Commit Graph

319 Commits

Author SHA1 Message Date
Giovanni Campagna
a26ded47d9 Add a private gtk-mutter protocol
Add a new interface, gtk_shell, than can be used by gtk to
retrieve a surface extension called gtk_surface, which will be
used to communicate with mutter all the GTK extensions to EWMH

https://bugzilla.gnome.org/show_bug.cgi?id=707128

Add support for GTK application menus

To do so, we need to be able to set surface state before creating
the MetaWindow, so we introduce MetaWaylandSurfaceInitialState as
a staging area.
The gtk-shell-surface implementation would either write to the
initial state, or directly to the window.

At the same, implement set_title and set_class too, because it's
easy enough.

https://bugzilla.gnome.org/show_bug.cgi?id=707128
2013-09-03 10:57:28 -04:00
Jasper St. Pierre
2d35e07fae wayland: Add support for set_opaque_region / set_input_region
https://bugzilla.gnome.org/show_bug.cgi?id=707019
2013-08-29 15:31:52 -04:00
Giovanni Campagna
aa15c09d54 Merge tag 'xrandr_branch_point' into wayland-kms-base
Conflicts:
	src/Makefile.am
	src/core/display.c
	src/core/screen-private.h
	src/core/screen.c
2013-08-27 10:07:39 +02:00
Giovanni Campagna
f09b9573f0 window: ignore skip-taskbar hint on parentless dialogs
Dialogs that don't have a parent should not be skip-taskbar,
otherwise they get lost and there is no way to recover them
(because they're not autoraised when activating the parent),
but toolkits and applications set the hint anyway.

https://bugzilla.gnome.org/show_bug.cgi?id=673399
2013-08-27 09:57:06 +02:00
Giovanni Campagna
bd3d5df9ce Remove HAVE_WAYLAND ifdefs
Wayland support is always enabled now.

https://bugzilla.gnome.org/show_bug.cgi?id=705497
2013-08-26 15:00:29 +02:00
Florian Müllner
bbf9358eba window: Fix a compiler warning 2013-08-23 23:09:58 -04:00
Giovanni Campagna
214f31257b Rework and consolidate monitor handling in MetaScreen
Consolidate all places that deal with output configuration in
MetaScreen, which gets it either from XRandR or from a dummy static configuration.
We still need to read the Xinerama config, even when running xwayland,
because we need the indices for _NET_WM_FULLSCREEN_MONITORS, but
now we do it only when needed.

https://bugzilla.gnome.org/show_bug.cgi?id=705670
2013-08-18 00:22:54 +02:00
Jasper St. Pierre
0e098249b1 Merge remote-tracking branch 'origin/master' into wayland 2013-08-13 10:44:09 -04:00
Jasper St. Pierre
12d2e1f600 Support _GTK_FRAME_EXTENTS
https://bugzilla.gnome.org/show_bug.cgi?id=705766
2013-08-13 10:40:15 -04:00
Giovanni Campagna
9a5f243f73 wayland: refactor window destruction and focus
The previous code was leaving focus fields dirty in MetaWaylandPointer
and MetaWaylandKeyboard at time (which could crash the X server
because of invalid object IDs)
The new code is more tighly integrated in the normal X11 code
for handling keyboard focus (meaning that the core idea of input
focus is also correct now), so that meta_window_unmanage() can
do the right thing. As a side benefit, clicking on wayland clients
now unfocus X11 clients.
For the mouse focus, we need to clear the surface pointer when
the metawindowactor is destroyed (even if the actual actor is
kept alive for effects), so that a repick finds a different pointer
focus.

https://bugzilla.gnome.org/show_bug.cgi?id=705859
2013-08-13 09:42:55 +02:00
Robert Bragg
40e820f551 Add support for stacking X and Wayland windows together
This breaks down the assumptions in stack-tracker.c and stack.c that
Mutter is only stacking X windows.

The stack tracker now tracks windows using a MetaStackWindow structure
which is a union with a type member so that X windows can be
distinguished from Wayland windows.

Some notable changes are:

Queued stack tracker operations that affect Wayland windows will not be
associated with an X serial number.

If an operation only affects a Wayland window and there are no queued
stack tracker operations ("unvalidated predictions") then the operation
is applied immediately since there is no server involved with changing
the stacking for Wayland windows.

The stack tracker can no longer respond to X events by turning them into
stack operations and discarding the predicted operations made prior to
that event because operations based on X events don't know anything
about the stacking of Wayland windows.

Instead of discarding old predictions the new approach is to trust the
predictions but whenever we receive an event from the server that
affects stacking we cross-reference with the predicted stack and check
for consistency. So e.g. if we have an event that says ADD window A then
we apply the predictions (up to the serial for that event) and verify
the predicted state includes a window A. Similarly if an event says
RAISE_ABOVE(B, C) we can apply the predictions (up to the serial for
that event) and verify that window B is above C.

If we ever receive spurious stacking events (with a serial older than we
would expect) or find an inconsistency (some things aren't possible to
predict from the compositor) then we hit a re-synchronization code-path
that will query the X server for the full stacking order and then use
that stack to walk through our combined stack and force the X windows to
match the just queried stack but avoiding disrupting the relative
stacking of Wayland windows. This will be relatively expensive but
shouldn't be hit for compositor initiated restacking operations where
our predictions should be accurate.

The code in core/stack.c that deals with synchronizing the window stack
with the X server had to be updated quite heavily. In general the patch
avoids changing the fundamental approach being used but most of the code
did need some amount of re-factoring to consider what re-stacking
operations actually involve X or not and when we need to restack X
windows we sometimes need to search for a suitable X sibling to restack
relative too since the closest siblings may be Wayland windows.
2013-08-10 19:13:49 -04:00
Robert Bragg
f9a11b3b18 wayland: Adds basic hybrid X + Wayland support
This adds support for running mutter as a hybrid X and Wayland
compositor. It runs a headless XWayland server for X applications
that presents wayland surfaces back to mutter which mutter can then
composite.

This aims to not break Mutter's existing support for the traditional X
compositing model which means a single build of Mutter can be
distributed supporting the traditional model and the new Wayland based
compositing model.

TODO: although building with --disable-wayland has at least been tested,
I still haven't actually verified that running as a traditional
compositor isn't broken currently.

Note: At this point no input is supported

Note: multiple authors have contributed to this patch:
Authored-by: Robert Bragg <robert@linux.intel.com>
Authored-by: Neil Roberts <neil@linux.intel.com>
Authored-by: Rico Tzschichholz.
Authored-by: Giovanni Campagna <gcampagna@src.gnome.org>
2013-08-10 19:13:48 -04:00
Robert Bragg
531be6c413 Track the X Shape input region and use it for picking
We now track whether a window has an input shape specified via the X
Shape extension. Intersecting that with the bounding shape (as required
by the X Shape extension) we use the resulting rectangles to paint
window silhouettes when picking. As well as improving the correctness of
picking this should also be much more efficient because typically when
only picking solid rectangles then the need to actually render and issue
a read_pixels request can be optimized away and instead the picking is
done on the cpu.
2013-08-10 19:13:48 -04:00
Florian Müllner
2103ff6a5c window: Don't force attached dialogs to be border-only
Originally attached dialogs did not have a titlebar, which the code
still assumes though it hasn't been true for a while; nowadays, the
actual look of attached dialogs is controlled by the theme.
As GTK+ recently gained the ability to set custom titlebars, we need
to support attached dialogs with either full borders (WM decorations)
or border-only (GTK+ titlebar).
Just remove the left-over assumption to make it work as expected.

https://bugzilla.gnome.org/show_bug.cgi?id=702764
2013-06-24 20:19:33 +02:00
Adel Gadllah
8ab136b7ea window: Make sure override_redirect window have correct monitor info
We need to update window->monitor on override_redirect windows as well, other
wise they may end up with an invalid struct which triggers and assert when
meta_window_is_monitor_sized is called.

https://bugzilla.gnome.org/show_bug.cgi?id=702564
2013-06-24 17:32:22 +02:00
Adel Gadllah
5205821fb9 window: Reuse current pointer position for monitor checks
Avoid a round trip to the xserver we already have the current position
anyway. Querying from the server on every move can cause the compositor to
stall during movement.
2013-06-23 21:24:41 +02:00
Jasper St. Pierre
df8234c5e3 window: Properly handle focusing override redirect windows
If an app pops up an OR window and sets input focus to it, like
Steam does, we'll think the focus window is null, causing us to
think the app is not focused.

OR windows should not be special if they get input focus, where
the input focus would be set to NULL. Instead, the window should
be marked as focused.

https://bugzilla.gnome.org/show_bug.cgi?id=647706
2013-05-22 13:46:15 -04:00
Dan Winship
7a4c808e43 display: clean up focus_window vs expected_focus_window
Mutter previously defined display->focus_window as the window that the
server says is focused, but kept display->expected_focus_window to
indicate the window that we have requested to be focused. But it turns
out that "expected_focus_window" was almost always what we wanted.

Make MetaDisplay do a better job of tracking focus-related requests
and events, and change display->focus_window to be our best guess of
the "currently" focused window (ie, the window that will be focused at
the time when the server processes the next request we send it).

https://bugzilla.gnome.org/show_bug.cgi?id=647706
2013-05-22 13:46:15 -04:00
Jasper St. Pierre
e430e051b7 window: Clean up the set_focused_internal function
Move things out of an indentation layer, and reshuffle
things around.

https://bugzilla.gnome.org/show_bug.cgi?id=647706
2013-05-22 12:35:04 -04:00
Jasper St. Pierre
696d9d2fa9 window: Merge got_focus/lost_focus to a new function
Make it a static function for now, but this will be a private
function soon, replacing meta_window_lost_focus. This should
contain no functional changes, only cosmetic indentation changes,
so best viewed with ignorews=1 or -w or -b, you know the drill.

https://bugzilla.gnome.org/show_bug.cgi?id=647706
2013-05-22 12:35:04 -04:00
Jasper St. Pierre
f6dd081acd window: Refactor "got focus" code
Just move this out to a separate function.

https://bugzilla.gnome.org/show_bug.cgi?id=647706
2013-05-22 12:35:04 -04:00
Jasper St. Pierre
a487d4dd01 window: Eliminate a potential race condition with _NET_WM_MOVERESIZE
Clients using _NET_WM_MOVERESIZE to start a drag operation may encounter
a race condition if the user presses and releases a mouse button very
fast, getting "stuck" in a grab state. While this is easily fixed with
the user pressing the button or hitting Escape as the EWMH spec suggests,
its's still a bit of annoyance for users.

After starting a grab operation, check that the button is actually pressed
by the client, and if not, cancel the grab operation. This prevents the
stuck grab in a race-free way, although it requires an extra round-trip
to the server.

With client-side decorations becoming more popular, the use of
_NET_WM_MOVERESIZE is on the rise, thus this bug is seen more frequently
than before.

https://bugzilla.gnome.org/show_bug.cgi?id=699777
2013-05-14 14:46:20 -04:00
Jasper St. Pierre
50b9042ac2 window: Add an accessor for whether the window can close
The shell will use this to determine whether to show a close
button in the overview.

https://bugzilla.gnome.org/show_bug.cgi?id=699269
2013-05-09 15:34:37 -04:00
Pavel Vasin
970a446bd8 window: Add missing chain-up for finalize()
https://bugzilla.gnome.org/show_bug.cgi?id=698710
2013-04-29 14:58:31 +02:00
Owen W. Taylor
8a33880c00 Don't configure a window before it responds to _NET_WM_SYNC_REQUEST
In 97a4cc8c, we accidentally lost the check that kept us from
sending multiple configures to a window before it responds to
_NET_WM_SYNC_REQUEST. So _NET_WM_SYNC_REQUEST stopped working
properly. Add a check back with the same effect.

https://bugzilla.gnome.org/show_bug.cgi?id=696091
2013-03-18 15:10:18 -04:00
Owen W. Taylor
7a8c45dda8 Handle spontaneous frames during a resize
During a resize, if we don't have a configure pending, then a counter
change shouldn't trigger anything other than the normal drawing:
it's just a spontaneous frame from the application. So don't try
to update the position or remove our timeout ID.

https://bugzilla.gnome.org/show_bug.cgi?id=696091
2013-03-18 15:10:18 -04:00
Owen W. Taylor
5ceffe86ee MetaScreen: Add tracking of whether there are fullscreen windows
Trying to track the fullscreen status outside of Mutter, as GNOME Shell
was doing previously, was very prone to errors, because Mutter has a
very tricky definition of when a window is set to be fullscreen and
*actually* acting like a fullscreen window.

 * Add meta_screen_get_monitor_in_fullscreen() and an
   ::in-fullscreen-changed signal. This allows an application to
   track when there are fullscreen windows on a monitor.

 * Do the computation of fullscreen status in a "later" function that
   runs after showing, so we properly take focus into account.

 * To get ordering of different phases right, add more values
   to MetaLaterType.

 * Add auto-minimization, similar to what was added to GNOME Shell
   earlier in this cycle - if a window is set to be fullscreen, but
   not actually fullscreen, minimize.

https://bugzilla.gnome.org/show_bug.cgi?id=649748
2013-03-18 13:05:22 -04:00
Florian Müllner
49df033b4e window: Remove obsolete code
Since the tile mode is now always reset on maximize(), this code
no longer does anything (not to mention that side-by-side tiled
windows haven't snapped back for a while now).

https://bugzilla.gnome.org/show_bug.cgi?id=682779
2013-03-18 16:43:08 +01:00
Florian Müllner
ac3465ffbb window: Reset tile-mode on maximization
We used to restore side-by-side tiling when unmaximizing, so we
kept the tile-mode during maximization. Since commit 10d53fc7d
there's no longer a good reason to do so, and it can result in
tile previews being shown erroneously on window drag operations
without motion (double-click on titlebar), so reset the tile
mode in maximize().

https://bugzilla.gnome.org/show_bug.cgi?id=682779
2013-03-18 16:43:08 +01:00
Florian Müllner
884ab602cb window: Force NORMAL layer for tile preview
The tile preview is expected to be shown underneath the focus window.
However the code that restacks the preview broke when override-redirect
windows were moved to a separate window group.
To fix, special-case tile previews to put them in the NORMAL layer.

https://bugzilla.gnome.org/show_bug.cgi?id=696053
2013-03-18 16:20:07 +01:00
Tomeu Vizoso
2efed44257 Add a meaningful name and description to all sections/files in the API docs
https://bugzilla.gnome.org/show_bug.cgi?id=695641
2013-03-14 18:11:04 -04:00
Tomeu Vizoso
be46869782 docs: Fix cross-reference generation
https://bugzilla.gnome.org/show_bug.cgi?id=676856
2013-03-14 18:02:43 -04:00
Tomeu Vizoso
e0b698d365 Fix gobs of gtk-doc warnings
https://bugzilla.gnome.org/show_bug.cgi?id=676856
2013-03-14 18:02:43 -04:00
Owen W. Taylor
97a4cc8c9b Make handling of windows that don't respond to _NET_WM_SYNC_REQUEST reliable
Previously, we were handling failure to respond to _NET_WM_SYNC_REQUEST
in the code path for throttling motion events. But this meant that
if a window didn't respond to _NET_WM_SYNC_REQUEST and there were no
motion events - for a keyboard resize, or after the end of the grab
operation - it would end up in a stuck state.

Use a separate per-window timeout to reliably catch the failure to respond
to _NET_WM_SYNC_REQUEST.

https://bugzilla.gnome.org/show_bug.cgi?id=694046
2013-03-14 08:01:28 -04:00
Owen W. Taylor
592374bc62 Fix freezing of windows with keyboard resizing
During resizing we froze window updates when configuring the
window, and unfroze the window updates when processing the
next resize. This wasn't absolutely reliable, because we might
not have a next resize. Instead tie window freezing more
directly to the current sync request value - a window is
frozen until it catches up with the last value we sent it
in _NET_WM_SYNC_REQUEST.

Testing with unresponsive clients showed that there was a bug
where window->disable_sync once set, would not actually disable
sync, but it *would* disable noticing that the client was
unresponsive for the next resize. Fix that by checking for
->disable_sync before sending _NET_WM_SYNC_REQUEST.

https://bugzilla.gnome.org/show_bug.cgi?id=694046
2013-03-14 07:59:57 -04:00
Owen W. Taylor
6e02fb80c4 Always send _NET_WM_FRAME_DRAWN for newly created windows
Send a _NET_WM_FRAME_DRAWN for each newly created window, as required
by the specification. This avoids a race where a window might be created
frozen but already unfrozen by the time we first see fetch the
counter value.

Remove a duplicate call to meta_compositor_set_updates_frozen() which
was called before the MetaWindowActor is created and hence did nothing.

https://bugzilla.gnome.org/show_bug.cgi?id=694771
2013-03-04 15:36:13 -05:00
Pavel Vasin
98b0a37442 Free memory allocated by XIQueryPointer()
https://bugzilla.gnome.org/show_bug.cgi?id=695135
2013-03-04 19:07:23 +01:00
Giovanni Campagna
e6d5e98c9d MetaWindow: fix annotation
caller-allocates makes sense only for structure, and gjs will complain
if used on anything else

https://bugzilla.gnome.org/show_bug.cgi?id=694801
2013-02-27 18:00:01 +01:00
Adel Gadllah
2a773e0c85 window: Add get_all_monitors
Add a method that returns the indices of the monitors a window
is on.

https://bugzilla.gnome.org/show_bug.cgi?id=646861
2013-02-25 22:05:42 +01:00
Ray Strode
2cafb8be2d window: fix meta_window_is_remote across hostname changes
meta_window_is_remote compares a cached copy of the system hostname
with the hostname of the client window
(as presented by the WM_CLIENT_MACHINE property).

Of course, the system hostname can change at any time, so caching
it is wrong. Also, the WM_CLIENT_MACHINE property won't necessarily
change when the system hostname changes, so comparing it with the
new system hostname is wrong, too.

This commit makes the code call gethostname() at the time
WM_CLIENT_MACHINE is set, check whether it's remote then, and cache
that value, rather than comparing potentially out of sync hostnames
later.

https://bugzilla.gnome.org/show_bug.cgi?id=688716
2013-02-20 16:02:10 -05:00
Adel Gadllah
e9709b7ff9 window: Add is_screen_sized method
Add a method that returns whether the window occupies
the whole screen (i.e all montiors).
2013-02-18 14:10:27 +01:00
Adel Gadllah
a6a9d3f448 window: Add meta_window_is_monitor_sized
We duplicate this checks in multiple places so lets add this to
avoid code duplication.
2013-02-17 21:42:01 +01:00
Owen W. Taylor
d900d83522 MetaWindowActor: Go back to freezing affecting actor geometry
We do, in fact, need freezing to affect window geometry, so that
move-resize operations (such as an interactive resize from the
left, or a resize of a popup centered by the application) occur
atomically.

So to make map effects work properly, only exclude the initial
placement of a window from freezing. (In the future, we may want
to consider whether pure moves of a window being done in response
to a user drag should also be excluded from freezing.)

Rename meta_window_sync_actor_position() to
meta_window_sync_actor_geometry() for clarity.

https://bugzilla.gnome.org/show_bug.cgi?id=693922
2013-02-15 21:48:25 -05:00
Owen W. Taylor
5876f2e3e5 Fix corner cases where _NET_WM_FRAME_DRAWN might be missed
The WM spec requires _NET_WM_FRAME_DRAWN to *always* be sent when
there is an appropriate update to the sync counter value. We were
potentially missing _NET_WM_FRAME_DRAWN when an application did a
spontaneous update during an interactive resize and during effects.
Refactor the code to always send _NET_WM_FRAME_DRAWN, even when
a window is frozen.

https://bugzilla.gnome.org/show_bug.cgi?id=693833
2013-02-14 16:21:26 -05:00
Owen W. Taylor
aeb589c176 MetaWindow: fix detecting the response to an extended _NET_WM_SYNC_REQUEST
During resizing, An odd counter value (indicating the beginning of a frame)
shouldn't cause us to redraw and start a new frame, only an even counter
value. This was causing the frozen state for the window frame counter to
overlap the frozen state for the resize, causing the window not to be
updated.

https://bugzilla.gnome.org/show_bug.cgi?id=693833
2013-02-14 16:21:14 -05:00
Owen W. Taylor
0503f6bb9a Consistently use meta_grab_op_is_resizing() for _NET_WM_SYNC_REQUEST
In different places we checked the grab op differently when determing
whether we are using _NET_WM_SYNC_REQUEST. This was somewhat covered
up previously by the fact that we only had a sync alarm when using
_NET_WM_SYNC_REQUEST, but that is no longer the case, so consistently
use meta_grab_op_is_resizing() everywhere.

https://bugzilla.gnome.org/show_bug.cgi?id=685463
2013-02-13 09:48:27 -05:00
Owen W. Taylor
87fe9685b5 Distinguish "no delay" frames from spontaneous drawing
When a client is drawing as hard as possible (without sleeping
between frames) we need to draw as soon possible, since sleeping
will decrease the effective frame rate shown to the user, and
can also result in the system never kicking out of power-saving
mode because it doesn't look fully utilized.

Use the amount the client increments the counter value by when
ending the frame to distinguish these cases:

 - Increment by 1: a no-delay frame
 - Increment by more than 1: a non-urgent frame, handle normally

https://bugzilla.gnome.org/show_bug.cgi?id=685463
2013-02-13 09:48:27 -05:00
Owen W. Taylor
790bfcad9a MetaWindow: always resize the frame first when we have synchronization
Resizing the frame triggers creation of a new backing pixmap for the
window, so we should do that first before we resize the client window
and mess up the contents of the old backing pixmap.

https://bugzilla.gnome.org/show_bug.cgi?id=685463
2013-02-13 09:48:26 -05:00
Owen W. Taylor
fbfab93c63 Send _NET_WM_FRAME_DRAWN messages
When the application provides the extended second counter for
_NET_WM_SYNC_REQUEST, send a client message with completion
information after the next redraw after each counter update
by the application.

https://bugzilla.gnome.org/show_bug.cgi?id=685463
2013-02-13 09:48:26 -05:00
Owen W. Taylor
70c0d39fa7 Add support for an extended style of _NET_WM_SYNC_REQUEST_COUNTER
If an application provides two values in _NET_WM_SYNC_REQUEST_COUNTER,
use that as a signal that the applications wants an extended behavior
where it can update the counter as well as the window manager. If the
application updates the counter to an odd value, updates of the
window are frozen until the counter is updated again to an even value.

https://bugzilla.gnome.org/show_bug.cgi?id=685463
2013-02-13 09:48:26 -05:00