Our layer shell implementation assigns every layer surface to an output
on creation. It tracks this output using the output field on the
underlying wlr_layer_surface_v1 structure. As such, much of the existing
code assumes that output is always non-NULL and omits NULL checks
accordingly.
However, there are currently two cases where we destroy a
sway_layer_surface and output is NULL. The first is when we can't find
an output to assign the surface to and destroy it immediately after
creation. The second is when we destroy a surface in response to its
output getting destroyed, as we set output to NULL in
handle_output_destroy() before we call wlr_layer_surface_v1_destroy(),
which is what calls the appropriate unmap and destroy callbacks.
The former case doesn't cause any problems, since we haven't even
allocated a sway_layer_surface at that point or registered any
callbacks. The latter case, however, currently triggers a crash (#6120)
if a popup is visible, since our popup_handle_unmap() implementation
can't handle a NULL output.
To fix this issue, keep output set until right before we free the
sway_layer_surface. All we need to do is remove some of the cleanup
logic from handle_output_destroy(), since as of commit c9060bcc12
("layer-shell: replace close() with destroy()") that same logic is
guaranteed to be happen later when wlroots calls handle_destroy() as
part of wlr_layer_surface_v1_destroy().
This lets us remove some NULL checks from other unmap/destroy callbacks,
which is nice. We also don't need to check that the wlr_output points to
a valid sway_output anymore, since we unset that pointer after disabling
the output as of commit a0bbe67076 ("Address emersions comments on
output re-enabling") Just to be safe, I've added assertions that the
wlr_output is non-NULL wherever we use it.
Fixes#6120.
The existing code gives this error when compiled with GCC 12:
../sway/server.c: In function ‘server_init’:
../sway/server.c:217:75: error: ‘%d’ directive output may be truncated writing between 1 and 11 bytes into a region of size 8 [-Werror=format-truncation=]
217 | snprintf(name_candidate, sizeof(name_candidate), "wayland-%d", i);
| ^~
../sway/server.c:217:66: note: directive argument in the range [-2147483647, 32]
217 | snprintf(name_candidate, sizeof(name_candidate), "wayland-%d", i);
| ^~~~~~~~~~~~
../sway/server.c:217:17: note: ‘snprintf’ output between 10 and 20 bytes into a destination of size 16
217 | snprintf(name_candidate, sizeof(name_candidate), "wayland-%d", i);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Because i is never negative, this is a false positive, but it is easy to
change i to unsigned to silence the error.
active_keyboard may be NULL, in which case an invalid pointer could be
passed to wlr_input_method_keyboard_grab_v2_send_modifiers. This
procedure call is unnecessary since wlroots commit 372a52ec "input
method: send modifiers in set_keyboard", so the call can simply be
removed.
Fixes#6836.
Currently, a floating window that's been fullscreened can send us
xdg_toplevel::move, and we'll enter seatop_move_floating, which lets us
drag the surface around while it's fullscreen. We don't want
this--fullscreen surfaces should always be aligned to the screen--so add
the same check that seatop_default already does when entering this mode.
Tested with Weston's weston-fullscreen demo, which sends a move request
if you click anywhere on its surface.
When REAPER submenu is closed `XCB_CLIENT_MESSAGE` with type
`NET_ACTIVE_WINDOW` is sent to set focus to parent menu.
Closes: https://github.com/swaywm/sway/issues/6324
Commit 37d7bc6998 ("transaction: Only wait for ack from visible
views") introduced a check which uses view_is_visible() to check if a view
is still visible on the screen. However view_is_visible() will early
return in case the node is in the destroying state. This is incorrect
for transactions, since a destroying view which is visible will trigger
configure events for other clients. This bug was visible when repeatedly
opening and closing two views side by side, since we ignore the
destroying node we get a frame where the still open view is shown with
the old configure values and the rest is the desktop background. The
next frame is than correct again.
Fix this by considering destroying views as visible, we correctly wait
for them and send the configure events to other views in time, fixing
the background flicker.
Fixes#6473
`popup_unconstrain` uses view coordinates to init the output box for
popups. However wlroots expects the box to be set in a toplevel surface
coordinate system, which is not always equal to view. The difference
between those is a window geometry set via xdg-shell.
GTK4 reserves some space for client-side decoration and thus has a
window with top left corner not matching to (0, 0) of a surface. The box
calculated without taking that into account was slightly shifted
compared to the actual output and allowed to position part of the popup
off screen.
SUID privilege drop is needed for the "builtin"-backend of libseat,
which copied our old "direct" backend behavior for the sake of
compatibility and ease of transition.
libseat now has a better alternative in the form of seatd-launch. It
uses the normal seatd daemon and libseat backend and takes care of SUID
for us.
Add a soft deprecation warning to highlight our future intent of
removing this code. The deprecation cycle is needed to avoid surprises
when sway no longer drops privileges.
Followup on 4e4898e90f.
If a view quickly maps and unmaps repeatedly, there will be multiple
destroyed containers with same view in a single transaction. Each of
these containers will then try to destroy this view, resulting in use
after free.
The container should only destroy the view if the view still belongs
to the container.
Simple reproducer: couple XMapWindow + XUnmapWindow in a loop followed
by XDestroyWindow.
See #6605
We currently track the focus of a seat in two ways: we use a list called
focus_stack to track the order in which nodes have been focused, with
the first node representing what's currently focused, and we use a
variable called has_focus to indicate whether anything has focus--i.e.
whether we should actually treat that first node as focused at any given
time.
In a number of places, we treat has_focus as implying that a focused
node exists. If it's true, we attempt to dereference the return value of
seat_get_focus(), our helper function for getting the first node in
focus_list, with no further checks. But this isn't quite correct with
the current implementation of seat_get_focus(): not only does it return
NULL when has_focus is false, it also returns NULL when focus_stack
contains no items.
In most cases, focus_stack never becomes empty and so this doesn't
matter at all. Since focus_stack stores a history of focused nodes, we
rarely remove nodes from it. The exception to this is when a node itself
goes away. In that case, we call seat_node_destroy() to remove it from
focus_stack and free it. But we don't unset has_focus if we've removed
the final node! This lets us get into a state where has_focus is true
but seat_get_focus() returns NULL, leading to a segfault when we try to
dereference it.
Fix the issue both by updating has_focus in seat_node_destroy() and by
adding an assertion in seat_get_focus() that ensures focus_stack and
has_focus are in sync, which will make it easier to track down similar
issues in the future.
Fixes#6395.
[1] There's some discussion in #1585 from when this was implemented
about whether has_focus is actually necessary; it's possible we could
remove it entirely, but for the moment this is the architecture we have.
cairo_image_surface_create can fail, e.g. when running out of
memory or when the size is too big. Avoid crashing in this case.
Closes: https://github.com/swaywm/sway/issues/6531
Now output_begin_destroy emits the node::destroy event similar to
workspace_begin_destroy. It currently has no listeners, since they
listen to output::disable or wlr_output::destroy instead.
