It may happen that you e.g. fullscreen a 800x600 surface with
buffer_scale 1 (e.g. a 800x600 buffer) on an output that is
otherwise scale 2. In this case we want to temporarily set
the output scale to 1, as we're really scanning out of a
scale 1 buffer. This causes us to e.g. report the input
positions in the right place, etc.
When we restore the original mode we also restore the original
scale.
Note that the scale change is a purely compositor internal change,
to clients it still looks like the output is scale 2.
We changed the protocol to always list modes in physical pixel
units (not scaled). And we removed the scaled mode flag. This
just updates the DRM and X11 compositors and the gl and pixman renderers
to handle this.
This lets you try fullscreen in different methods, sizes, scales,
translations, etc. You can verify both output and input (via mouse over
of the rectangles).
It was erronously using output->current->height in one
place where it should use output->height. This may cause
it to create an invalid clipped coordinate in case of output
scaling or transform, because the next round "prev" would
end up NULL.
backtrace:
(gdb) bt
#0 0xb7704424 in __kernel_vsyscall ()
#1 0xb757ddde in raise (sig=5) at ../nptl/sysdeps/unix/sysv/linux/pt-raise.c:42
#2 <signal handler called>
#3 weston_config_destroy (config=0x0) at config-parser.c:508
#4 0xb75cbc0e in x11_destroy (ec=0x93506b0) at compositor-x11.c:1473
#5 0x0804e0e9 in main (argc=1, argv=0xbffe5354) at compositor.c:3337
The current config parser, parses the ini file and pulls out the values
specified by the struct config_section passed to parse_config_file() and
then throw the rest away. This means that every place we want to get
info out of the ini file, we have to parse the whole thing again. It's not
a big overhead, but it's also not a convenient API.
This patch adds a parser that parses the ini file to a data structure and
puts that in weston_compositor->config along with API to query comfig
keys from the data structure. The old parser is still available, but
we'll transition to the new approach over the next few commits.
Default output scale of 256 makes little sense. Actually this is a type
mismatch between wl_fixed and int, probably a leftover from when the
scale factor was proposed as a fixed point number.
Scale 256 probably causes the Window creation to fail, but that actually
leads to a segfault in Mesa libEGL later:
Program received signal SIGSEGV, Segmentation fault.
0 dri2_create_window_surface (drv=0x645060, disp=0x646610, conf=<optimized out>, window=<optimized out>, attrib_list=<optimized out>)
at platform_x11.c:291
291 surf->SwapInterval = 1;
Here 'surf' is NULL.
1 0x00007ffff76c0709 in eglCreateWindowSurface (dpy=0x646610, config=<optimized out>, window=58720261, attrib_list=0x0) at eglapi.c:534
2 0x0000000000421549 in gl_renderer_output_create (output=0x673ae0, window=58720261) at gl-renderer.c:1661
3 0x00007ffff41c456b in x11_compositor_create_output (c=0x6388b0, x=0, y=0, width=1024, height=640, fullscreen=0, no_input=0,
configured_name=0x0, transform=0, scale=256) at compositor-x11.c:928
4 0x00007ffff41c5ca0 in x11_compositor_create (display=0x631950, fullscreen=0, no_input=0, use_pixman=0, argc=0x7fffffffda7c,
argv=0x7fffffffdd18, config_fd=14) at compositor-x11.c:1596
5 0x00007ffff41c63db in backend_init (display=0x631950, argc=0x7fffffffda7c, argv=0x7fffffffdd18, config_fd=14) at compositor-x11.c:1746
6 0x000000000040fcb7 in main (argc=1, argv=0x7fffffffdd18) at compositor.c:3293
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Cc: Alexander Larsson <alexl@redhat.com>
This patch does miscellanous improvements with raw surfaces:
* some frames markers are sent to identify a single frame made of
multiple surface updates
* we send the dirty sub-rectangles instead of the full bouncing box
* the size of the fragmentation buffer is now honored, so that our big
surface updates don't look like a DoS
* the subtile and image flipping are done in one step (not requiring a
temporary tile)
* we don't care about the size of the dirty region and always use the
preferred codec for surface update
Last FreeRDP don't send Synchronize packets anymore, so send the
first screen refresh when we're "connected". The client cursor is
also disabled during this step.
This patch fixes NSC codec initialisation that was not done (an
erronous copy'n paste).
The remoteFx context must be resetted when we go through an activation
sequence
The stream utils of FreeRDP have changed recently, this patch fixes
the compositor compilation against FreeRdp master.
The backend_init() prototype has changed too, this fixes it.
Whether or not a shm pool is used for resizing is now configurable at
build time (--disable-resize-optimization).
[pq: removed an unnecessary hunk from the patch]
Scale-crop mode scales the wallpaper to tightly fill the whole output,
but preserving wallpaper aspect ratio. If aspect ratio differs from the
output's, the wallpaper is centered cutting it from top/bottom or
left/right.
Add this to the weston.ini man page, and explain all three modes.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
On Raspberry Pi, weston-desktop-shell is so slow to start, that the
compositor has time to run the fade-in before the wallpaper is up. The
user launching Weston sees the screen flipping to black, the fbcon
fading in, and then the desktop popping up.
To fix this, wait for the weston-desktop-shell to draw
everything before starting the initial fade-in. A new request is
added to the private desktop-shell protocol to signal it. If a
desktop-shell client does not support the new request, the fade-in
happens already at bind time.
If weston-desktop-shell crashes, or does not send the 'desktop_ready'
request in 15 seconds, the compositor will fade in anyway. This should
avoid a blocked screen in case weston-desktop-shell malfunction.
shell_fade_startup() does not directly start the fade-in but schedules
an idle callback, so that the compositor can process all pending events
before starting the fade clock. Otherwise (on RPi) we risk skipping part
of the animation. Yes, it is a hack, that should have been done in
window.c and weston-desktop-shell instead.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
There is no need to support weston_plane anymore.
The max-planes option is removed as unused.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Replace the GL renderer with the new rpi-renderer on the Raspberry Pi
backend. This makes Weston on rpi not use EGL or GL anymore, at all.
The weston_plane feature is disabled, since the rpi-renderer does the
same, but better.
Add a command line option to select the output transform. It is not a
weston.ini option for now, since the rpi backend does not read the
configuration file yet. Hopefully that will be done later with some
shared code.
Add the rpi options to 'weston --help' output.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Dispmanx is the prorietary display API on the Raspberry Pi, which
provides hardware compositing. Every visible surface is assigned a
Dispmanx element, and the hardware or firmware will do all compositing
onto screen. The API supports translation, scaling, flips, discrete
rotations in 90-degree steps, alpha channel on the surfaces, and
full-surface alpha on top.
Previously, Dispmanx capabilities were used via the weston_plane
mechanism, where surfaces were assigned to planes when possible, and
otherwise transparently falling back to GLESv2 compositing. Because we
have no way to use the same memory buffer as a GL texture and a Dispmanx
resource, we had to prepare for both. In the worst case, that means one GL
texture, and two (double-buffered case) Dispmanx resources, all the size
of a whole surface, for all surfaces. This was eating memory fast. To
make things worse (and less slow), the wl_shm buffer was kept around,
since it was copied to either a texture or a resource as needed. This
caused all clients to need two buffers. In a Dispmanx-only renderer, we
can drop the GL texture, and we can release wl_shm buffer immediately
after the first copy, so clients become effectively single-buffered. So
from the worst case of 5 buffers per surface, we go down to 3 or just
2 (single-buffered Dispmanx element, one wl_shm buffer in the client)
buffers per surface.
As this will replace the GL renderer on rpi, we cannot fall back to the
GLESv2 compositing anymore. We lose arbitrary surface rotation, but we
lose also the GL fallback, which caused glitches.
This patch depends on new RaspberryPi firmware. Older firmware may not
render ARGB surfaces correctly, solid color surfaces maybe cause a
performance hit, and the output may completely fail in case the firmware
does not fall back internal off-line compositing properly as needed.
This new rpi-renderer support surface translation and scaling, but not
rotation or transpose (not even in 90-deg steps). In theory, 90-deg step
surface rotation is possible to support. Output transformations are
supported, but flipped variants do not seem to work right.
As a detail, menus and other surfaces that are simply translated with
respect to another surface caused falling back to the GL renderer. The
rpi-renderer handles them directly.
This patch only adds the new renderer, but does not hook it up into use.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Both GL and pixman renderer (pixman probably only because GL did?)
return the screen capture image as y-flipped, therefore Weston y-flips
it again. However, the future rpi-renderer can produce only right-way-up
(non-flipped) screen captures, and does not need an y-flip.
Add a capability flag for y-flip, which the rpi-renderer will not set,
to get screen captures the right way up.
The wcap recording code needs yet another temporary buffer for the
non-flipped case, since the WCAP format is flipped, and the code
normally overwrites the input image as it compresses it. This becomes
difficult, if the compressor is supposed to flip while processing.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
The upcoming rpi-renderer cannot handle arbitrary rotations. Introduce
Weston capability bits, and add a bit for arbitrary rotation. GL and
Pixman renderers support it.
Shell or any other module must not produce surface transformations with
rotation, if the capability bit is not set. Do not register the surface
rotation binding in desktop shell, if arbitary rotation is not
supported.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
The array of arguments supplied to execv must be NULL terminated. If
unitialized values are used as pointers the exec call may fail with a
EFAULT error ("Bad address").
https://bugs.freedesktop.org/show_bug.cgi?id=64874
If you specify e.g. scale=2 in an output section in weston.ini
we scale all modes by that factor.
We also correctly scale cursor positioning, but ATM there is no
scaling of the cursor sprite itself.
If you specify e.g. scale=2 in weston.ini an output section for the
X11 backend we automatically upscale all normal surfaces by this
amount. Additionally we respect a buffer_scale set on the buffer to
mean that the buffer is already in a scaled form.
This works with both the gl and the pixman renderer. The non-X
backends compile and work, but don't support changing the output
scale (they do downscale as needed due to buffer_scale though).
This also sends the new "scale" and "done" events on wl_output,
making clients aware of the scale.
Rather than storing the shadow_image in the untransformed space
and rotating on copy to hw_buffer we store both on the transformed
space. This means a copy between them is a straight copy, and that
apps supplying correctly transformed surface buffers need not
change them.
We also correctly handle all output transform including the previously
unhandled flipped ones, as well as client supplied buffer_transforms (which
were previously ignored).
We also simplify the actual rendering by just converting any damage
region to output coordinates and set it on a clip and composite
the whole buffer, letting pixman do the rectangle handling. This
means we always do all the transforms, including the surface positioning
as a pixman_image transform. This simplifies the code and sets us up
for handling scaling at a later stage.
The transform looks complicated, but in practice it ends up being
an integer translation almost always, so it will hit the pixman
fastpaths.
