compositor-drm: Delete drm_backend_set_modes

Even if we do have a framebuffer matching the mode, we immediately
schedule a repaint, meaning we either do work for no reason, or show
stale content before we bring up the new content.

Delete this and just let repaint deal with it.

Differential Revision: https://phabricator.freedesktop.org/D1481

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Armin Krezović <krezovic.armin@gmail.com>

compositor-drm: Store width and height inside drm_fb

This will be used so we can later determine the compatibility of drm_fbs
without needing to introspect external state.

Differential Revision: https://phabricator.freedesktop.org/D1487

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Armin Krezović <krezovic.armin@gmail.com>

compositor-drm: Use signed int for width/height

This makes it sign-compatible with weston_output->{width,height}.

Differential Revision: https://phabricator.freedesktop.org/D1486

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Quentin Glidic <sardemff7+git@sardemff7.net>
Reviewed-by: Armin Krezović <krezovic.armin@gmail.com>

compositor-drm: Use fb->fd consistently

Everyone else uses fb->fd rather than pulling the FD back out of GBM.
Use that in the destroy callback too.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Armin Krezović <krezovic.armin@gmail.com>
Differential Revision: https://phabricator.freedesktop.org/D1406

compositor-drm: Extract EGL destroy to helper

No functional change.

Differential Revision: https://phabricator.freedesktop.org/D1484

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Armin Krezović <krezovic.armin@gmail.com>

compositor-drm: Simplify drm_sprite_crtc_supported

No need to walk the CRTC list every time looking for CRTC indices, when we
already have the CRTC index stashed away. Taking the plane as an argument
also simplifies things a little for callers, and future-proofs for a
potential future KMS API which passes a list of supported CRTC IDs rather
than a bitmask of supported CRTC indices.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Armin Krezović <krezovic.armin@gmail.com>
Differential Revision: https://phabricator.freedesktop.org/D1407

compositor-drm: Comment struct members

Clarify the difference between crtc_id (DRM object) and pipe (index into
drmModeRes->crtcs array, possible_crtcs bitmask).

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Quentin Glidic <sardemff7+git@sardemff7.net>
Reviewed-by: Armin Krezović <krezovic.armin@gmail.com>
Differential Revision: https://phabricator.freedesktop.org/D1405

libweston-desktop: don't crash when getting the pid for X clients

X client's don't have a wl_client associated with their
weston_desktop_client, so make sure to not use it.

Signed-off-by: Giulio Camuffo <giuliocamuffo@gmail.com>
Reviewed-by: Quentin Glidic <sardemff7+git@sardemff7.net>

tests: Skip Xwayland test if binary isn't available

We know we're not going to succeed if the binary isn't installed, so
skip the test in that case.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

libweston: Make module loading safe against long paths

Avoid any buffer overflows here by checking we don't go over PATH_MAX
with stupid module names.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>

compositor-wayland: Destroy cursor images earlier

Destroying a wl_cursor will attempt to access the wl_display, which
we have just freed. Avoid a segfault by destroying the cursor images
before we destroy the display.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Dima Ryazanov <dima@gmail.com>

Don't prepend protocol/ to include paths

No need to add protocol/, as it's already handled by an explicit
compiler include path.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Dima Ryazanov <dima@gmail.com>

editor: Load a file if specified on command line

Add support for basic text file loading, to facilitate more expansive
testing of its UTF-8 text editing support.

Signed-off-by: Bryce Harrington <bryce@osg.samsung.com>
Reviewed-by: Yong Bakos <ybakos@humanoriented.com>

editor: Use parse_options() from shared for command line options

Also add a basic --help option

Signed-off-by: Bryce Harrington <bryce@osg.samsung.com>
Reviewed-by: Yong Bakos <ybakos@humanoriented.com>

compositor-wayland: Set frame callback for Pixman

Fixing 89c2f637b9, also set the output's frame_cb for the Pixman
renderer, not just GL. Fixes a segfault when using compositor-wayland
with --use-pixman.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Dima Ryazanov <dima@gmail.com>

gl-renderer: Fix an invalid write when closing a Weston window

Call eglMakeCurrent before destroying the native EGL window, similar to what
other sample clients are already doing.

Signed-off-by: Dima Ryazanov <dima@gmail.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

compositor-wayland: Fix a use after free

When a window is being closed, the frame_done callback often runs after
the output is already destroyed, i.e:

  wayland_output_start_repaint_loop
  input_handle_button
    wayland_output_destroy
  frame_done

To fix this, destroy the callback before destroying the output.

(Also, fix the type of output in frame_done: it's passed in
a wayland_output, not a weston_output.)

Signed-off-by: Dima Ryazanov <dima@gmail.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

libweston-desktop/xwayland: add is_mapped handling for XWAYLAND

The xwayland window type XWAYLAND is not handled by the shell at all,
instead libweston-desktop maps such surfaces itself. However, it forgot
to set weston_surface::is_mapped and weston_view::is_mapped.

weston_surface::is_mapped affects the behaviour of weston_view_unmap()
and weston_surface_attach().

weston_view::is_mapped affects the behaviour of weston_view_unmap() and
weston_view_destroy().

When manually mapping a window of type XWAYLAND, mark both the view and
surface as mapped. This follows the expections in libweston, even though
the meaning of is_mapped is not clearly defined for either surface or
view.

Also, when the XWAYLAND window is manually unmapped, unmap the
weston_surface. This updates weston_surface::is_mapped to reflect the
state. However, as a side-effect it will also unmap all sibling views,
should any exist.

v2: rename surface_base to wsurface
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Quentin Glidic <sardemff7+git@sardemff7.net>
Acked-by: Daniel Stone <daniels@collabora.com>

libweston-desktop/xwayland: XWAYLAND surfaces are never 'added'

Add documentation (asserts) that show that windows of types XWAYLAND are
never registered with the shell.

Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Quentin Glidic <sardemff7+git@sardemff7.net>

libweston-desktop/xwayland: clarify 'added' logic

Setting to a constant is much easier to read and grep for than setting to
a computed variable.

There are no functional changes.

Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Quentin Glidic <sardemff7+git@sardemff7.net>

libweston-desktop/xwayland: window type XWAYLAND cannot have a parent

Add an assert to ensure that a window of type XWAYLAND is never
attempted with a parent. Following the code though, the assert can be
made even stricter by allowing only TRANSIENT to have a parent.

This is essentially adding documentation.

v2: use stricter assert
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Daniel Stone <daniels@collabora.com>

XWM: debug position and size on map request

Helps debugging initial placement problems.

Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Daniel Stone <daniels@collabora.com>

XWM: debug position on create_notify

Helps debugging X11 window positioning issues.

Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Daniel Stone <daniels@collabora.com>

compositor-x11: fix segfault when use_pixman is true

Signed-off-by: Ryo Munakata <ryomnktml@gmail.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

Don't include version.h from compositor.h

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewd-by: Bryce Harrington <bryce@osg.samsung.com>

tests: Extend subsurface place_{above,below} hierarchy

Following on from b8c16c995b, extend the family tree being tested by
place_above and place_below a little, ensuring that subsurfaces can't be
placed above or below surfaces which are related to them, but aren't
their immediate parent or sibling.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>

build: fix wayland-backend distcheck

Fix the following error from 'make distcheck':

  CC       libweston/wayland_backend_la-compositor-wayland.lo
../../libweston/compositor-wayland.c:54:51: fatal error: xdg-shell-unstable-v6-client-protocol.h: No such file or directory
 #include "xdg-shell-unstable-v6-client-protocol.h"

Files generated with the scanner belong in nodist_*_SOURCES, not with
the regular sources.

Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Daniel Stone <daniels@collabora.com>

weston-terminal: Fix crash on first keystroke

Since 894b3rcc634 weston-terminal will crash on first keystroke if you
fail to create an xkb compose state.  This can happen if you don't have
a Compose file.

Instead, now we just return uncomposed symbols.

Signed-off-by: Derek Foreman <derekf@osg.samsung.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

compositor: allow using nested parent as a subsurface sibling

The parent of a subsurface can be used as a sibling in the place_below
and place_above calls. However this did not work when the parent is
nested, so fix the sibling check and add a test to check this case.

Signed-off-by: Daniel Stone <daniels@collabora.com>

xwayland: Create the drag-and-drop window in weston_wm_dnd_init

Just to keep it hidden so far... A lot of the plumbing necessary to
handle x11->wayland drag and drop is missing, and the current
partial handling gets in the middle for X11 drag-and-drop itself
to work.

The approach is well directed, but needs some further work, till
then, just keep our fake drag-and-drop target hidden. This allows
drag-and-drop to work between X11 clients in Xwayland, and avoids
a crash with (currently unhandled) wl_resource-less data sources.

Fixes https://bugs.freedesktop.org/show_bug.cgi?id=94218

Signed-off-by: Carlos Garnacho <carlosg@gnome.org>
Reviewed-by: Daniel Stone <daniels@collabora.com>

libweston: Add move (without scale) animation

Signed-off-by: Quentin Glidic <sardemff7+git@sardemff7.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>

desktop-shell/client: Add left/right panel position (basic) support

Signed-off-by: Quentin Glidic <sardemff7+git@sardemff7.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>

desktop-shell/client: Introduce helper variables

These variables will be much more useful in the following commit.
The indentation is off to avoid future diff noise.

Signed-off-by: Quentin Glidic <sardemff7+git@sardemff7.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>

desktop-shell/client: Parse the clock format once

Signed-off-by: Quentin Glidic <sardemff7+git@sardemff7.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>

desktop-shell/client: Use a proper enum for clock format

Signed-off-by: Quentin Glidic <sardemff7+git@sardemff7.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>

desktop-shell: Properly position the panel surface

Now weston actually supports putting the panel at the bottom of the
screen.

Signed-off-by: Quentin Glidic <sardemff7+git@sardemff7.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>

desktop-shell: Compute panel size without using output

Signed-off-by: Quentin Glidic <sardemff7+git@sardemff7.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>

desktop-shell/client: Add bottom panel support

Signed-off-by: Quentin Glidic <sardemff7+git@sardemff7.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>

gl: Don't declare variables in for loop

Fixes compilation error introduced by 43cea54c:

  libweston/gl-renderer.c:2862:2: error: ‘for’ loop initial declarations
  are only allowed in C99 mode
    for (unsigned i = 0; i < ARRAY_LENGTH(swap_damage_ext_to_entrypoint);
    i++) {
    ^

Signed-off-by: Bryce Harrington <bryce@osg.samsung.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

compositor-wayland: Port to xdg-shell-v6

v2:

 - Keep wl_shell code around until xdg_shell is declared stable.

Signed-off-by: Armin Krezović <krezovic.armin@gmail.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

gl-renderer: add support of WL_SHM_FORMAT_YUYV

This patch allow gl-renderer to accept WL_SHM_FORMAT_YUYV buffers.
This is the pixel format supported by most of the USB webcams.

v2:
 - fix hsub Vs vsub inversion

Signed-off-by: Vincent Abriou <vincent.abriou@st.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

compositor-x11: Move vfunc setting from set_size to enable

Signed-off-by: Armin Krezović <krezovic.armin@gmail.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

compositor-wayland: Destroy shm buffers on output disable

Reviewed-by: Armin Krezović <krezovic.armin@gmail.com>

compositor-wayland: Properly clean up on backend destroy

Also remove a wrong XXX comment.

Reviewed-by: Armin Krezović <krezovic.armin@gmail.com>

compositor-wayland: Simplify fullscreen output surface handling

Reviewed-by: Armin Krezović <krezovic.armin@gmail.com>

compositor-wayland: Move vfunc setting from set_size to enable

Reviewed-by: Armin Krezović <krezovic.armin@gmail.com>

weston-editor: Free preferred_language in text_entry_destroy

Signed-off-by: Silvan Jegen <s.jegen@gmail.com>
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
Signed-off-by: Daniel Stone <daniels@collabora.com>

clients/nested-client: call eglTerminate/eglReleaseThread at exit

Do a minimalistic teardown at program exist.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

tests/buffer-count-test: teardown after the test

Might be a bit of an overkill, but still. One should cleanup after
themselves.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

gl-renderer: use weston_platform_destroy_egl_surface wrapper

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

clients/window: use weston_platform_destroy_egl_surface wrapper

v2: Use correct (destroy) API (Dan)

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

clients/subsurfaces: use weston_platform_destroy_egl_surface wrapper

... over a direct eglDestroySurface call. Provides symmetry in the
create/destroy paths.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

clients/simple-egl: use weston_platform_destroy_egl_surface wrapper

... over a direct eglDestroySurface call. Provides symmetry in the
create/destroy paths.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

clients/nested-client: destroy the EGL surface during teardown

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

shared/platform.h: introduce weston_platform_destroy_egl_surface()

Introduce the weston_platform_destroy_egl_surface() wrapper to
complement the weston_platform_create_egl_surface() one.

We'll use the former with the next patches trhoughout weston to
consistently destroy the surface as needed.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

gl-renderer: add support for EGL_KHR_swap_buffers_with_damage

Extension is identical to the EXT one, yet we need to check for the KHR
abbreviated extension name + entry-point.

v2: s/foo/swap_damage_ext_to_entrypoint/ (Eric, Daniel)

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com> (v1)
Reviewed-by: Daniel Stone <daniels@collabora.com>

simple-egl: add support for EGL_KHR_swap_buffers_with_damage

Functionally identical to the EXT version of the extension.

v2: s/foo/swap_damage_ext_to_entrypoint/ (Eric, Daniel)
v3: do the above sed for real (Frank)

[daniels: Fixed signed vs. unsigned warning.]

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com> (v1)
Reviewed-by: Daniel Stone <daniels@collabora.com>

libweston: Move text_backend_* to weston.h

Signed-off-by: Quentin Glidic <sardemff7+git@sardemff7.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>

Ignore the OSC code for desktop notifications

In Fedora, bash is configured to display a desktop notification when a command
finishes (and the terminal is not focused). weston-terminal complains about it;
let's silence it.

Signed-off-by: Dima Ryazanov <dima@gmail.com>
Acked-by: Daniel Stone <daniels@collabora.com>

libweston: remove unused function declaration of weston_compositor_top

Signed-off-by: Ryo Munakata <ryomnktml@gmail.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

xwayland: Fix X11 lock file size confusion

The X11 lock file was somewhat opaque. Into a sized array of 16
characters, we previously read 11 bytes. 61beda653b fixed the parsing of
this input to ensure that we only considered the first 10 bytes: this
has the effect of culling a LF byte at the end of the string.

This commit more explicitly NULLs the entire string before reading, and
trims trailing LF characters only.

It also adds some documentation by way of resizing pid, an explicit size
check on snprintf's return, and comments.

Verified manually that it emits lock files with a trailing \n, as Xorg
does. Also verified manually that it ignores misformatted lock files,
but accepts either \n or \0 in the trailing position.

Related Mutter issue: https://bugzilla.gnome.org/show_bug.cgi?id=774613

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>

text-input: Clear out context->input pointer

If we destroy the text_input (e.g. due to surface deactivation) whilst
the IM manager (the client holding the input_method_context resource,
usually weston-keyboard) sends an event, we can hit a segfault in
text-backend. This is because we free the text_input structure, but
don't actually clear the context->input structure even when we send the
deactivate event.

This is clearly intended to be catered for, since context->input is
always checked for NULL before we relay any events.

This is enough to fix one cause of text-test failing, but it's
ultimately error-prone until we have a no-op test-shell; there is still
a race where weston-desktop-shell can launch weston-keyboard before
text-test manages to bind zwp_text_input_manager.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Maniphest Tasks: https://phabricator.freedesktop.org/T7615
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>

xwayland: Fix crash when run with no input device

Starting an xterm with no input device led to a crash
because weston_wm_pick_seat() was returning garbage and
weston_wm_selection_init() was trying to use the garbage.

Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
Reviewed-by: Giulio Camuffo <giuliocamuffo@gmail.com>
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>

xwayland: fix lock file parsing

Patch 139fcabe7cdb1f2296bf02ef917aaab84e00cd4e "xwayland: Improve error
checking for strtol call" caused a regression in the X11 unix socket
lock file parsing. Before that patch, only the first 10 characters were
considered for parsing. After the patch, the newline as the 11th
character caused strtol() to stop parsing at the 10th character which
was then considered an error as not the whole input was consumed.

The effect of the regression was that no X11 lock files were ever deemed
stale, hence stale lock files were never removed. Up till now, I have
accumulated 37 lock files, and Weston complaining for each of them on
every start it cannot parse them.

Fix this by terminating the string at the expected newline character.

Also, it looks like 'pid' was being used uninitialized, risking strtol()
reading past the end of the array. This patch fixes that too.

Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Daniel Stone <daniels@collabora.com>

Get rid of the window_create_menu function

It's currently unused, and there's actually no way to use it correctly.

The caller cannot free the menu that was created:
- the function only returns the window, not the menu
- there's no public API to destroy a menu object

Signed-off-by: Dima Ryazanov <dima@gmail.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

terminal: Silence maybe-uninitialized warning

clients/terminal.c: In function 'redraw_handler':
clients/terminal.c:213:28: warning: 'machine.unicode' may be used
uninitialized in this function [-Wmaybe-uninitialized]
  struct utf8_state_machine machine;
                            ^~~~~~~

clients/terminal.c: In function 'handle_char':
clients/terminal.c:213:28: warning: 'machine.unicode' may be used
uninitialized in this function [-Wmaybe-uninitialized]

Warning produced by GCC 5.3 and 6.1, with -O3.

'I found it weird that the compiler wouldn’t see that, so I re-checked
the code.

I think with -O3, this specific "for" is compile-time unlooped, and
utf8_next_char inlined.  And there is *one* path that can keep
machine.state to utf8state_start, thus triggering the warning.

Without -O3, the function is globally tagged as “changing unicode”, so
no warning is produced.

[...]

Side note: I picked 0 as the default value, but maybe in this case
0xfffd would be better?'

Signed-off-by: Quentin Glidic <sardemff7+git@sardemff7.net>
Acked-by: Emil Velikov <emil.l.velikov@gmail.com>

input: Update keyboard serial on press and release

Other compositors such as mutter update the keyboard serial for both key
press and key release, unlike weston which updates it only on key press.

When dealing with popup windows which require a match in serials, if the
event that caused the popup to be shown is a key release, then the popup
would be dismissed.

This occurs when navigating gtk+ sub-menus using the keyboard.

Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Bugzilla: https://bugzilla.gnome.org/show_bug.cgi?id=768017

gl-renderer: Use EGL_KHR_no_config_context

This patch makes use of recently implemented
EGL_KHR_no_config_context extension in Mesa,
which superseeds EGL_MESA_configless_context.

See also (and the follow-up patch):

https://lists.freedesktop.org/archives/mesa-dev/2016-September/128510.html

v2:

 - Extend existing infrastructure for EGL_MESA_configless_context
   per suggestion from Emmanuel Gil Peyrot.

Signed-off-by: Armin Krezović <krezovic.armin@gmail.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

clients: Fix weston-scaler source-only mode to use integer width and height

If only the source of a viewport is set, the width and height must
be integer or the protocol mandates that the compositor generate an
error.  This is because using only the source is a crop, and the
width and height become the surface size - all surface sizes must
be integer.

Weston was fixed to generate this error in bb32ccc0, however the
test app continued to use fractional co-ordinates when run as
weston-scaler -s  (which only sets the viewport source)

This leaves fractional width/height for the other cases, but uses
integer for the crop-only mode.  The descriptions in the help text
are still accurate with this change, but weston-scaler -s no longer
exits with an error.

Signed-off-by: Derek Foreman <derekf@osg.samsung.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

xdg-shell: validate positioner object

According to the xdg-shell v6 protocol a positioner object is only
complete if both the size and its anchor rectangle are set. Ensure the
weston clients do this and let weston be more strict on checking if a
client has done so.

This also fixes weston-terminal popups not showing up on gnome-shell
3.22.

Signed-off-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Yong Bakos <ybakos@humanoriented.com>

ivi-shell: describe members of type wl_list

I wrote comments on which list they are used with

Signed-off-by: Emre Ucan <eucan@de.adit-jv.com>
[Pekka: fix surface_list to layer_list, add ivi_layout_transition_set::transition_list]
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>

ivi-shell: clear unused struct members

I removed unused members of ivi_layout* data
structs.

Signed-off-by: Emre Ucan <eucan@de.adit-jv.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>

gl-renderer: Refactor gl_renderer_output_window_create()

This change refactors gl_renderer_output_window_create() to separate out
window surface creation code from output common creation code.

Signed-off-by: Miguel A Vico Moya <mvicomoya@nvidia.com>
Reviewed-by: Andy Ritger <aritger@nvidia.com>
Reviewed-by: James Jones <jajones@nvidia.com>
[Pekka: rebased and removed unused 'gr' and 'ec']
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>

gl-renderer: Add <platform_attribs> param to gl_renderer_display_create

This change adds <platform_attribs> parameter to
gl_renderer_display_create() in case we ever want to pass non-NULL.

Signed-off-by: Miguel A Vico Moya <mvicomoya@nvidia.com>
Reviewed-by: Andy Ritger <aritger@nvidia.com>
Reviewed-by: James Jones <jajones@nvidia.com>
[Pekka: removed notes about EGLOutput]
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>

gl-renderer: Accept non-NULL empty <visual_id> arrays

This change modifies egl_choose_config() to accept a non-NULL but empty
<visual_id> array (i.e. n_ids == 0)

Signed-off-by: Miguel A Vico Moya <mvicomoya@nvidia.com>
Reviewed-by: Andy Ritger <aritger@nvidia.com>
Reviewed-by: James Jones <jajones@nvidia.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>

gl-renderer: Rename <attribs> param of gl_renderer to <config_attribs>

This change renames <attribs> parameter of gl_renderer_display_create()
and gl_renderer_output_window_create() to <config_attribs> to explain
which attribs it is.

Signed-off-by: Miguel A Vico Moya <mvicomoya@nvidia.com>
Reviewed-by: Andy Ritger <aritger@nvidia.com>
Reviewed-by: James Jones <jajones@nvidia.com>
[Pekka: remove notes about EGLOutput]
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>

gl-renderer: Rename gl_renderer::output_create to output_window_create

No functional change. This patch renames gl_renderer_output_create() to
gl_renderer_output_window_create(), which is something more descriptive
of what the function does.

Signed-off-by: Miguel A Vico Moya <mvicomoya@nvidia.com>
Reviewed-by: Andy Ritger <aritger@nvidia.com>
Reviewed-by: James Jones <jajones@nvidia.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>

gl-renderer: Rename gl_renderer_create to gl_renderer_display_create

No functional change. This patch only renames gl_renderer_create() to
gl_renderer_display_create(), which is something more descriptive of
what the function does.

Signed-off-by: Miguel A Vico Moya <mvicomoya@nvidia.com>
Reviewed-by: James Jones <jajones@nvidia.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>

gl-renderer: conditionally call query_buffer while gl_renderer_attach

While gl_renderer_attach, query_buffer should be call only if the
query_buffer function exists ie when has_bind_display is true.

v2:
 - Take into account Giulio's remark. Use has_bind_display viariable to test if
   EGL_WL_bind_wayland_display extension is supported.

Signed-off-by: Vincent Abriou <vincent.abriou@st.com>
Reviewed-by: Giulio Camuffo <giuliocamuffo@gmail.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

libweston: fix building issue when EGL support is not enabled

weston-egl-ext.h has been include in compositor-xx.c file in order to
define EGL_PLATFORM_xxx_KHR extensions used by the compositors.
But in case EGL support is not enabled, all EGL related definition must
be skipped except EGL_PLATFORM_xxx_KHR that must be still defined to
allow compositor-xx.c to build.

Signed-off-by: Vincent Abriou <vincent.abriou@st.com>
Reviewed-by: Giulio Camuffo <giuliocamuffo@gmail.com>

xwayland: Move includes to avoid deprecation warnings

compositor.h already helpfully defines WL_HIDE_DEPRECATED for us, so we
don't get warnings about wl_buffer (in particular) being deprecated when
we have wayland-server headers defining it as deprecated, and then
wayland-client headers using the type.

Move it to be before all our other includes, so we actually make use of
it.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Yong Bakos <ybakos@humanoriented.com>
Tested-by: Yong Bakos <ybakos@humanoriented.com>

clients: Make XKB compose support conditional

Debian Jessie's version of libxkbcommon is too old for compose support,
so rather than force people to upgrade, let's make it conditional.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Daniel Díaz <daniel.diaz@linaro.org>
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>

Add configuration option for no input device.

As it has been discussed in the past [1], running Weston
without any input device at launch might be beneficial for
some use cases.

Certainly, it's best for the vast majority of users (and
the project) to require an input device to be present, as
to avoid frustration and hassle, but for those brave souls
that so prefer, this patch lets them run without any input
device at all.

This introduces a simple configuration in weston.ini:
  [core]
  require-input=true

True is the default, so no behavioral change is introduced.

[1] https://lists.freedesktop.org/archives/wayland-devel/2015-November/025193.html

Signed-off-by: Daniel Díaz <daniel.diaz@linaro.org>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>

libweston-desktop: destroy wl_shell_surface after the wl_surface is destroyed

The wl_shell_surface spec says that it is destroyed automatically by the
server when the wl_surface is destroyed, and indeed it does not have a
destroy request. So, do that.

Signed-off-by: Giulio Camuffo <giuliocamuffo@gmail.com>
Reviewed-by: Quentin Glidic <sardemff7+git@sardemff7.net>

compositor-wayland: Convert fullscreen flag to bool

Signed-off-by: Armin Krezović <krezovic.armin@gmail.com>
Reviewed-by: Quentin Glidic <sardemff7+git@sardemff7.net>

simple-im: Rename source file to be consistent with other client tools

Except for weston-info, client source files are not prefixed "weston-".

Signed-off-by: Bryce Harrington <bryce@osg.samsung.com>
Reviewed-by: Jan Arne Petersen <janarne@gmail.com>

simple-im: Use the appropriate enum names to xkb calls

XKB_KEYMAP_COMPILE_NO_FLAGS and XKB_CONTEXT_NO_FLAGS are both defined as
0 so no functional change here, just improved code clarity.

Signed-off-by: Bryce Harrington <bryce@osg.samsung.com>
Reviewed-by: Jan Arne Petersen <janarne@gmail.com>

simple-im: Use returns from main() consistently

Signed-off-by: Bryce Harrington <bryce@osg.samsung.com>
Reviewed-by: Jan Arne Petersen <janarne@gmail.com>

window: Fix 'devlivery' misspelling

clients: Add XKB compose key support

This adds single-symbol compose support using libxkbcommon's compose
functionality.  E.g., assuming you have the right alt key defined as
your compose key, typing <RAlt>+i+' will produce í, and <RAlt>+y+= will
produce ¥.  This makes compose key work for weston-editor,
weston-terminal, weston-eventdemo, and any other clients that use
Weston's window.* routines for accepting and managing keyboard input.

Compose sequences are loaded from the system's standard tables.  As
well, libxkbcommon will transparently load custom sequences from the
user's ~/.XCompose file.

Note that due to limitations in toytoolkit's key handler interface, only
compose sequences resulting in single symbols are supported.  While
libxkbcommon supports multi-symbol compose strings, support for passing
text buffers to Weston clients is left as future work.

This largely obviates the need for the weston-simple-im input method
client, which had provided a very limited compose functionality that was
only available in clients implementing the zwp_input_method protocol,
and with no mechanism to load system or user-specified compose keys.

Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=53648
Signed-off-by: Bryce Harrington <bryce@osg.samsung.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>

compositor-wayland: Convert draw_initial_frame to boolean

Signed-off-by: Armin Krezović <krezovic.armin@gmail.com>
Reviewed-by: Quentin Glidic <sardemff7+git@sardemff7.net>

compositor-wayland: Convert sprawl flag to boolean

Signed-off-by: Armin Krezović <krezovic.armin@gmail.com>
Reviewed-by: Quentin Glidic <sardemff7+git@sardemff7.net>

compositor-wayland: Convert use-pixman flag to boolean

Signed-off-by: Armin Krezović <krezovic.armin@gmail.com>
Reviewed-by: Quentin Glidic <sardemff7+git@sardemff7.net>

gl-renderer: add support of WL_SHM_FORMAT_NV12

This patch allow gl-renderer to accept WL_SHM_FORMAT_NV12 buffers.

Signed-off-by: Vincent Abriou <vincent.abriou@st.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

gl-renderer: add support of WL_SHM_FORMAT_YUV420

This patch allow gl-renderer to accept WL_SHM_FORMAT_YUV420 buffers.

In a gstreamer pipeline, the support of the WL_SHM_FORMAT_YUV420 by
weston avoid pixel conversion between software decoders and waylandsink.
Indeed, software decoders output I420 (YUV420 planar) that will
match with WL_SHM_FORMAT_YUV420.

Signed-off-by: Vincent Abriou <vincent.abriou@st.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

libweston: include weston-egl-ext.h in drm, x11 and wayland compositor

As to what is done for gl-renderer.c, weston-egl-ext.h should be
include in compositor-drm.c, compositor-x11.c and compositor-wayland.c.
This fix building issue with GPU that does not have EGL_PLATFORM_xxx_KHR
in their extension header file eglext.h.

Signed-off-by: Vincent Abriou <vincent.abriou@st.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

compositor-rdp: Properly destroy the renderer and pixman image

Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Signed-off-by: Armin Krezović <krezovic.armin@gmail.com>

libweston: Drop requirement of setting mm_width/mm_height in backends

They were required for transitional phase in order not to
break previous weston_output_init(). Now, they can even
be initialized on enable, or left with defaults if backend
doesn't support them.

Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Signed-off-by: Armin Krezović <krezovic.armin@gmail.com>

libweston: Remove weston_backend_output_config structure

Reviewed-by: Quentin Glidic <sardemff7+git@sardemff7.net>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Signed-off-by: Armin Krezović <krezovic.armin@gmail.com>

weston: Rename weston_output_init_pending() to weston_output_init()

v2:

 - Rebased for latest changes.

v3:

 - Rebased for changes in wayland backend.

Reviewed-by: Quentin Glidic <sardemff7+git@sardemff7.net>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Signed-off-by: Armin Krezović <krezovic.armin@gmail.com>