data_allocated = TRUE;
data = g_malloc (width*height*4);
bytes_per_line = width *4;
for (ypos=0; ypos<height; ypos++)
for (xpos=0; xpos<width; xpos++)
{
char *p = data + width*4*ypos + xpos * 4;
guint32 *pixel = (guint32 *)p;
if ((xpos > width/2 && xpos <= (width/2) + width/4)
|| (ypos > height/2 && ypos <= (height/2) + height/4))
*pixel=0xff0000ff;
else
*pixel=0xff00ff00;
}
}
#endif
if (x != 0 || y != 0 ||
width != priv->pixmap_width || height != priv->pixmap_height)
clutter_texture_set_area_from_rgb_data (CLUTTER_TEXTURE (texture),
(guint8 *)data,
TRUE,
x, y,
width, height,
bytes_per_line,
4,
CLUTTER_TEXTURE_RGB_FLAG_BGR |
CLUTTER_TEXTURE_RGB_FLAG_PREMULT,
&error);
else
clutter_texture_set_from_rgb_data (CLUTTER_TEXTURE (texture),
(guint8 *)data,
TRUE,
width, height,
bytes_per_line,
4,
CLUTTER_TEXTURE_RGB_FLAG_BGR |
CLUTTER_TEXTURE_RGB_FLAG_PREMULT,
&error);
if (error)
{
g_warning ("Error when uploading from pixbuf: %s",
error->message);
g_error_free (error);
}
if (data_allocated)
g_free (data);
if (priv->have_shm)
XFree (image);
}
/**
* clutter_x11_texture_pixmap_new:
*
* Creates a new #ClutterX11TexturePixmap which can be used to display the
* contents of an X11 Pixmap inside a Clutter scene graph
*
* Return value: A new #ClutterX11TexturePixmap
*
* Since: 0.8
*/
ClutterActor *
clutter_x11_texture_pixmap_new (void)
{
ClutterActor *actor;
actor = g_object_new (CLUTTER_X11_TYPE_TEXTURE_PIXMAP, NULL);
return actor;
}
/**
* clutter_x11_texture_pixmap_new_with_pixmap:
* @pixmap: the X Pixmap to which this texture should be bound
*
* Creates a new #ClutterX11TexturePixmap for @pixmap
*
* Return value: A new #ClutterX11TexturePixmap bound to the given X Pixmap
*
* Since: 0.8
*/
ClutterActor *
clutter_x11_texture_pixmap_new_with_pixmap (Pixmap pixmap)
{
ClutterActor *actor;
actor = g_object_new (CLUTTER_X11_TYPE_TEXTURE_PIXMAP,
"pixmap", pixmap,
NULL);
return actor;
}
/**
* clutter_x11_texture_pixmap_new_with_window:
* @window: the X window to which this texture should be bound
*
* Creates a new #ClutterX11TexturePixmap for @window
*
* Return value: A new #ClutterX11TexturePixmap bound to the given X window.
*
* Since: 0.8
**/
ClutterActor *
clutter_x11_texture_pixmap_new_with_window (Window window)
{
ClutterActor *actor;
actor = g_object_new (CLUTTER_X11_TYPE_TEXTURE_PIXMAP,
"window", window,
NULL);
return actor;
}
/**
* clutter_x11_texture_pixmap_set_pixmap:
* @texture: the texture to bind
* @pixmap: the X Pixmap to which the texture should be bound
*
* Sets the X Pixmap to which the texture should be bound.
*
* Since: 0.8
*/
void
clutter_x11_texture_pixmap_set_pixmap (ClutterX11TexturePixmap *texture,
Pixmap pixmap)
{
Window root;
int x, y;
unsigned int width, height, border_width, depth;
Status status = 0;
gboolean new_pixmap = FALSE, new_pixmap_width = FALSE;
gboolean new_pixmap_height = FALSE, new_pixmap_depth = FALSE;
ClutterX11TexturePixmapPrivate *priv;
g_return_if_fail (CLUTTER_X11_IS_TEXTURE_PIXMAP (texture));
priv = texture->priv;
clutter_x11_trap_x_errors ();
status = XGetGeometry (clutter_x11_get_default_display(),
(Drawable)pixmap,
&root,
&x,
&y,
&width,
&height,
&border_width,
&depth);
if (clutter_x11_untrap_x_errors () || status == 0)
{
if (pixmap != None)
g_warning ("Unable to query pixmap: %lx", pixmap);
pixmap = None;
width = height = depth = 0;
}
if (priv->image)
{
XDestroyImage (priv->image);
priv->image = NULL;
}
if (priv->pixmap != pixmap)
{
if (priv->pixmap && priv->owns_pixmap)
XFreePixmap (clutter_x11_get_default_display (), priv->pixmap);
priv->pixmap = pixmap;
new_pixmap = TRUE;
/* The damage object is created on the window if there is any
* but if there is no window, then we create it directly on
* the pixmap, so it needs to be recreated with a change in
* pixmap.
*/
if (priv->automatic_updates && new_pixmap && !priv->window)
{
free_damage_resources (texture);
create_damage_resources (texture);
}
}
if (priv->pixmap_width != width)
{
priv->pixmap_width = width;
new_pixmap_width = TRUE;
}
if (priv->pixmap_height != height)
{
priv->pixmap_height = height;
new_pixmap_height = TRUE;
}
if (priv->depth != depth)
{
priv->depth = depth;
new_pixmap_depth = TRUE;
}
/* NB: We defer sending the signals until updating all the
* above members so the values are all available to the
* signal handlers. */
g_object_ref (texture);
if (new_pixmap)
g_object_notify (G_OBJECT (texture), "pixmap");
if (new_pixmap_width)
g_object_notify (G_OBJECT (texture), "pixmap-width");
if (new_pixmap_height)
g_object_notify (G_OBJECT (texture), "pixmap-height");
if (new_pixmap_depth)
g_object_notify (G_OBJECT (texture), "pixmap-depth");
free_shm_resources (texture);
if (priv->depth != 0 &&
priv->pixmap != None &&
priv->pixmap_width != 0 &&
priv->pixmap_height != 0)
{
if (CLUTTER_ACTOR_IS_REALIZED (texture))
clutter_x11_texture_pixmap_update_area (texture,
0, 0,
priv->pixmap_width,
priv->pixmap_height);
}
/*
* Keep ref until here in case a notify causes removal from the scene; can't
* lower the notifies because glx's notify handler needs to run before
* update_area
*/
g_object_unref (texture);
}
/**
* clutter_x11_texture_pixmap_set_window:
* @texture: the texture to bind
* @window: the X window to which the texture should be bound
* @automatic: %TRUE for automatic window updates, %FALSE for manual.
*
* Sets up a suitable pixmap for the window, using the composite and damage
* extensions if possible, and then calls
* clutter_x11_texture_pixmap_set_pixmap().
*
* If you want to display a window in a #ClutterTexture, you probably want
* this function, or its older sister, clutter_glx_texture_pixmap_set_window()
*
* Since: 0.8
*/
void
clutter_x11_texture_pixmap_set_window (ClutterX11TexturePixmap *texture,
Window window,
gboolean automatic)
{
ClutterX11TexturePixmapPrivate *priv;
XWindowAttributes attr;
Display *dpy = clutter_x11_get_default_display ();
g_return_if_fail (CLUTTER_X11_IS_TEXTURE_PIXMAP (texture));
priv = texture->priv;
if (!clutter_x11_has_composite_extension())
return;
if (priv->window == window && automatic == priv->window_redirect_automatic)
return;
if (priv->window)
{
clutter_x11_remove_filter (on_x_event_filter_too, (gpointer)texture);
clutter_x11_trap_x_errors ();
XCompositeUnredirectWindow(clutter_x11_get_default_display (),
priv->window,
priv->window_redirect_automatic ?
CompositeRedirectAutomatic : CompositeRedirectManual);
XSync (clutter_x11_get_default_display (), False);
clutter_x11_untrap_x_errors ();
}
priv->window = window;
priv->window_redirect_automatic = automatic;
priv->window_mapped = FALSE;
priv->destroyed = FALSE;
if (window == None)
return;
clutter_x11_trap_x_errors ();
{
if (!XGetWindowAttributes (dpy, window, &attr))
{
XSync (dpy, False);
clutter_x11_untrap_x_errors ();
g_warning ("bad window 0x%x", (guint32)window);
priv->window = None;
return;
}
XCompositeRedirectWindow
(dpy,
window,
automatic ?
CompositeRedirectAutomatic : CompositeRedirectManual);
XSync (dpy, False);
}
clutter_x11_untrap_x_errors ();
if (priv->window)
{
XSelectInput (dpy, priv->window,
attr.your_event_mask | StructureNotifyMask);
clutter_x11_add_filter (on_x_event_filter_too, (gpointer)texture);
}
if (priv->automatic_updates)
{
free_damage_resources (texture);
create_damage_resources (texture);
}
g_object_ref (texture);
g_object_notify (G_OBJECT (texture), "window");
clutter_x11_texture_pixmap_set_mapped (texture,
attr.map_state == IsViewable);
clutter_x11_texture_pixmap_sync_window (texture);
g_object_unref (texture);
}
/**
* clutter_x11_texture_pixmap_sync_window:
* @texture: the texture to bind
*
* Resets the texture's pixmap from its window, perhaps in response to the
* pixmap's invalidation as the window changed size.
*
* Since: 0.8
*/
void
clutter_x11_texture_pixmap_sync_window (ClutterX11TexturePixmap *texture)
{
ClutterX11TexturePixmapPrivate *priv;
Pixmap pixmap;
g_return_if_fail (CLUTTER_X11_IS_TEXTURE_PIXMAP (texture));
priv = texture->priv;
if (priv->destroyed)
return;
if (!clutter_x11_has_composite_extension())
{
clutter_x11_texture_pixmap_set_pixmap (texture, priv->window);
return;
}
if (priv->window)
{
XWindowAttributes attr;
Display *dpy = clutter_x11_get_default_display ();
gboolean mapped = FALSE;
gboolean notify_x = FALSE;
gboolean notify_y = FALSE;
gboolean notify_override_redirect = FALSE;
Status status;
/* NB: It's only valid to name a pixmap if the window is viewable.
*
* We don't explicitly check this though since there would be a race
* between checking and naming unless we use a server grab which is
* undesireable.
*
* Instead we gracefully handle any error with naming the pixmap.
*/
clutter_x11_trap_x_errors ();
pixmap = XCompositeNameWindowPixmap (dpy, priv->window);
status = XGetWindowAttributes (dpy, priv->window, &attr);
if (status != 0)
{
notify_x = attr.x != priv->window_x;
notify_y = attr.y != priv->window_y;
notify_override_redirect = attr.override_redirect != priv->override_redirect;
priv->window_x = attr.x;
priv->window_y = attr.y;
priv->override_redirect = attr.override_redirect;
}
/* We rely on XGetWindowAttributes() implicitly syncing with the server
* so if there was an error naming the pixmap that will have been
* caught by now. */
if (clutter_x11_untrap_x_errors ())
pixmap = None;
g_object_ref (texture); /* guard against unparent */
if (pixmap)
{
clutter_x11_texture_pixmap_set_pixmap (texture, pixmap);
priv->owns_pixmap = TRUE;
}
clutter_x11_texture_pixmap_set_mapped (texture, mapped);
/* could do more clever things with a signal, i guess.. */
if (notify_override_redirect)
g_object_notify (G_OBJECT (texture), "window-override-redirect");
if (notify_x)
g_object_notify (G_OBJECT (texture), "window-x");
if (notify_y)
g_object_notify (G_OBJECT (texture), "window-y");
g_object_unref (texture);
}
}
static void
clutter_x11_texture_pixmap_set_mapped (ClutterX11TexturePixmap *texture,
gboolean mapped)
{
ClutterX11TexturePixmapPrivate *priv;
priv = texture->priv;
if (mapped != priv->window_mapped)
{
priv->window_mapped = mapped;
g_object_notify (G_OBJECT (texture), "window-mapped");
}
}
static void
clutter_x11_texture_pixmap_destroyed (ClutterX11TexturePixmap *texture)
{
ClutterX11TexturePixmapPrivate *priv;
priv = texture->priv;
if (!priv->destroyed)
{
priv->destroyed = TRUE;
g_object_notify (G_OBJECT (texture), "destroyed");
}
/*
* Don't set window to None, that would destroy the pixmap, which might still
* be useful e.g. for destroy animations -- app's responsibility.
*/
}
/**
* clutter_x11_texture_pixmap_update_area:
* @texture: The texture whose content shall be updated.
* @x: the X coordinate of the area to update
* @y: the Y coordinate of the area to update
* @width: the width of the area to update
* @height: the height of the area to update
*
* Performs the actual binding of texture to the current content of
* the pixmap. Can be called to update the texture if the pixmap
* content has changed.
*
* Since: 0.8
**/
void
clutter_x11_texture_pixmap_update_area (ClutterX11TexturePixmap *texture,
gint x,
gint y,
gint width,
gint height)
{
g_return_if_fail (CLUTTER_X11_IS_TEXTURE_PIXMAP (texture));
g_signal_emit (texture, signals[UPDATE_AREA], 0, x, y, width, height);
Adds initial clipped redraw support to Clutter A new (internal only currently) API, _clutter_actor_queue_clipped_redraw can be used to queue a redraw along with a clip rectangle in actor coordinates. This clip rectangle propagates up to the stage and clutter backend which may optionally use the information to optimize stage redraws. The GLX backend in particular may scissor the next redraw to the clip rectangle and use GLX_MESA_copy_sub_buffer to present the stage subregion. The intention is that any actors that can naturally determine the bounds of updates should queue clipped redraws to reduce the cost of updating small regions of the screen. Notes: » If GLX_MESA_copy_sub_buffer isn't available then the GLX backend ignores any clip rectangles. » queuing multiple clipped redraws will result in the bounding box of each clip rectangle being used. » If a clipped redraw has a height > 300 pixels then it's promoted into a full stage redraw, so that the GPU doesn't end up blocking too long waiting for the vsync to reach the optimal position to avoid tearing. » Note: no empirical data was used to come up with this threshold so we may need to tune this. » Currently only ClutterX11TexturePixmap makes use of this new API. This is done via a new "queue-damage-redraw" signal that is emitted when the pixmap is updated. The default handler queues a clipped redraw with the assumption that the pixmap is being painted as a rectangle covering the actors transformed allocation. If you subclass ClutterX11TexturePixmap and change how it's painted you now also need to override the signal handler and queue your own redraw. Technically this is a semantic break, but it's assumed that no one is currently doing this. This still leaves a few unsolved issues with regards to optimizing sub stage redraws that need to be addressed in further work so this can only be considered a stepping stone a this point: » Because we have no reliable way to determine if the painting of any given actor is being modified any optimizations implemented using _clutter_actor_queue_redraw_with_clip must be overridable by a subclass, and technically must be opt-in for existing classes to avoid a change in semantics. E.g. consider that a user connects to the paint signal for ClutterTexture and paints a circle instead of a rectangle. In this case any original logic to queue clipped redraws would be incorrect. » Currently only the implementation of an actor has enough information with which to queue clipped redraws. E.g. It is not possible for generic code in clutter-actor.c to queue a clipped redraw when hiding an actor because actors have no way to report a "paint box". (remember actors can draw outside their allocation and actors with depth may also be projected outside of their allocation) » The current plan is to add a actor_class->get_paint_cuboid() virtual so actors can report a bounding cube for everything they would draw in their current state and use that to queue clipped redraws against the stage by projecting the paint cube into stage coordinates. » Our heuristics for promoting clipped redraws into full redraws to avoid blocking the GPU while we wait for the vsync need improving: » vsync issues aren't relevant for redirected/composited applications so they should use different heuristics. In this case we instead need to trade off the cost of blitting when using glXCopySubBuffer vs promoting to a full redraw and flipping instead.
2009-11-30 17:47:55 +00:00
/* The default handler for the "queue-damage-redraw" signal is
* clutter_x11_texture_pixmap_real_queue_damage_redraw which will queue a
* clipped redraw. */
g_signal_emit (texture, signals[QUEUE_DAMAGE_REDRAW],
0, x, y, width, height);
}
/**
* clutter_x11_texture_pixmap_set_automatic:
* @texture: a #ClutterX11TexturePixmap
* @setting: %TRUE to enable automatic updates
*
* Enables or disables the automatic updates ot @texture in case the backing
* pixmap or window is damaged
*
* Since: 0.8
*/
void
clutter_x11_texture_pixmap_set_automatic (ClutterX11TexturePixmap *texture,
gboolean setting)
{
ClutterX11TexturePixmapPrivate *priv;
Display *dpy;
g_return_if_fail (CLUTTER_X11_IS_TEXTURE_PIXMAP (texture));
priv = texture->priv;
if (setting == priv->automatic_updates)
return;
dpy = clutter_x11_get_default_display();
if (setting)
create_damage_resources (texture);
else
free_damage_resources (texture);
priv->automatic_updates = setting;
}
Adds initial clipped redraw support to Clutter A new (internal only currently) API, _clutter_actor_queue_clipped_redraw can be used to queue a redraw along with a clip rectangle in actor coordinates. This clip rectangle propagates up to the stage and clutter backend which may optionally use the information to optimize stage redraws. The GLX backend in particular may scissor the next redraw to the clip rectangle and use GLX_MESA_copy_sub_buffer to present the stage subregion. The intention is that any actors that can naturally determine the bounds of updates should queue clipped redraws to reduce the cost of updating small regions of the screen. Notes: » If GLX_MESA_copy_sub_buffer isn't available then the GLX backend ignores any clip rectangles. » queuing multiple clipped redraws will result in the bounding box of each clip rectangle being used. » If a clipped redraw has a height > 300 pixels then it's promoted into a full stage redraw, so that the GPU doesn't end up blocking too long waiting for the vsync to reach the optimal position to avoid tearing. » Note: no empirical data was used to come up with this threshold so we may need to tune this. » Currently only ClutterX11TexturePixmap makes use of this new API. This is done via a new "queue-damage-redraw" signal that is emitted when the pixmap is updated. The default handler queues a clipped redraw with the assumption that the pixmap is being painted as a rectangle covering the actors transformed allocation. If you subclass ClutterX11TexturePixmap and change how it's painted you now also need to override the signal handler and queue your own redraw. Technically this is a semantic break, but it's assumed that no one is currently doing this. This still leaves a few unsolved issues with regards to optimizing sub stage redraws that need to be addressed in further work so this can only be considered a stepping stone a this point: » Because we have no reliable way to determine if the painting of any given actor is being modified any optimizations implemented using _clutter_actor_queue_redraw_with_clip must be overridable by a subclass, and technically must be opt-in for existing classes to avoid a change in semantics. E.g. consider that a user connects to the paint signal for ClutterTexture and paints a circle instead of a rectangle. In this case any original logic to queue clipped redraws would be incorrect. » Currently only the implementation of an actor has enough information with which to queue clipped redraws. E.g. It is not possible for generic code in clutter-actor.c to queue a clipped redraw when hiding an actor because actors have no way to report a "paint box". (remember actors can draw outside their allocation and actors with depth may also be projected outside of their allocation) » The current plan is to add a actor_class->get_paint_cuboid() virtual so actors can report a bounding cube for everything they would draw in their current state and use that to queue clipped redraws against the stage by projecting the paint cube into stage coordinates. » Our heuristics for promoting clipped redraws into full redraws to avoid blocking the GPU while we wait for the vsync need improving: » vsync issues aren't relevant for redirected/composited applications so they should use different heuristics. In this case we instead need to trade off the cost of blitting when using glXCopySubBuffer vs promoting to a full redraw and flipping instead.
2009-11-30 17:47:55 +00:00