153f37bdf5 (#1145) removed the
wlr_xwayland_is_unamanged function while fixing OR, because it was
belieived that it's supposed to work around the broken OR handling.
This was a misunderstanding. is_unmanaged is (while sort of a hack)
intended to work around inherent differences between "real" X sessions
and our Xwayland/wayland situation.
The main reason it exists is to support applications like rofi and dzen,
while not handing focus to other OR windows (which should *not* be
required).
Traditionally, these applications just grabbed input from X and didn't
need to be focused by any logic in the WM. Which of course doesn't work
in wayland compositors. So we have to give them focus in some way.
Giving *every* OR window focus, breaks other applications that don't
expect focus to change.
A testcase that was pointed out to me where wlr_xwayland_is_unamanged was
breaking things is https://github.com/swaywm/sway/issues/2128 (syncplay,
gitk, gitgui)
Supposedly it broke using keyboard to navigate the menus.
I can't reproduce this with this patch. The popups can be navigated as
long as the parent has focus.
Just install a SIG_IGN handler, which is defined by POSIX.1-2001 to
reap the child.
To test, spawn any process (e.g. GUI application) with a keybinding,
close that process, verify it doesn't show up as a <defunct> zombie in
ps(1) output.
Implement the basic logic for tablet-v2 tablet_pad's grabs. And plug in
the default grab.
Features like "holding" the focus should be implemented via grabs, like
they are for pointer and keyboard.
The override_redirect flag can change on configure notify and
on map notify. This adds an event to know when it changes.
This removes wlr_xwayland_surface_is_unmanaged which was wrongly
using the window type to decide whether the view should be
unmanaged.
A similar patch was proposed to Weston, but has never been
merged upstream [1].
[1]: https://patchwork.freedesktop.org/patch/211161/
The previous naming was based on the input-device capability names from
libinput.
With code that uses the libinput_tablet_tool and mapping into tablet-v2,
this is confusing, so the name is changed to follow the names used in
the protocol.
It's possible that a non-default keyboard grab exists when we are trying
to change focus. For example, say there is an XDG popup when we click on
a different window. This popup's keyboard grab will swallow any
keyboard_notify_enter(), meaning the newly-clicked window won't receive
keyboard input.
So, we cancel any existing grabs in roots_seat_set_focus(). Before this
fix, a window would have been set as active but not receive keyboard
entry.
Fixes#233.
Signed-off-by: Genki Sky <sky@genki.is>
The user-visible issue is that newly-mapped xdg-shell* windows would
sometimes start with their top-left-corner, rather than their center, in
the center of the screen. This is because get_size() would
conservatively fall back on (width, height) == (0, 0) if both
set_window_geometry() had not been called, and it found
view->wlr_surface to be NULL.
But, view->wlr_surface is only set to non-NULL in view_map(). We call
get_size() before this. Fortunately, the wlr_surface in question is
accessible via view->xdg_shell{,_v6}->surface, so always fall back on
that. We can assert its presence instead of further falling back on
(width, height) == (0, 0).
Signed-off-by: Genki Sky <sky@genki.is>
Compositors now have more control over how the backend creates its
renderer. Currently all backends create an EGL/GLES2 renderer, so
the necessary attributes for creating the context are passed to a
user-provided callback function. It is responsible for initializing
provided wlr_egl and to return a renderer. On fail, return 0.
Fixes#987
Before this change, a view would lose focus after clicking something that's not keyboard-interactive. This would lead to edge cases with layer-shell windows like input methods, which are pointer-only-interactive, but are not intended to change the state of any focus.
After clicking on something non-interactive, the current view was getting deactivated, but still received keyboard events. roots_seat_set_focus now changes both together in this case.
Makes the xwayland startup process two phased.
The first phase just initialises the X11 sockets.
The second phase starts the Xwayland server itself.
When starting xwayland lazily the second phase will be postponed until
a client has connected to the X11 socket.
Changes in behaviour:
The DISPLAY environment is now set immediately after the X11 sockets
are created.
When the Xwayland server is killed or crashes, the sockets will not be
recreated, but reused.
Fixes#849: Start up Xwayland lazily
On the X11 backend the cursor position might be outside the output
window so no output is returned leading to the assert to trigger. Use
sane fallback instead of crashing.
When e.g. running rootston under X11 it would otherwise crash when
closing rootston like:
#0 0x00007f0197da7327 in wl_list_remove () at /usr/lib/x86_64-linux-gnu/libwayland-server.so.0
#1 0x000056306fcee7fb in handle_destroy (listener=0x5630723a2948, data=0x5630723a5d20) at ../rootston/layer_shell.c:273
#2 0x00007f019800a552 in wlr_signal_emit_safe (signal=0x5630723a5e30, data=0x5630723a5d20) at ../util/signal.c:29
#3 0x00007f0197fef808 in layer_surface_destroy (surface=0x5630723a5d20) at ../types/wlr_layer_shell.c:169
#4 0x00007f0197ff0001 in client_handle_destroy (resource=0x56307229c4c0) at ../types/wlr_layer_shell.c:371
#5 0x00007f0197da2f30 in () at /usr/lib/x86_64-linux-gnu/libwayland-server.so.0
#6 0x00007f0197da77f9 in () at /usr/lib/x86_64-linux-gnu/libwayland-server.so.0
#7 0x00007f0197da301d in wl_client_destroy () at /usr/lib/x86_64-linux-gnu/libwayland-server.so.0
#8 0x00007f0197da30d8 in () at /usr/lib/x86_64-linux-gnu/libwayland-server.so.0
#9 0x00007f0197da4c12 in wl_event_loop_dispatch () at /usr/lib/x86_64-linux-gnu/libwayland-server.so.0
#10 0x00007f0197da344a in wl_display_run () at /usr/lib/x86_64-linux-gnu/libwayland-server.so.0
#11 0x000056306fcef069 in main (argc=3, argv=0x7ffd22032528) at ../rootston/main.c:83
since the output_destroy got already removed in handle_output_destroy.
With -O2:
../rootston/desktop.c: In function ‘desktop_surface_at’:
../rootston/desktop.c:714:18: error: ‘roots_output’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
if ((surface = layer_surface_at(roots_output,
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
&roots_output->layers[ZWLR_LAYER_SHELL_V1_LAYER_BOTTOM],
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
ox, oy, sx, sy))) {
~~~~~~~~~~~~~~~
==32557==ERROR: AddressSanitizer: SEGV on unknown address 0x000000000000 (pc 0x000000425f96 bp 0x7fff8ac19de0 sp 0x7fff8ac19d20 T0)
==32557==The signal is caused by a READ memory access.
==32557==Hint: address points to the zero page.
#0 0x425f95 in roots_seat_set_focus ../rootston/seat.c:737
#1 0x40bcd6 in roots_cursor_press_button ../rootston/cursor.c:272
#2 0x40c1f7 in roots_cursor_handle_button ../rootston/cursor.c:298
#3 0x42179b in handle_cursor_button ../rootston/seat.c:58
#4 0x7f1651062367 in wlr_signal_emit_safe ../util/signal.c:29
#5 0x7f165101b532 in handle_pointer_button ../types/wlr_cursor.c:344
#6 0x7f1651062367 in wlr_signal_emit_safe ../util/signal.c:29
#7 0x7f1650ff633b in handle_pointer_button ../backend/libinput/pointer.c:85
#8 0x7f1650ff5291 in wlr_libinput_event ../backend/libinput/events.c:215
#9 0x7f1650ff3990 in wlr_libinput_readable ../backend/libinput/backend.c:35
#10 0x7f1650d88c11 in wl_event_loop_dispatch (/lib64/libwayland-server.so.0+0x9c11)
#11 0x7f1650d87449 in wl_display_run (/lib64/libwayland-server.so.0+0x8449)
#12 0x418e90 in main ../rootston/main.c:81
#13 0x7f164ff7ef29 in __libc_start_main (/lib64/libc.so.6+0x20f29)
#14 0x405829 in _start (/home/shared/wayland/wlroots/build/rootston/rootston+0x405829)
introduced by #680
Check whether the newly focused view is the same as the one currently
fullscreen on that output, or override redirect and don't unfullscreen
in these cases.
The output fullscreen surfaces are drawn in front of everything, without
consideration for view z-order.
If a view is brought to front, unset any fullscreen view that would
cover this view to make sure the view is visible.
It was broken because the damage extents were rotated about its
own center, not about the center of the surface.
This adds a new wlr_region_rotated_bounds that rotates regions.
This allows us to have only one code path (for both non-rotated
views and rotated views) and optimizes rendering for rotated
views.
