The end goal is to have all clutter backend code in src/backends. Input
is the larger chunk of it, which is now part of our specific
MutterClutterBackendX11, this extends to device manager, input devices,
tools and keymap.
This was supposed to be nice and incremental, but there's no sane way
to cut this through. As a result of the refactor, a number of private
Clutter functions are now exported for external backends to be possible.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/672
Mutter needs to know which framebuffer the paint nodes will be
drawn into, and using cogl_get_draw_framebuffer() directly is
not an option since ClutterRootNode only pushes the draw fb
at draw time.
Expose clutter_paint_node_get_framebuffer().
https://gitlab.gnome.org/GNOME/mutter/merge_requests/409
Incompressible events already pass through unmodified, so queuing them
just wasted time and memory.
We would however like to keep the ordering of events so we can only
apply this optimization if the queue is empty.
This reduces the input latency of incompressible events like touchpad
scrolling or drawing tablets by up to one frame. It also means the same
series of events now arrives at the client more smoothly and not in
bursts.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/711
Until now we would:
1. Enqueue modifier key event on the stage.
2. Update device modifier state.
3. Dequeue and process modifier key event with NEW device modifier state.
But if we consider optimizing out the queuing in some cases then there
will become a problem:
1. Process modifier key event with OLD device modifier state.
2. Update device modifier state.
To correct the above we now do:
1. Update device modifier state.
2. Queue/process modifier key event with NEW device modifier state.
It appears commit dd940a71 which introduced the old behaviour was correct
in the need to update the device modifier state, but is at least no longer
correct (if it ever was) that it should be done after queuing the event.
If queuing is working, as it is right now, then it makes no difference
whether the device modifier state is updated before or after. Because both
cases will come before the dequeing and processing.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/711
Thanks to the now removed global/context grabs, we can move pointer and
keyboard grabs back home to where they belong.
While at it, also add handling of CLUTTER_TABLET_DEVICE devices to
`on_grab_actor_destroy` and `clutter_input_device_get_grabbed_actor`.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/536
If possible, GLib will try to use the va_marshaller to pass the signal
arguments, rather than unboxing into and out of a `GValue`. This is much
more performant and especially good for often-thrown signals.
The original bug even mentions Clutter performance issues as a drive to
implement the va_marshaller in GLib (see
https://bugzilla.gnome.org/show_bug.cgi?id=661140).
https://gitlab.gnome.org/GNOME/mutter/merge_requests/700
Some of the marshallers we generate in `clutter-marshal.list` are also
available in GLib, so we don't need to generate them ourselves. Even
more, by passing NULL to `g_signal_new` in these cases will actually
internally optimize this even more by also setting the valist
marshaller, which is a little bit faster than the regular marshalling
using `GValue` and libffi.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/700
`g_object_notify()` actually takes a global lock to look up the property
by its name, which means there is a performance hit (albeit tiny) every
time this function is called. For this reason, always try to use
`g_object_notify_by_pspec()` instead.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/703
Since commit 0eab73dc, actors are only allocated when they are actually
visible. While this generally works well, it breaks - because of *course*
it does - ClutterClones when the clone source (or any of its ancestors)
is hidden.
Force an allocation in that case to allow the clone's paint to work as
intended.
https://gitlab.gnome.org/GNOME/mutter/issues/683
The type has been deprecated, so stop using it. The easiest replacement
would be to add our own struct, but considering that separate name/value
arrays are easier to free (g_auto!) and we already do the split in one
place, go that route.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/689
Glib stopped providing any fallback implementations on systems without
memmove() all the way back in 2013. Since then, the symbol is a simple
macro around memmove(); use that function directly now that glib added
a deprecation warning.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/689
Currently nothing in the clutter machinery prevents hidden portions
of the actor tree from calling queue_relayout() (and having it fully
honored).
But that allocation should not be necessary till the actor is shown,
and one of the things we do on show() is queueing a relayout/redraw
after flagging the actor as visible.
We can simply defer clutter_actor_allocate() calls till that show()
call, and leave the needs_allocate and other flags set so we ensure
the allocation is properly set then.
This should cut down some needless operations when invisible portions
of the actor tree change indirectly due to user interaction, or due
to background activity.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/677
Clutter has two motion event toggles: one in the global context, and
one per stage. The global one is deprecated, and currently unused.
Remove the global motion event handling.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/666
The “togglekeys” setting is to emit a sounds whenever the state of one
of the modifiers keys (CAPS lock, NUM Lock, SCROLL lock) is changed, it
has nothing to do with the rest of the accessibility settings.
Therefore, there is no need to reset the various timers used by
accessibility whenever the “togglekeys” setting is changed.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/614
ClutterStage:after-paint now does not guarantee a valid
implicit framebuffer pushed to the stack. Instead, use
the new 'paint-view' signal, that is emitted at a point
in the drawing routine where a framebuffer is pushed.
In addition to that, stop using the implicit framebuffer
API and port the actor-shader-effect test to read from
the view's framebuffer directly.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/623
Now that ClutterStageView is embraced as part of the public
set of Clutter classes, is it possible to give consumers
of this API more information and control over the drawing
routines of ClutterStage.
Introduce ClutterStage:paint-view, a signal that is emitted
for painting a specific view. It's defined as a RUN_LAST
signal to give anyone connecting to it the ability to run
before the view is actually painted, or after (using the
G_CONNECT_AFTER flag, or g_signal_connect_after).
This signal has a corresponding class handler, which allows
Mutter to have much finer control over the painting routines.
In fact, this will allow us to implement a "paint phase watcher"
mechanism in the following patches.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/623
ClutterStage:after-paint is supposed to be emitted after all
painting is done, but before the frame is finished. However,
as it is right now, it is being emitted after each view is
painted -- on multi-monitor setups, after-frame is being
emitted multiple times.
Send after-paint only once, after all views are painted and
before finishing the frame.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/623
As a compositor toolkit, it makes sense to allow consumers
of Clutter interact with the stage views themselves. As such,
ClutterStageView should be a public class.
As such, it is now included in clutter.h and should not be
included directly.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/623
When libwacom is configured disabled, this error appears:
../clutter/clutter/x11/clutter-input-device-xi2.c: In function ‘clutter_input_device_xi2_finalize’:
../clutter/clutter/x11/clutter-input-device-xi2.c:122:7: error: ‘device_xi2’ undeclared (first use in this function)
if (device_xi2->inhibit_pointer_query_timer)
Fix it with the "obvious" solution.
This code was added in c1303bd642.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/611
When a user moves their cursor the perceived behaviour is that it will
pick what is under the cursor. However this isn't how picking works.
Picking does a virtual redraw of the screen, so in some cases what gets
picked isn't the same as what the user could see on the previous frame.
It more represents what will be drawn on the next frame than what is on
screen at present.
It may be unsafe to change these semantics, and they are useful anyway.
Just document it better.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/214
After 4faeb12731, the maximum time allowed for an update to happen
is calculated as:
max_render_time_allowed = refresh_interval - 1000 * sync_delay;
However, extremely small refresh intervals -- that come as consequence
to extremely high refresh rates -- may fall into an odd numerical range
when refresh_interval < 1000 * sync_delay. That would give us a negative
time.
To be extra cautious about it, add another sanity check for this case.
Change suggested by Jasper St. Pierre.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/363
The presentation timing logic (via `master_clock_get_swap_wait_time`) now
works unconditionally. By "works" we mean that a result of zero from
`master_clock_get_swap_wait_time` actually means zero now. Previously
zero could mean either a successful result of zero milliseconds or that
the backend couldn't get an answer. And a non-zero result is the same as
before.
This works even if the screen is "idle" and even if the backend doesn't
provide presentation timestamps. So now our two fallback throttling
mechanisms of relying on `CLUTTER_FEATURE_SWAP_THROTTLE` and decimating
to `clutter_get_default_frame_rate` can be deleted.
Closes: https://gitlab.gnome.org/GNOME/mutter/issues/406 and
https://bugzilla.gnome.org/show_bug.cgi?id=781835https://gitlab.gnome.org/GNOME/mutter/merge_requests/363
If `last_presentation_time` is zero (unsupported) or just very old
(system was idle) then we would like to avoid that triggering a large
number of loop interations. This will get us closer to the right answer
without iterating.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/363
That could happen if the backend did not provide presentation timestamps,
or if the screen was not changing other than the hardware cursor:
if (stage_cogl->last_presentation_time == 0||
stage_cogl->last_presentation_time < now - 150000)
{
stage_cogl->update_time = now;
return;
}
By setting `update_time` to `now`, master_clock_get_swap_wait_time()
returns 0:
gint64 now = g_source_get_time (master_clock->source);
if (min_update_time < now)
{
return 0;
}
else
{
gint64 delay_us = min_update_time - now;
return (delay_us + 999) / 1000;
}
However, zero is a value unsupported by the default master clock
due to:
if (swap_delay != 0)
return swap_delay;
All cases are now handled by extrapolating when the next presentation
time would be and calculating an appropriate update time to meet that.
We also need to add a check for `update_time == last_update_time`, which
is a situation that just became possible since we support old (or zero)
values of `last_presentation_time`. This avoids getting more than one
stage update per frame interval when input events arrive without
triggering a stage redraw (e.g. moving the hardware cursor).
https://gitlab.gnome.org/GNOME/mutter/merge_requests/363
Instead of crazy refresh rates >1MHz falling back to 60Hz, just honour
them by rendering unthrottled (same as `sync_delay < 0`). Although I
wouldn't actually expect that path to ever be needed in reality, it just
ensures an infinite `while` loop never happens.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/363
When using evdev (for Wayland), the backend receives all device events
and queue them for clutter.
Hook up the pointer accessibility handlers in clutter's main processing
queue, so that we get better accuracy for pointer location.
We need to avoid doing this on X11 though because X11 relies on the raw
events for this to work reliably, so the same is already done in the
X11 backend when using X11.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/512
Pointer accessibility features requires to receive all pointer events
regardless of X11 grabs.
Add XI2 raw events mask and hook up the pointer accessibility handlers
to the raw motion and button press/release events.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/512
Add support for click assist, namely simulated secondary click (on a
long primary button press) and hover click support (simulate a click when
the pointer remains static for some time).
https://gitlab.gnome.org/GNOME/mutter/merge_requests/512
Add the required signaling in place in clutter device manager to notify
the upper layers (namely, the shell) whenever a click assist delay or
gesture is started or stopped.
This will allow the shell to implement a visual feedback for click
assist operations.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/512
For accessibility features, being either keyboard accessibility to
implement mousekeys, or pointer accessibility to implement simulated
secondary click or dwell click, we need to have a virtual device.
Add that virtual device in ClutterInputDevice so it can be used either
for keyboard or pointer accessibility.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/512
Certain arguments like `-fno-omit-frame-pointer` break GIR creation.
Lets handle this like we do for the rest of mutter and duplicate the
relevant arguments from `clutter_c_args`.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/601
Spotted while adding tracing to swap buffers, we only enter
the first part of the if condition when use_clipped_redraw
is TRUE, so it's pretty safe to assume it's TRUE.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/197
The idea here is to be able to visualize and immediately
understand what is happening. Something like:
```
[ view1 ] [ view2 ]
[---- Layout ---][------ Paint ------][ Pick ]
[================== Update =====================]
```
But with colors. A few of the previous profiling data
sections were removed, since they didn't really add to
reading the graph.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/197
Add debug flags based on meson's `debug` option instead of `buildtype`.
This allows custom build configurations to behave like a debug or release build.
Add `-fno-omit-frame-pointer` to Mutter/Cogl. Not to Clutter though, as that would
require more changes to how Clutter's gir is created
Remove `-DG_DISABLE_CAST_CHECKS` from Clutter in debug builds
Add `-DG_DISABLE_CHECKS`, `-DG_DISABLE_ASSERT` and `-DG_DISABLE_CAST_CHECKS` to all
non-debug builds but `plain`, which explicitly should not have any compile flags
Use `cc.get_supported_arguments`, so it becomes more obvious to the user which flags
are set during compilation
https://gitlab.gnome.org/GNOME/mutter/merge_requests/497
Commit df7d8e2cb highlights a crash on test_destroy_destroy, in fact it could
happen that calling clutter_actor_destroy on a child while iterating on the
list, would implicitly call test_destroy_remove that tries to modify the list
at the same time. Causing a memory error.
So instead of manually free the children list, just ensure that this list is
valid and that when the object destruction is done, this is free'd.
See: https://gitlab.gnome.org/GNOME/mutter/merge_requests/576https://gitlab.gnome.org/GNOME/mutter/merge_requests/581
Commit cabcad185 removed the call to cogl_set_source_color4ub() before
cogl_fill_path(), so instead of the previously assigned selection color,
the background is drawn with the last set source.
In order to honour the newly added framebuffer parameter and still apply
the correct color, switch from cogl_fill_path() to the (deprecated!)
cogl_framebuffer_fill_path() method.
https://gitlab.gnome.org/GNOME/mutter/issues/494
If an update (new frame) had been scheduled already before
`_clutter_stage_cogl_presented` was called then that means it was
scheduled for the wrong time. Because the `last_presentation_time` has
changed since then. And using an `update_time` based on an outdated
presentation time results in scheduling frames too early, filling the
buffer queue (triple buffering or worse) and high visual latency.
So if we do receive a presentation event when an update is already
scheduled, remember to reschedule the update based on the newer
`last_presentation_time`. This way we avoid overfilling the buffer queue
and limit ourselves to double buffering for less visible lag.
Closes: https://gitlab.gnome.org/GNOME/mutter/issues/334
Prerequisite: https://gitlab.gnome.org/GNOME/mutter/merge_requests/520https://gitlab.gnome.org/GNOME/mutter/merge_requests/281
Pango functions pango_unichar_direction() and pango_find_base_dir() have been
deprecated in pango 1.44, since these are used mostly clutter and gtk, copy the
code from pango and use fribidi dependency explicitly.
This is the same strategy used by Gtk.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/583
The `last_presentation_time` is usually a little in the past (although
sometimes in the future depending on the driver). When it's over 2ms
(`sync_delay`) in the past that would trigger the while loop to count up so
that the next `update_time` is in the future.
The problem with that is for common values of `last_presentation_time`
which are only a few milliseconds ago, incrementing `update_time` by
`refresh_interval` also means counting past the next physical frame that
we haven't rendered yet. And so mutter would skip that frame.
**Example**
Given:
```
last_presentation_time = now - 3ms
sync_delay = 2ms
refresh_interval = 16ms
next_presentation_time = last_presentation_time + refresh_interval
= now + 13ms
-3ms now +13ms +29ms +45ms
----|--+------------|---------------|---------------|----
: :
last_presentation_time next_presentation_time
```
Old algorithm:
```
update_time = last_presentation_time + sync_delay
= now - 1ms
while (update_time < now)
(now - 1ms < now)
update_time = now - 1ms + 16ms
update_time = now + 15ms
next_presentation_time = now + 13ms
available_render_time = next_presentation_time - max(now, update_time)
= (now + 13ms) - (now + 15ms)
= -2ms so the next frame will be skipped.
-3ms now +13ms +29ms +45ms
----|--+------------|-+-------------|---------------|----
: : :
: : update_time (too late)
: :
last_presentation_time next_presentation_time (a missed frame)
```
New algorithm:
```
min_render_time_allowed = refresh_interval / 2
= 8ms
max_render_time_allowed = refresh_interval - sync_delay
= 14ms
target_presentation_time = last_presentation_time + refresh_interval
= now - 3ms + 16ms
= now + 13ms
while (target_presentation_time - min_render_time_allowed < now)
(now + 13ms - 8ms < now)
(5ms < 0ms)
# loop is never entered
update_time = target_presentation_time - max_render_time_allowed
= now + 13ms - 14ms
= now - 1ms
next_presentation_time = now + 13ms
available_render_time = next_presentation_time - max(now, update_time)
= (now + 13ms) - now
= 13ms which is plenty of render time.
-3ms now +13ms +29ms +45ms
----|-++------------|---------------|---------------|----
: : :
: update_time :
: :
last_presentation_time next_presentation_time
```
The reason nobody noticed these missed frames very often was because
mutter has some accidental workarounds built-in:
* Prior to 3.32, the offending code was only reachable in Xorg sessions.
It was never reached in Wayland sessions because it hadn't been
implemented yet (till e9e4b2b72).
* Even though Wayland support is now implemented the native backend
provides a `last_presentation_time` much faster than Xorg sessions
(being in the same process) and so is less likely to spuriously enter
the while loop to miss a frame.
* For Xorg sessions we are accidentally triple buffering (#334). This
is a good way to avoid the missed frames, but is also an accident.
* `sync_delay` is presently just high enough (2ms by coincidence is very
close to common values of `now - last_presentation_time`) to push the
`update_time` into the future in some cases, which avoids entering the
while loop. This is why the same missed frames problem was also noticed
when experimenting with `sync_delay = 0`.
v2: adjust variable names and code style.
Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789186
and most of https://gitlab.gnome.org/GNOME/mutter/issues/571https://gitlab.gnome.org/GNOME/mutter/merge_requests/520
We only call _clutter_input_device_update with devices that are not
Keyboard devices. Also passing a Keyboard device to a function whose
primary purpose is picking should be considered a bug in the caller.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/547
We're bailing out of clutter_stage_cogl_add_redraw_clip() early without
doing anything if we're ignoring redraw clips, so no need to call it if
we already know that will be the case.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/547
GList's used in legacy code were free'd using a g_slist_foreach + g_slist_free,
while we can just use g_slist_free_full as per GLib 2.28.
So replace code where we were using this legacy codepath.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/576
GList's used in legacy code were free'd using a g_list_foreach + g_list_free,
while we can just use g_list_free_full as per GLib 2.28.
So replace code where we were using this legacy codepath.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/576
Moving an actor with a ClutterDeformEffect applied flickers because
the depth_testing, setting the depth testing test function to
COGL_DEPTH_TEST_FUNCTION_LEQUAL fixes the problem.
Fixes https://gitlab.gnome.org/GNOME/mutter/issues/507
When the pointer is grabbed, we send the crossing events that are
initiated by this pointer only to the actor that has the grab. For
grabbed touch sequences, we always capture and bubble the crossing
events right now.
Fix this and make grabbed pointers and touch sequences behave the same
by sending touch crossing events only to the grab actor.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/422
It's important to cancel click actions when we get a touch cancel event,
otherwise the long press event might get emitted after the compositor
took over the touches because it detected a gesture.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/396
Clutter does the nicety of connecting just created PangoContexts to
ClutterBackend signals in order to update it on resolution/font changes.
However the way the signals are disconnected (automatically via
g_signal_connect_object() auto-disconnect feature) may incur into
performance issues with a high enough number of ClutterActors with a
PangoContext (eg. ClutterText) as the lookup by closure is linear across
all signals and handlers.
Keep the handler IDs around, and disconnect them specifically on dispose
so it is more O(1)-ish.
Related: https://gitlab.gnome.org/GNOME/mutter/issues/556
