The code in _gdk_wayland_window_dispose was not safe against
being called twice - it would call g_hash_table_destroy twice
on the known_globals hash table, the second time operating on
freed memory. It was also leaking the list of async_roundtrips.
After fixing both of these issues, the displayclose testcase
now works on Wayland.
We call gdk_wayland_window_hide_surface when the window gets
destroyed, and in this case, the frame clock might not exist
anymore.
This was showing up in the displayclose testcase.
While we do not have subwindows in Wayland, we do create an
artificial root window. When the display is closed, the root
window gets destroyed, causing recursing to be true for the
toplevel windows.
As the protocol is still considered unstable (meaning not backward
compatible), we should, as stated in the protocol, only bind the version
advertised is the version we implement.
https://bugzilla.gnome.org/show_bug.cgi?id=753856
The extra reference will be held from GdkEventPrivate data, so there's
a common place to all events. Without this, events queued after devices/
capabilities disappear (eg. on TTY switch) might hold invalid pointers.
Windowing level operations on those devices (queries, grabs...) are
expected to fail at that time, but we should hold meaningful data for
the regular event handling paths.
https://bugzilla.gnome.org/show_bug.cgi?id=753185
Prior to this patch, the ID of the GtkApplication was always used for
clients which were GtkApplications. This would only be guaranteed to be
correct for D-Bus activatable programs. As a result, all
non-D-Bus-activatable applications would set the wrong ID making the
shell unable to find the corresponding .desktop file.
This change makes it so that the GDK backend always uses the name
passed to g_set_prgname, or the default value if not explicitly set, as
this more often corresponds to the .desktop file.
This means that in order to make D-Bus activatable applications set the
correct application ID, they must, for now, manually call
g_set_prgname() with their application ID (basename of the .desktop
file).
If g_get_prgname() returns NULL, fallback to gdk_get_program_class()
even though it will most likely never be correct according to the
xdg_surface.set_app_id specification.
https://bugzilla.gnome.org/show_bug.cgi?id=746435
wl_log() currently logs using G_LOG_LEVEL_ERROR
(which is fatal). The wayland client library doesn't
expect this behavior. It uses wl_log to log recoverable
errors.
This commit changes the log level to G_LOG_LEVEL_DEBUG
instead.
https://bugzilla.gnome.org/show_bug.cgi?id=753635
On wayland, the gestures protocol defines a wl_pointer_gestures global
object, that will match in number with wl_seats, swipe and pinch
interfaces can be obtained from it, which events are translated into
GdkEventTouchpadSwipe/Pinch events.
Each gesture type has its separate GdkEvent struct, and begin/update/
end/cancel event types.
There is support for multi-finger swipe (3-4 fingers), and 2-finger
rotate/pinch gestures.
Only a drag context which was created with 'drag_begin' will be
guaranteed to have a source window at all times. Thus, in finalize we
cannot assume we can retrieve a GdkDisplay from the source_window
pointer since it may be NULL. Though, the display is only needed for
contexts created via 'drag_begin' thus we can retrieve it after
checking that is the case.
https://bugzilla.gnome.org/show_bug.cgi?id=749339
We need to be mapped to have a gtk_surface and thus be able to do
requests on it so we need to save the modal hint and apply it when we
get mapped so that code that sets the hint before showing a window
doesn't get ignored.
https://bugzilla.gnome.org/show_bug.cgi?id=753138
When dealing with selection events, we might see windows from
other screens in the requestor field. The current x11 backend
code fails to wrap these in a foreign GdkWindow, since we
don't have the corresponding GdkScreen anymore. Work around
this by creating such 'foreign screens' on demand. We still
maintain the 1:1 relation between the display and the screen
returned by gdk_display_get_default_screen().
https://bugzilla.gnome.org/show_bug.cgi?id=721398
The shadow will be drawn in the wrong place in those cases so all we can
do is disable it. This fixes double shadows drawn around menus, popups
and tooltips.
https://bugzilla.gnome.org/show_bug.cgi?id=734984
If a drag ends inside a known window, set the dest_window field
in the drag context. This information is needed to implemented
notebook tab dragging.
https://bugzilla.gnome.org/show_bug.cgi?id=752638
-1 means that we have no specific preference for an initial
fullscreen monitor, and -1 is less than the number of monitors,
so we would end up accessing invalid memory. Prevent that.
https://bugzilla.gnome.org/show_bug.cgi?id=752875
Call chain:
- input_data_cb
- broadway_server_read_all_input_nonblocking (input)
- broadway_input_free (input)
(now input is invalid)
attempt to use input->active -> use-after-free
Make broadway_server_read_all_input_nonblocking return a boolean, TRUE
if the input was valid, FALSE otherwise. This allows input_data_cb to
detect whether the input was gone or not.
https://bugzilla.gnome.org/show_bug.cgi?id=741685
Windows does not send any release key event for one of the shift keys
when both shift keys were pressed together. This commit solves
the problem by sending the extra release key event for the shift key
which was released as first, when the other shift key is released.
Other modifiers (e.g. Ctrl, Alt) do not have this problem.
https://bugzilla.gnome.org/show_bug.cgi?id=751721
So a window can be maximized/zoomed again after being moved away from
its maximized position. This makes the zoom button on non-CSD windows
work as before.
Instead of using the default zoom behaviour use the internal
maximized state for selecting our own zoom target. This makes
zooming work for CSD windows where for some reason the
given default zoom target is the current window frame itself
resulting in a shadowless window of the same size.
While this makes the zoom button behave a bit different as expected
it makes things more consistent with other platforms and fixes CSD
zooming.
Prior to this patch, the hotspot would be passed in buffer coordinate
space. Where this were ever tested, i.e. in a patched mutter, the
server interpreted them incorrectly, which meant it went undiscovered.
In the updated mutter patches the incorrect behavior in GTK+ was
discovered due to the behavior in mutter was corrected.
In the themed cursor case, the dimensions were not correctly scaled
either, but this had no negative visible effect because the dimension is
only used for reporting damage tracking, and passing a bigger damage
region than surface has no negative visible effects.
https://bugzilla.gnome.org/show_bug.cgi?id=752616
When using frame times from _NET_WM_FRAME_DRAWN and _NET_WM_FRAME_TIMINGS, we
were treating them as local monotonic times, but they are actually extended-precision
versions of the server time, and need to be translated to monotonic times in the
case where the X server and client aren't running on the same system.
This fixes rendering stalls when using X over a remote ssh connection.
https://bugzilla.gnome.org/show_bug.cgi?id=741800
Avoid using a stale timestamp (from the last user interaction with the
application) when a message arrives from D-Bus requesting that a new
window be created.
In this case the most-correct thing that we can do is to use no
timestamp at all.
We modify gdk_x11_display_set_startup_notification_id() to allow a NULL
value to mean "reset everything" and then call this function
unconditionally on receipt of D-Bus activation requests. The result
will be that a missing desktop-startup-id in the platform-data struct
will reset the timestamp.
Under their default configuration metacity and mutter will both map
windows presented with no timestamp in the foreground. This could
result in false-positive, but there is very little we can do about that
without the original timestamp from the user event.
https://bugzilla.gnome.org/show_bug.cgi?id=752000
If we don't find Xft values in the X resource db, simply fall
back to the values that are hardcoded in /etc/X11/Xresources
anyway. Extra trickery with likely-made-up screen dimensions
is not going to yield better results, and only makes for a
deeper rabbit hole when debugging.
Support was added for GDK_HINT_ASPECT in
gdk_quartz_window_set_geometry_hints though with one restriction:
min_aspect and max_aspect have to be equal, which I believe corresponds
to the most common usage. A warning will be printed if this condition is
not met but min_aspect will be used anyway.
Under Wayland, fullscreen/maximized windows may not cover the entire
area when a size increment is specified.
Ignore size increments for fullscreen/maximized windows just like most
window managers do under X11 so that windows with size increments can
still be fullscreen or fully maximized under Wayland as well.
https://bugzilla.gnome.org/show_bug.cgi?id=751368
We used to "invalidate" scroll valuators, so the next scroll event could
be used as the base for the next scroll deltas. This has the inconvenience
that it invariably consumes the first event received after enter and,
due to interactions with WM overeager passive button grabs, there's a
possibility we don't scroll at all if we receive interleaved "smooth
scroll" XI_Motion events and XI_Enter events (Normally triggered by regular
scroll wheels in mice).
In order to fix this, and at the expense of some sync-call overhead on
XI_Enter events (one XIQueryDevice call per slave device), query the
current scroll valuator state for all the slaves of the entered pointer,
so we do know beforehand the right base values. If new devices are plugged
while the pointer is on top of the client, the initialized scroll values
will match the valuators'.
https://bugzilla.gnome.org/show_bug.cgi?id=750994https://bugzilla.gnome.org/show_bug.cgi?id=750870
gdk_x11_device_xi2_window_at_position() may attempt to push/pop
a few error trap pairs while traversing the window tree. Move it
outside the server grab, and around the multiple XIQueryPointer
calls we may do here.
https://bugzilla.gnome.org/show_bug.cgi?id=751739
We mistakenly forced the "STRING" type, which was able to confuse higher
layer helpers like gtk_selection_data_get_uris(). This fixes a crash
happening anytime a drop is attempted on a GtkPlacesSidebar.
Currently, due to the lack of progress information in the Wayland DnD
protocol, we assume a DnD operation is finished after the first
data_source.send is finished (It's either that or leaving stuck grabs).
This however breaks previous assumptions that dest widgets can request
the data multiple times, even in response to GtkWidget::drag-motion.
This leaves us with a NULL owner for the DnD atom when we aren't
finished receiving wl_data_source events yet, causing a crash.
This commit fixes the crash, the behavior left is still far from
desirable though...
And force the ungrab on it, instead of the slave, in the case of
local DnD drop. This avoids confusions on the pointer events spawn
from DnD, as GDK doesn't think anymore those are from a slave device.
Most namely, it fixes the stuck grab when finishing DnD on the
same app it was started from.
We currently only hold the last offer received, which is wrong, as both
are independent and have different life cycles.
This means we have to store per-selection wl_data_offer and targets, and
maintain these as appropriate from the clipboard/DnD specific entry points.
This oddly can be reproduced with weston+weston-dnd, when dragging
anything from GTK+ into weston-dnd, it will insist on picking its
custom application/x-wayland-dnd-flower mimetype, and this request
forwarded by the compositor, even if GTK+ didn't announce it on
its wl_data_source mimetype list. (What should probably happen here
is that the request is silenced, and/or weston-dnd picks (null))
This should be harmless, we are leaking though the fd in that case,
because the emission of GdkEventSelection on an unhandled mimetype
results in NOP. In order to avoid this, we should check whether the
mimetype is supported at all on the backend code and possibly close
the fd, this involves storing these in the first place.
https://bugzilla.gnome.org/show_bug.cgi?id=751414
If the other peer requests data too fast (too rare/unlikely though),
we might receive multiple gdk_wayland_selection_request_target() calls
with no ending gdk_wayland_selection_check_write(), in which case the
fd is leaked as no GOutputStream was created to take over it.
https://bugzilla.gnome.org/show_bug.cgi?id=751414
We weren't catching all the places where the AsyncWriteData operation
should be cancelled, which could happen if we repeatedly request the
same target on different fds.
https://bugzilla.gnome.org/show_bug.cgi?id=751414
At the moment we create the AsyncWriteData, the ownership of the
fd is granted to the GOutputStream, and the fd set to -1, so at
this moment we're just silently getting EBADFD.
This partially reverts 25885ca600, the initialization of .fd
to -1 is valid and stays though.
https://bugzilla.gnome.org/show_bug.cgi?id=751414
On X11 this is something the windowing system does for us, which the
wayland backend should emulate, being grabs completely client-side.
So, if the grab and current focus windows differ, make sure we emit
focus/crossing events as it corresponds to the grab device.
This was being done so only on pointers. Internally, a GdkDeviceGrabInfo
is kept for each of the master pointer/keyboard, failing to do this for
keyboards results in a stuck keyboard grab.
https://bugzilla.gnome.org/show_bug.cgi?id=748892
The fd must be closed on async_write_data_free(), but we should also
initialize it to -1 so gdk_wayland_selection_check_write() doesn't wrongly
pick the stdin fd.
https://bugzilla.gnome.org/show_bug.cgi?id=751414
A subsurface positioning operation only takes effect when the parent
surfaces state is applied. If a subsurface is mapped and positioned, but
the parent surface state is not immediately committed, the relative
position of the subsurface is undefined and may be placed incorrectly.
To avoid this undefined state, always request that the parent surface
should be committed after mapping a subsurface so that the position
operation will take effect.
https://bugzilla.gnome.org/show_bug.cgi?id=751098
Background patterns are often updated when style changes. In many cases,
the new pattern will match the previous. We can optimize out the
invalidation that will occur upon resetting the same pattern.
An pass_through window is something you can draw in but does not
affect event handling. Normally if a window has with no event mask set
for a particular event then input events in it go to its parent window
(X11 semantics), whereas if pass_through is enabled the window below
the window will get the event. The later mode is useful when the
window is partially transparent. Note that an pass-through windows can
have child windows that are not pass-through so they can still get events
on some parts.
Semantically, this behaves the same as an regular window with
gdk_window_set_child_input_shapes() called on it (and re-called any
time a child is changed), but its far more efficient and easy to use.
This allows us to fix the testoverlay input stacking test.
https://bugzilla.gnome.org/show_bug.cgi?id=750568https://bugs.freedesktop.org/show_bug.cgi?id=90917
If a menu was not attached to any widget, we try to calculate its
position given where the grabbed pointer is and what window has its
focus. Previously we failed to do so if a "transfer window" was used
for the grab, and this patch adds a code path that, if the menu window
itself didn't have the grab, look for the transfer window and get the
grab device from there.
https://bugzilla.gnome.org/show_bug.cgi?id=748951
If a position was already explicitly set, don't try to guess the
position of popup menus by looking at the pointer position, just use
the set coordinates.
https://bugzilla.gnome.org/show_bug.cgi?id=748951
According to the xdg-shell protocol specification the (x, y) coordinates
passed when creating a popup surface is relative to top left corner of
the parent surface, but prior to this patch, if the parent surface
was an xdg_surface, we'd position it relative to top left corner of the
window geometry of that xdg_surface.
https://bugzilla.gnome.org/show_bug.cgi?id=749717
