UProf is a small library that aims to help applications/libraries provide
domain specific reports about performance. It currently provides high
precision timer primitives (rdtsc on x86) and simple counters, the ability
to link statistics between optional components at runtime and makes report
generation easy.
This adds initial accounting for:
- Total mainloop time
- Painting
- Picking
- Layouting
- Idle time
The timing done by uprof is of wall clock time. It's not based on stochastic
samples we simply sample a counter at the start and end. When dealing with
the complexities of GPU drivers and with various kinds of IO this form of
profiling can be quite enlightening as it will be able to represent where
your application is blocking unlike tools such as sysprof.
To enable uprof accounting you must configure Clutter with --enable-profile
and have uprof-0.2 installed from git://git.moblin.org/uprof
If you want to see a report of statistics when Clutter applications exit you
should export CLUTTER_PROFILE_OUTPUT_REPORT=1 before running them.
Just a final word of caution; this stuff is new and the manual nature of
adding uprof instrumentation means it is prone to some errors when modifying
code. This just means that when you question strange results don't rule out
a mistake in the instrumentation. Obviously though we hope the benfits out
weigh e.g. by focusing on very key stats and by having automatic reporting.
The ClutterStage:use-alpha property is used to let a stage know that it
should honour the alpha component of the ClutterStage:color property.
If :use-alpha is set to FALSE the stage always uses the full opacity
when clearing itself before a paint(); otherwise, the alpha value is
used.
The stage's pick id can be written to the framebuffer when we call
cogl_clear so there's no need for the stage to also chain up in it's pick
function resulting in clutter-actor.c also emitting a rectangle for the
stage.
There is a new internal Cogl function called _cogl_check_driver_valid
which looks at the value of the GL_VERSION string to determine whether
the driver is supported. Clutter now calls this after the stage is
realized. If it fails then the stage is marked as unrealized and a
warning is shown.
_cogl_features_init now also checks the version number before getting
the function pointers for glBlendFuncSeparate and
glBlendEquationSeparate. It is not safe to just check for the presence
of the functions because some drivers may define the function without
fully implementing the spec.
The GLES version of _cogl_check_driver_valid just always returns TRUE
because there are no version requirements yet.
Eventually the function could also check for mandatory extensions if
there were any.
http://bugzilla.openedhand.com/show_bug.cgi?id=1875
We can not process events for a stage that has been destroyed so we
should make sure that the events for the stage are removed from the
global event queue during dispose.
http://bugzilla.openedhand.com/show_bug.cgi?id=1882
Because Cogl defines the origin of viewport and window coordinates to be
top-left it always needs to know the size of the current window so that Cogl
window/viewport coordinates can be transformed into OpenGL coordinates.
This also fixes cogl_read_pixels to use the current draw buffer height
instead of the viewport height to determine the OpenGL y coordinate to use
for glReadPixels.
The only backend that tried to implement offscreen stages was the GLX backend
and even this has apparently be broken for some time without anyone noticing.
The property still remains and since the property already clearly states that
it may not work I don't expect anyone to notice.
This simplifies quite a bit of the GLX code which is very desireable from the
POV that we want to start migrating window system code down to Cogl and the
simpler the code is the more straight forward this work will be.
In the future when Cogl has a nicely designed API for framebuffer objects then
re-implementing offscreen stages cleanly for *all* backends should be quite
straightforward.
Instead of using ClutterActor for the base class of the Stage
implementation we should extend the StageWindow interface with
the required bits (geometry, realization) and use a simple object
class.
This require a wee bit of changes across Backend, Stage and
StageWindow, even though it's mostly re-shuffling.
First of all, StageWindow should get new virtual functions:
* geometry:
- resize()
- get_geometry()
* realization
- realize()
- unrealize()
This covers all the bits that we use from ClutterActor currently
inside the stage implementations.
The ClutterBackend::create_stage() virtual function should create
a StageWindow, and not an Actor (it should always have been; the
fact that it returned an Actor was a leak of the black magic going
on underneath). Since we never guaranteed ABI compatibility for
the Backend class, this is not a problem.
Internally to ClutterStage we can finally drop the shenanigans of
setting/unsetting actor flags on the implementation: if the realization
succeeds, for instance, we set the REALIZED flag on the Stage and
we're done.
As an initial proof of concept, the X11 and GLX stage implementations
have been ported to the New World Order(tm) and show no regressions.
The Stage:offscreen property hasn't been tested for ages, and it
should really just use a FBO, not indirect rendering on a X Pixmap
only on X11. There are better ways anyway to get the current
contents of ClutterStage as a buffer anyway.
We might remove it at any later date, or actually make it work
properly.
It might be desirable for some applications and/or platforms to get
every motion event that was delivered to Clutter from the windowing
backend. By adding a per-stage flag we can bypass the throttling
done when processing the events.
http://bugzilla.openedhand.com/show_bug.cgi?id=1665
A lot of applications change the size of the stage from the default
before the stage is initially shown. The size change won't take affect
until the first allocation run. However we want the window to be at
the correct size when we first map it so we should force an allocation
run before showing the stage.
There was an explicit call to XResizeWindow in
clutter_stage_x11_show. This is not needed anymore because
XResizeWindow will already have been called by the allocate method.
To allow for flushing of batched geometry within Cogl we can't support users
directly calling glReadPixels. glReadPixels is also awkward, not least
because it returns upside down image data.
All the unit tests have been swithed over and clutter_stage_read_pixels now
sits on top of this too.
The clutter_context_get_default() function is private, but shared
across Clutter. For this reason, it should be prefixed by '_' so
that the symbol is hidden from the shared object.
Merge branch 'premultiplication'
[cogl-texture docs] Improves the documentation of the internal_format args
[test-premult] Adds a unit test for texture upload premultiplication semantics
[fog] Document that fogging only works with opaque or unmultipled colors
[test-blend-strings] Explicitly request RGBA_888 tex format for test textures
[premultiplication] Be more conservative with what data gets premultiplied
[bitmap] Fixes _cogl_bitmap_fallback_unpremult
[cogl-bitmap] Fix minor copy and paste error in _cogl_bitmap_fallback_premult
Avoid unnecesary unpremultiplication when saving to local data
Don't unpremultiply Cairo data
Default to a blend function that expects premultiplied colors
Implement premultiplication for CoglBitmap
Use correct texture format for pixmap textures and FBO's
Add cogl_color_premultiply()
The fixed function fogging provided by OpenGL only works with unmultiplied
colors (or if the color has an alpha of 1.0) so since we now premultiply
textures and colors by default a note to this affect has been added to
clutter_stage_set_fog and cogl_set_fog.
test-depth.c no longer uses clutter_stage_set_fog for this reason.
In the future when we can depend on fragment shaders we should also be
able to support fogging of premultiplied primitives.
Merge branch 'master-clock-updates'
* master-clock-updates: (22 commits)
Change the paint forcing on the Text cache text
[timelines] Improve marker hit check and don't fudge the delta
Revert "[timeline] Don't clamp the elapsed time when a looping tl reaches the end"
[tests] Don't add a newline to the end of g_test_message calls
[test-timeline] Add a marker at the beginning of the timeline
[timeline] Don't clamp the elapsed time when a looping tl reaches the end
[master-clock] Throttle if no redraw was performed
[docs] Update Clutter's API reference
Force a paint instead of calling clutter_redraw()
Fix clutter_redraw() to match the redraw cycle
Run the repaint functions inside the redraw cycle
Remove useless manual timeline ticking
Move elapsed-time calculations into ClutterTimeline
Limit the frame rate when not syncing to VBLANK
Decrease the main-loop priority of the frame cycle
Avoid motion-compression in test-picking test
Compress events as part of the frame cycle
Remove stage update idle and do updates from the master clock
Call g_main_context_wakeup() when we start running timelines
Remove unused msecs_delta member
...
Someone might hide the previously focused actor in the focus-out signal
handler and as key focus still appears to be on that actor we'd get
re-entrant call and get glib critical from g_object_weak_unref
This changes clutter_stage_get_key_focus() to return stage/NULL during
focus-out signal emission. It used to be the actor focus-out was being
emitted on.
Fixes bug:
http://bugzilla.openedhand.com/show_bug.cgi?id=1547
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
A flag in the master clock is now set whenever the dispatch caused an
actual redraw of a stage. If this flag is not set during the prepare
and check functions then it will resort to limiting the redraw
attempts to the default frame rate as if vblank syncing was
disabled. Otherwise if a timeline is running that does not cause the
scene to change then it would busy-wait with 100% CPU until the next
frame.
This fix was suggested by Owen Taylor in:
http://bugzilla.openedhand.com/show_bug.cgi?id=1637
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
We do not need the whole redraw machinery inside
clutter_stage_read_pixels(): instead, we want Clutter to drop everything,
paint and call glReadPixels() with the current buffer.
The clutter_redraw() function is used by embedding toolkits to
force a redraw on a stage. Since everything is performed by
toggling a flag inside the Stage itself and then letting the
master clock advance, we need a ClutterStage method to ensure
that we start the master clock and redraw.
clutter-master-clock.c clutter-master-clock.h: When the
SYNC_TO_VBLANK feature is not available, wait for 1/frame_rate
seconds since the start of the last frame before drawing the next
frame. Add _clutter_master_clock_start_running() to abstract
the usage of g_main_context_wakeup()
clutter-stage.c: Add _clutter_master_clock_start_running()
clutter-main.c: Update docs for clutter_set_default_frame_rate()
clutter_get_default_frame_rate() to no longer talk about timeline
frame rates.
test-text-perf.c test-text.c: Set a frame rate of 1000fps so that
frame-rate limiting doesn't affect the result.
http://bugzilla.openedhand.com/show_bug.cgi?id=1637
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
Instead of trying to guess about which motion events are
extraneous, queue up all events until we process a frame.
This allows us to look ahead and reliably compress consecutive
sequence of motion events.
clutter-main.c: Feed received events to the stage for queueing.
Remove old compression code. Remove clutter_get_motion_events_frequency()
clutter_set_motion_events_frequency()
clutter-stage.c: Keep a queue of pending events.
clutter-master-clock.c: Add processng of queued events to the
clock source dispatch function.
http://bugzilla.openedhand.com/show_bug.cgi?id=1637
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
When a redraw is queued on a stage, simply set a flag; then in
the check/prepare functions of the master clock source, check
for stages that need redrawing.
This avoids the complexity of having multiple competing sources
at the same priority and makes the update ordering more reliable and
understandable.
http://bugzilla.openedhand.com/show_bug.cgi?id=1637
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
The clutter_stage_fullscreen() and clutter_stage_unfullscreen() are
a GDK-ism. The underlying implementation is already using an accessor
with a boolean parameter.
This should take the amount of collisions between properties, methods
and signals to zero.
The :fullscreen property is very much confusing as it is implemented.
It can be written to a value, but the whole process might fail. If we
set:
g_object_set (stage, "fullscreen", TRUE, NULL);
and the fullscreen process fails or it is not implemented, the value
will be reset to FALSE (if we're lucky) or left TRUE (most of the
times).
The writability is just a shorthand for invoking clutter_stage_fullscreen()
or clutter_stage_unfullscreen() depending on a boolean value without
using an if.
The :fullscreen property also greatly confuses high level languages,
since the same symbol is used:
- for a method name (Clutter.Stage.fullscreen())
- for a property name (Clutter.Stage.fullscreen)
- for a signal (Clutter.Stage::fullscreen)
For these reasons, the :fullscreen should be renamed to :fullscreen-set
and be read-only. Implementations of the Stage should only emit the
StageState event to change from normal to fullscreen, and the Stage
will automatically update the value of the property and emit a notify
signal for it.
An implementaton of realize() never needs to set the
CLUTTER_ACTOR_REALIZED flag, though it can unset the flag if
things fail unexpectedly. (Previously, stage backend implementations
had to do this since clutter_actor_realize() wasn't used; this
is no longer the case.)
http://bugzilla.openedhand.com/show_bug.cgi?id=1634
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
The clutter_redraw() function is used by libraries embedding
Clutter inside another toolkit, instead of queueing a redraw
on the embedded stage. This means that clutter_redraw() should
perform the same sequence of actions done by the redraw idle
callback.
The mapping and unmapping of the X11 stage implementation is
a bit bong. It's asynchronous, for starters, when it really
can avoid it by tracking the state internally.
The ordering of the map/unmap sequence is also broken with
respect to the resizing.
By tracking the state internally into StageX11 we can safely
remove the MapNotify and UnmapNotify X event handling.
In theory, we should use _NET_WM_STATE a lot more, and reuse
the X11 state flags for fullscreening as well.
Instead of passing a boolean value, the ::allocate virtual function
should use a bitmask and flags. This gives us room for expansion
without breaking API/ABI, and allows to encode more information to
the allocation process instead of just changes of absolute origin.
Sometimes it is necessary for third party code to have a
function called during the redraw process, so that you can
update the scenegraph before it is painted.
The stencil buffer is always cleared the first time a clip is used
that needs it and the stencil test is disabled otherwise so there is
no need to clear before a paint.
When the stage state changes between active/deactive, send out
key-focus-in/key-focus-out signal for the current key focused
actor on the stage.
Fixes bug:
http://bugzilla.openedhand.com/show_bug.cgi?id=1503
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
When showing a Stage for the first time we end up realizing the stage
implementation before realizing the wrapper. This leads to segmentation
faults or errors coming from the backend because we're fumbling the
state and realization sequence.
The fix for bug 1138 broke multi-stage support on GLX, causing
X11 to segfault with the following stack trace:
Backtrace:
0: /usr/X11R6/bin/X(xf86SigHandler+0x7e) [0x80c91fe]
1: [0xb7eea400]
2: /usr/lib/xorg/modules/extensions//libglx.so [0xb7ae880c]
3: /usr/lib/xorg/modules/extensions//libglx.so [0xb7aec0d6]
4: /usr/X11R6/bin/X [0x8154c24]
5: /usr/X11R6/bin/X(Dispatch+0x314) [0x808de54]
6: /usr/X11R6/bin/X(main+0x4b5) [0x8074795]
7: /lib/i686/cmov/libc.so.6(__libc_start_main+0xe5) [0xb7c75775]
8: /usr/X11R6/bin/X(FontFileCompleteXLFD+0x21d) [0x8073a81]
which I can only track down to clutter_backend_glx_ensure_current()
being passed a NULL stage -- something that happens when a stage
is not correct realized. That should lead to a glXMakeCurrent(None)
and not to a segmentation fault, though.
The stage is chaining up to the ClutterGroup::paint instead of
the ClutterGroup::pick method. This works anyway because we
detect the stage by default, but it's not a reliable solution
in case we decide to change the picking further on.
The master clock is currently advanced using a frame source driven
by the default frame rate. This breaks the sync to vblank because
the vblanking rate could be different than 60 Hz -- or it might be
completely disabled (e.g. with CLUTTER_VBLANK=none).
We should be using the main loop to check if we have timelines
playing, and if so queue a redraw on the stages we own.
We should also prepare the subsequent frame at the end of the redraw
process, so if there are new redraw we will have the scene already
in place.
This makes Clutter redraw at the maximum frame rate, which is
limited by the vblanking frequency.
Commit 515350a7 renamed ::focus-in and ::focus-out to ::key-focus-in
and ::key-focus-out respectively. One signal emission for ::focus-out
escaped the renaming in ClutterStage.
For consistency, and since those signals are key-related, the
::focus-in signal is not ::key-focus-in and the ::focus-out
signal is now ::key-focus-out.
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
Redundant clearing of depth and stencil buffers every render can be very
expensive, so cogl now gives control over which auxiliary buffers are
cleared.
Note: For now clutter continues to clear the color, depth and stencil buffer
each paint.
Currently, all timelines install a timeout inside the TimeoutPool
they share. Every time the main loop spins, all the timeouts are
updated. This, in turn, will usually lead to redraws being queued
on the stages.
This behaviour leads to the potential starvation of timelines and
to excessive redraws.
One lesson learned from the games developers is that the scenegraph
should be prepared in its entirety before the GL paint sequence is
initiated. This means making sure that every ::new-frame signal
handler is called before clutter_redraw() is invoked.
In order to do so a TimeoutPool is not enough: we need a master
clock. The clock will be responsible for advancing all the active
timelines created inside a scene, but only when the stage is
being redrawn.
The sequence is:
+ queue_redraw() is invoked on an actor and bubbles up
to the stage
+ if no redraw() has already been scheduled, install an
idle handler with a known priority
+ inside the idle handler:
- advance the master clock, which will in turn advance
every playing timeline by the amount of milliseconds
elapsed since the last redraw; this will make every
playing timeline emit the ::new-frame signal
- queue a relayout
- call the redraw() method of the backend
This way we trade multiple timeouts with a single frame source
that only runs if a timeline is playing and queues redraws on
the various stages.
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
Bug 1513 - Allow passing in ClutterPickMode to
clutter_stage_get_actor_at_pos()
At the moment, clutter_stage_get_actor_at_pos() uses CLUTTER_PICK_ALL
internally to find an actor. It would be useful to allow passing in
ClutterPickMode to clutter_stage_get_actor_at_pos(), so that the caller
can specify CLUTTER_PICK_REACTIVE as a criteria.
