There were a few cases of 'goto out' in main() that did not set ret to
EXIT_FAILURE. Shell failing to init is the one I hit when writing tests
for ivi-shell.
Rather than adding a few more 'ret = EXIT_FAILURE', make that the
default and remove the redundant assignments. When Weston exits
properly ec->exit_code will take care of the exit code.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Emilio Pozuelo Monfort <emilio.pozuelo@collabora.co.uk>
Make the sanity check more explicit and log a warning if it happens.
Small negative values are ok because it just means the compositor is
lagging behind, or more likely the user specified a too long repaint
window.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-By: Derek Foreman <derekf@osg.samsung.com>
Reviewed-by: Bryce Harrington <b.harrington@samsung.com>
This timer delays the output_repaint towards the end of the refresh
period, reducing the time from repaint to present.
The length of the repaint window can be set in weston.ini.
The call to weston_output_schedule_repaint_reset() is delayed by one
more period. If we exit the continuous repaint loop (set
output->repaint_scheduled to false) in finish_frame, we may call
start_repaint_loop() unnecessarily. The problem case was actually
observed with two outputs on the DRM backend at 60 Hz, and 7 ms
repaint-window. During a window move, one output was constantly falling
off the continuous repaint loop and introducing additional one frame
latency, leading to jerky window motion. This code now avoids the
problem.
Changes in v2:
- Rename repaint_delay_timer to repaint_timer and
output_repaint_delay_handler to output_repaint_timer_handler.
- When computing the delay, take the current time into account. The timer
uses a relative timeout, so we have to subtract any time already gone.
Note, that 'gone' may also be negative. DRM has a habit of predicting
the page flip timestamp so it may be still in the future when we get the
completion event.
- Do also a sanity check 'msec > 1000'. In the unlikely case that
something fails to provide a good timestamp, never delay for more than
one second.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-By: Derek Foreman <derekf@osg.samsung.com>
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
Create a new function weston_compositor_read_presentation_clock() to
wrap the clock_gettime() call for the Presentation clock.
Reading the presentation clock is never supposed to fail, but if it
does, this will notify about it. I have not seen it fail yet, though.
This prepares for new testing features in the future that might allow
controlling the presentation clock. Right now it is just a convenience
function for clock_gettime().
All presentation clock readers are converted to call this new function
except rpi-backend's rpi_flippipe_update_complete(), because it gets its
clock id via a thread-safe mechanism. There shouldn't be anything really
thread-unsafe in weston_compositor_read_presentation_clock() at the
moment, but might be in the future, and weston core is not expected to
need to be thread-safe.
This is based on the original patch by
Cc: Derek Foreman <derekf@osg.samsung.com>
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-By: Derek Foreman <derekf@osg.samsung.com>
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
This will be used by pixman-renderer.
v2: Fix doc typo.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-By: Derek Foreman <derekf@osg.samsung.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Add matrix representations of these two transformations. Future patches
will leverage these to simplify the coordinate transformation code.
[Pekka: commit message]
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-By: Derek Foreman <derekf@osg.samsung.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
We can greatly simplify weston_output_transform_coordinate now by simply
multiplying by the output matrix and converting the result to fixed point.
This patch fixes zoomed input behaviour on the nested backends (x11,
wayland) which use absolute input coordinates. And probably also
absolute input devices. The patch that broke this was "zoom: Use pixels
instead of GL coordinates".
Signed-off-By: Derek Foreman <derekf@osg.samsung.com>
[Pekka: adjusted coding style and message]
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-By: Derek Foreman <derekf@osg.samsung.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Previously, weston_output.matrix was in GL coordinates and therefore only
really useful for the GL backend.
This breaks zoom, which will be fixed by the following patch:
zoom: Use pixels instead of GL coordinates
[Pekka: added a comment to compositor.h, message]
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-By: Derek Foreman <derekf@osg.samsung.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Add API for setting a clip ('scissor' in the code) rectangle per view,
in surface coordinates. Ivi-shell requires this feature to be able to
implement the IVI Layer Manager API, which includes clipping of
surfaces.
The names weston_view_set_mask() and weston_view_set_mask_infinite()
mirror the existing weston_layer_set_mask*() functions.
This view clipping complements the weston_layer clipping, because view
clipping is defined in surface local coordinates, while layer
mask/clipping is defined in global coordinates.
View clipping requires explicit support from the renderers. Therefore a
new Weston capability bit is added: WESTON_CAP_VIEW_CLIP_MASK. Shells
(and all users) of this new API are required to check the capability bit
is set before using the API. Otherwise the rendering will not be what
they expect.
View clips are inherited through the transformation inheritance
mechanism. However, there are restrictions. The clip rectangle can be
set only on the root view of a transformation inheritance tree. The
additional transformations in child views must not rotate the coordinate
axes. These restrictions avoid corner cases in clip inheritance, and
keep the renderer implementations as simple as they are right now.
Renderers only need to do an additional intersection with the clip
rectangle which is always aligned to the surface coordinate system.
For more details, see the API documentation in the patch.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Nobuhiko Tanibata <NOBUHIKO_TANIBATA@xddp.denso.co.jp>
Tested-by: Nobuhiko Tanibata <NOBUHIKO_TANIBATA@xddp.denso.co.jp>
Reviewed-By: Derek Foreman <derekf@osg.samsung.com>
Expand weston_compositor_pick_view() so it is easier to read. Use
short-hand variables, that make it easier to add one more test in the
future.
Write the output coordinate pointers only when returning non-NULL.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Nobuhiko Tanibata <NOBUHIKO_TANIBATA@xddp.denso.co.jp>
Tested-by: Nobuhiko Tanibata <NOBUHIKO_TANIBATA@xddp.denso.co.jp>
Reviewed-By: Derek Foreman <derekf@osg.samsung.com>
All things everywhere, except this one case, assume weston_plane::damage
is in global coordinates. Document it.
view_accumulate_damage() is wrong in converting damage to plane
coordinates (similar to global coordinate except translated). Fix this
by removing the unwanted translation, and use only global coordinates.
We have not seen this bug manifest in real life because we get lucky:
the origin of the primary plane is always at 0, 0. We do not use
non-primary planes, except cursor plane on DRM backend where the actual
damage coordinates are ignored.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
There are two call sites, one is already having a pixman_box32_t it
needs to call view_compute_bbox() with. The other call site will have a
box32_t when view clipping gets implemented.
Change view_compute_bbox() to take a pixman_box32_t as the input
argument, and convert call sites.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
Also fixes a theoretical memory leak as the region was never fini'd.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
The fix is not trivial, so I want to document the problem before I
forget about it again.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
This is an optional API that will be implemented by the renderers. It
allows to fetch the current contents of a surface, essentially the
buffer contents from a client buffer, converted to an RGBA format.
This is meant as a debugging API. The implementations may be heavy and
cause a stall, so they are not intended to be used often during normal
operations.
Renderers are expected to convert whatever data a surface has to a
single RGBA format.
Changes in v2:
- remove stride and format arguments from the API
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
v1 Tested-by: Nobuhiko Tanibata <NOBUHIKO_TANIBATA@xddp.denso.co.jp>
Reviewed-by: Daniel Stone <daniels@collabora.com>
weston_view::transform.boundingbox is made to include the layer mask,
which removes the need for masked_boundingbox.
The following were using boundingbox when they should have used
masked_boundingbox:
- drm_output_prepare_overlay_view() uses boundingbox to compute overlay
position, source and destination coordinates.
- drm_assign_planes() uses boundingbox for view overlap checks.
- is_view_not_visible() uses boundingbox, but nothing will show outside
the layer mask.
- weston_surface_assign_output() intersects boundingbox with output
region to choose the primary output for a surface.
- weston_view_assign_output() intersects boundingbox with output region
to pick the outputs the view is on.
This patch essentially changes all those cases to use the masked
boundingbox.
Therefore there are no cases which would need the boundingbox without
the layer mask, and we can convert boundingbox into masked and remove
the left-over member.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
[v2: don't move the decl of 'mask' in weston_view_update_transform]
Reviewed-By: Giulio Camuffo <giuliocamuffo@gmail.com>
Turns out there were no users of weston_view::transform.opaque,
everything was already using transform.masked_opaque. Therefore
repurpose transform.opaque as masked_opaque, and remove masked_opaque
member.
Now this opaque region in global coordinates is clipped by the layer
mask, if set. There are no cases where you would need the opaque region
without the effect of layer mask.
Also add a note in compositor.h, that changing view's layer counts as
changing geometry, which requires calling weston_view_geometry_dirty()
to let all derived state update.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-By: Giulio Camuffo <giuliocamuffo@gmail.com>
PRESENTATION_FEEDBACK_KIND_ZERO_COPY is a flag that needs to be set for
each surface separately. Some surfaces may be zero-copy (as defined by
Presentation feedback) while some are not.
A complication with Weston is that a surface may have multiple views on
screen. All copies (views) of the surface are required to be zero-copy
for the ZERO_COPY flag to be set.
Backends set per-view feedback flags during the assing_planes hook, and
then Weston core collects the flags from all views of a surface.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Tested-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Change weston_output_finish_frame() signature so that backends are
required to set the flags, that will be reported on the Presentation
'presented' event. This is meant for output-wide feedback flags. Flags
that vary per wl_surface are subject for the following patch.
All start_repaint_loop functions use the special private flag
PRESENTATION_FEEDBACK_INVALID to mark, that this call of
weston_output_finish_frame() cannot trigger the 'presented' event. If it
does, we now hit an assert, and should then investigate why a fake update
triggered Presentation feedback.
DRM:
Page flip is always vsync'd, and always gets the completion timestamp
from the kernel which should correspond well to hardware. Completion is
triggered by the kernel/hardware.
Vblank handler is only used with the broken planes path, therefore do
not report VSYNC, because we cannot guarantee all the planes updated at
the same time. We cannot set the INVALID, because it would abort the
compositor if the broken planes path was ever used. This is a hack that
will get fixed with nuclear pageflip support in the future.
fbdev:
No vsync, update done by copy, no completion event from hardware, and
completion time is totally fake.
headless:
No real output to update.
RDP:
Guessing that maybe no vsync, fake time, and copy make sense (pixels
sent over network). Also no event that the pixels have been shown?
RPI:
Presumably Dispmanx updates are vsync'd. We get a completion event from
the driver, but need to read the clock ourselves, so the completion time
is somewhat unreliable. Zero-copy flag not implemented though it would
be theoretically possible with EGL clients (zero-copy is a per-surface
flag anyway, so in this patch).
Wayland:
No information how the host compositor is doing updates, so make a safe
guess without assuming vsync or hardware completion event. While we do
get some timestamp from the host compositor, it is not the completion
time. Would need to hook to the Presentation extension of the host
compositor to get more accurate flags.
X11:
No idea about vsync, completion event, or copying. Also the timestamp is
a fake.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Tested-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Acked-by: Mario Kleiner <mario.kleiner.de@gmail.com>
load_modules currently ignores errors signalled by both
weston_load_module and module_init, and instead always returns 0. Its
return value appears to be checked in callers, so we most likely want to
propagate any errors.
Signed-off-by: Ondřej Majerech <majerech.o@gmail.com>
Reviewed-by: Derek Foreman <derekf@osg.samsung.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Logging is activated and deactivated with the debug key binding 't'.
When activated, it creates a new log file, where it records the events.
The log file contains events and detailed object information entries in
JSON format, and is meant to be parsed in sequence from beginning to the
end.
The emitted events are mostly related to the output repaint cycle, like
when repaint begins, is submitted to GPU, and when it completes on a
vblank. This is recorded per-output. Also some per-surface events are
recorded, including when surface damage is flushed.
To reduce the log size, events refer to objects like outputs and
surfaces by id numbers. Detailed object information is emitted only as
needed: on the first object occurrence, and afterwards only if
weston_timeline_object::force_refresh asks for it.
The detailed information for surfaces includes the string returned by
weston_surface::get_label. Therefore it is important to set
weston_timeline_object::force_refresh = 1 whenever the string would
change, so that the new details get recorded.
A rudimentary parser and SVG generator can be found at:
https://github.com/ppaalanen/wesgr
The timeline logs can answer questions including:
- How does the compositor repaint cycle work timing-wise?
- When was the vblank deadline missed?
- What is the latency from surface commit to showing the new content on
screen?
- How long does it take to process the scenegraph?
v2: weston_surface::get_description renamed to get_label.
v3: reafctor a bit into fprint_quoted_string().
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
When printing out logs from Weston's actions, mainly for debugging, it
can be very difficult to identify the different surfaces. Inspecting
the configure function pointer is not useful, as the configure functions
may live in modules.
Add vfunc get_label to weston_surface, which will produce a short,
human-readable description of the surface, which allows identifying it
better, rather than just looking at the surface size, for instance.
Set the label function from most parts of Weston, to identify cursors and
drag icons, and panels, backgrounds, screensavers and lock surfaces, and
the desktop shell's application surfaces.
v2: renamed 'description' to 'label', so we get
weston_surface_set_label_func().
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
calloc (and zalloc) set the allocated memory to 0, so there's really no
need to do it manually.
Signed-off-by: Bryce Harrington <bryce@osg.samsung.com>
Reviewed-by: Marek Chalupa <mchqwerty@gmail.com>
[Pekka: dropped the src/evdev.c hunk.]
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Weston's idle timeout can already be set via the '-i' command-line
option, but this patch lets users specify it also via weston.ini.
Note that the command-line option takes precedence over the .ini,
should the option be set by both.
This patch also Updates weston.ini man page with idle-timeout bits
https://bugs.freedesktop.org/show_bug.cgi?id=83921
Signed-off-by: Frederic Plourde <frederic.plourde@collabora.co.uk>
Reviewed-by: Bryce Harrington <b.harrington@samsung.com>
Most of the backends do their own parsing of transform strings, so let's
put that all in the same place (compositor.c/h)
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
If there are several things that cause an exit, keep the error code from
the first one, not the last one. The latter ones could be just fallout
from the first.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Currently, once we've reached our main's wl_display_run(), we always
return ret=EXIT_SUCCESS when weston terminates through wl_display_terminate.
This patch makes it possible to specify another return value by setting
prior to terminating Weston. This is useful for automated tests that want
to report test failures to the overlying testing infrastructure.
Signed-off-by: Frederic Plourde <frederic.plourde@collabora.co.uk>
[Pekka: fixed some tabs.]
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
commit 6ae7bc982f accidentally made weston_output_mode_switch_temporary
send done events when it shouldn't.
Signed-off-by: Derek Foreman <derekf@osg.samsung.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
This breaks weston_output_mode_switch() into 3 functions:
weston_output_mode_set_native()
weston_output_mode_switch_to_temporary()
weston_output_mode_switch_to_native()
Differences from previous behaviour:
SET_NATIVE didn't set current_scale (now it does)
SET_TEMPORARY could set mode and scale independently - now it can't.
Signed-off-by: Derek Foreman <derekf@osg.samsung.com>
Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Commit 280e7dd918 introduced a bug in the
return value of weston_subsurface_is_synchronized().
Signed-off-by: Carlos Olmedo Escobar <carlos.olmedo.e@gmail.com>
Reviewed-by: Derek Foreman <derekf@osg.samsung.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
For functions that test if something is true/valid and return a 1
or 0, it makes sense to switch to bool.
Reviewed-by: Bryce Harrington <b.harrington@samsung.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
With the more accurate definition of wl_surface roles in Wayland,
enforce the restriction: a role is always set permanently, and
attempting to change it is a protocol error.
This patch is based on Jasper's patch:
http://lists.freedesktop.org/archives/wayland-devel/2014-August/016811.html
The difference in this patch compared to his are:
- send role errors on the interface whose request triggers it, not on
wl_surface
- an interface could have several requests assigning different roles,
cannot use wl_interface as the unique key; use an arbitary string
instead
- ensure in window-manager.c that create_shell_surface() ->
create_common_surface() is never called with surface->configure set,
to avoid compositor abort
- use wl_resource_post_no_memory() where appropriate instead of
hand-rolling it with wl_resource_post_error()
Ideally we would not add weston_surface::role_name field, but use
weston_surface::configure. At the moment this is not possible though,
because at least shell.c uses several different roles with the same
configure function. Drag'n'drop uses two configure functions for the
same role. The configure hook is also reset in several places,
which is not good for role tracking.
This patch overlooks the wl_surface roles assigned in privileged
extensions: screensaver, panel, background, lock, input panel.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Jasper St. Pierre <jstpierre@mecheye.net>
Add 'msc' field to weston_output to maintain the refresh counter, and
use it in presentation_feedback.presented.
Make compositor-drm update the per-output refresh counter with the
values reported by DRM. If the DRM reported value jumps backwards,
assume it wrapped around once.
Other backends do not update weston_output::msc, and there
presentation_feedback will always deliver refresh counter as zero.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
v3 Reviewed-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Implement the presentation.feedback request, and the
presentation_feedback protocol interface. Feedback information is
delivered to clients as the backend reports it, except the refresh
counter (MSC) which is always reported as zero.
Changes in v4:
* add 'flags' argument to 'presented' event without implementation
Changes in v5:
* remove the 'destroy' method implementation for feedback objects
[Pekka Paalanen: do not leak struct feedback.]
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
v3 Reviewed-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Add presentation clock setters that verify the given clock actually
works. Offer an automatic choice of a software fallback clock, when a
backend has to always use clock_gettime() to approximate the
presentation time.
The DRM backend already queried the DRM about the clock id, just let the
DRM backend set the presentation clock from that.
For all other backends which do not get a timestamp from the driver,
call the software clock setter to choose a suitable clock.
Report the chosen clock via presentation.clock_id event to clients.
In finish_frame(), upgrade the argument from uint32_t milliseconds to
struct timespec which can accurately hold the presentation clock values.
This will be needed when weston_output_finish_frame() starts to send out
presentation_feedback.presented events.
While at it, replace gettimeofday() calls with clock_gettime() using the
chosen presentation clock, so we manufacture presentation timestamps
from the presentation clock when the gfx drivers cannot give us a proper
timestamp.
Rpi patch is more verbose due to not having the compositor pointer
available in rpi_flippipe_update_complete(). Explicitly carry the clock
id with flippipe so it is available in the thread.
Changes in v4:
* rpi debug build fix
v4 Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
v3 Reviewed-by: Mario Kleiner <mario.kleiner.de@gmail.com>
You can bind to the global interface, and it delivers a fake clock id.
All requests on it raise an error.
Changes in v4:
* queuing methods were extractracted for a later series
[Louis-Francis Ratté-Boulianne: split queuing feature]
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Signed-off-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Reviewed-by: Mario Kleiner <mario.kleiner.de@gmail.com>
When backend_init returns NULL, we goto out_signals, which wants to
free(modules), but in this particular code path, modules hasn't been
initialised leading to a "Double-free or corruption" error message.
Initialising modules to NULL makes the free a no-op in this scenario.
Signed-off-by: Ondřej Majerech <majerech.o@gmail.com>
Reviewed-by: Ryo Munakata <ryomnktml@gmail.com>
weston_surface_update_transform() no longer exists, except in comments.
Fix that.
[Pekka Paalanen: don't lose the full comment in compositor-drm.c.]
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
The destructor set on the wl_output resources needs the weston_output
to be allocated, because it removes the resource from its list.
So unset the destructor on all the resources when destroying an
output.
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Pekka Paalanen
Jason Ekstrand
Derek Foreman
Ondřej Majerech
Pekka Paalanen
Bryce Harrington
Frederic Plourde
Carlos Olmedo Escobar
Giulio Camuffo