Namely the single-planar YUYV, the two-planar NV12, and the
three-planar YUV420, using the shaders already present in Weston.
Signed-off-by: Emmanuel Gil Peyrot <emmanuel.peyrot@collabora.com>
Reviewed-by: Derek Foreman <derekf@osg.samsung.com>
Maniphest Tasks: T13
Differential Revision: https://phabricator.freedesktop.org/D334
This struct serves as renderer data for linux-dmabuf buffers, and can
contain multiple struct egl_image, simplifying this latter in the
common non-dmabuf case.
Signed-off-by: Emmanuel Gil Peyrot <emmanuel.peyrot@collabora.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Derek Foreman <derekf@osg.samsung.com>
Differential Revision: https://phabricator.freedesktop.org/D333
This allows renderers to use that struct to create their own dmabufs,
in case they can’t import the one provided by the client directly but
know how to convert it into a format they can render.
Signed-off-by: Emmanuel Gil Peyrot <emmanuel.peyrot@collabora.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Differential Revision: https://phabricator.freedesktop.org/D332
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
Reviewed-by: Derek Foreman <derekf@osg.samsung.com>
Clean up display connection via eglTerminate() in case of EGL
initialisation error.
Signed-off-by: Dawid Gajownik <gajownik@gmail.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Import dmabuf as an EGLImage, and hold on to the EGLImage until we are
signalled a content change. On content change, destroy the EGLImage and
re-import to trigger GPU cache flushes.
We hold on to the EGLImage as long as possible just in case the client
does other imports that might later make re-importing fail.
As dmabuf protocol uses drm_fourcc codes, we need libdrm for
drm_fourcc.h. However, we are not doing any libdrm function calls, so
there is no new need to link to libdrm.
RFCv1 changes:
- fix error if dmabuf exposed unsupported
- always use GL_TEXTURE_EXTERNAL_OES with dmabuf
v2 changes:
- improve support check and error handling
- hold on to the imported EGLImage to avoid the dmabuf becoming
unimportable in the future
- send internal errors with linux_dmabuf_buffer_send_server_error()
- import EGL_EXT_image_dma_buf_import extension headers
- use heuristics to decide between GL_TEXTURE_2D and
GL_TEXTURE_EXTERNAL_OES
- add comment about Mesa requirements
- change y-invert from per-plane boolean to per-buffer flag
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Signed-off-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
This is a reference-counted holder of an EGLImage. For now, direct
EGLImage usage is simply converted to use egl_image. Use of reference
counting will come in a later patch.
v2:
- this is a new patch, split from gl-renderer dmabuf import support
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Normally we need to check if a seat's [device_type]_count is > 0 before
we can use the associated pointer. However, in a binding you're
guaranteed that the seat has a device of that type. If we pass in
that type instead of the seat, it's obvious we don't have to test it.
The bindings can still get the seat pointer via whatever->seat if they
need it.
This is preparation for a follow up patch that prevents direct access
to seat->device_type pointers, and this will save us a few tests at
that point.
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
Signed-off-by: Derek Foreman <derekf@osg.samsung.com>
EGLGetDisplay() doesn't generate a GL error, so we shouldn't print one.
I've also renamed the goto labels so it's a little clearer when to use them.
Signed-off-by: Derek Foreman <derekf@osg.samsung.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
To help reduce code duplication and also 'kitchen-sink' includes
the ARRAY_LENGTH macro was moved to a stand-alone file and
referenced from the sources consuming it. Other macros will be
added in subsequent passes.
Signed-off-by: Jon A. Cruz <jonc@osg.samsung.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
If the GL implementation doesn't provide an XRGB visual we may still be
able to proceed with an ARGB one. Since we're not changing the scanout
buffer format, and our current rendering loop always results in saturated
alpha in the frame buffer, it should be Just Fine(tm) - and probably better
than just exiting.
This is a workaround for https://bugs.freedesktop.org/show_bug.cgi?id=89689
Reviewed-By: Bryce Harrington <bryce@osg.samsung.com>
Tested-By: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-By: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Signed-off-by: Derek Foreman <derekf@osg.samsung.com>
Currently we pass either a single format or no formats to the gl renderer
create and output_create functions. We extend this to any number of
formats so we can allow fallback formats if we don't get our first pick.
Reviewed-By: Bryce Harrington <bryce@osg.samsung.com>
Reviewed-By: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Signed-off-by: Derek Foreman <derekf@osg.samsung.com>
EGL_MESA_configless_context is a display extension. The query for client
extensions was overwriting the pointer, so it was being searched from
the client extensions instead.
Fix any confusion here by moving all client extension checks into
another function. Drop a useless cast.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Daniel Stone <daniels@collabora.com>
An EGL implementation may support client extensions without supporting
EGL_EXT_platform_base. In such a case, we should return 0 to fall back
to the old eglGetDisplay() way.
Cc: Manuel Bachmann <manuel.bachmann@open.eurogiciel.org>
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Some DRI drivers, including VMware vmwgfx, do not support
calling eglQueryString() with a EGL_NO_DISPLAY parameter.
Just as we do in gl_renderer_supports(), which returns 0
but does not fail in this case, do not fail in
gl_renderer_setup_egl_extensions().
Signed-off-by: Manuel Bachmann <manuel.bachmann@open.eurogiciel.org>
[Pekka: split the patch]
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Reviewed-by: Daniel Stone <daniels@collabora.com>
This means compositors don't need to call supports() manually and
create() will just return -1 in the failure case as before. This also
means we can deal with the case of eglGetProcAddress returning
non-NULL but not actually being available at runtime.
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
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>
Taking the easy way, always do a rendering pass when copying any real
buffer or texture. Will handle YUV formats, and makes it easy to always
return data the right y-direction up.
All the FBO GL state is created and torn down on every invocation, so this
is a pretty naive implementation.
If there was a wl_shm buffer giving the content to the surface, and the
stride of the buffer was greater than width * bytes_per_pixel, then this
implementation will return stride long rows, not width.
Changes in v2:
- simplify pack_color()
- remove stride and format 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>
Add a new buffer type identifying the solid color contents which do not
have a real buffer.
Solid color surfaces now pretend to have 1x1 pixel content data.
This helps the future surface_get_data_size() implementation.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
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>
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>
Pixman uses y-x banded rectangles to represent regions. We use these
y-x banded rectangles to generate triangle fans, resulting in more
geometry than strictly necessary to draw the screen.
This patch combines the bands to reduce geometry for complex scenes.
Acked-by: "Jasper St. Pierre" <jstpierre@mecheye.net>
Tested-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
output_rotate_damage shifted an array of pixman regions with a loop. Now
it uses an index into that array.
This currently only saves 1 pixman_region32_copy, but we can now raise
BUFFER_DAMAGE_COUNT without a performance impact if we run into a
configuration where this is useful.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
this adds a mechanism to mask the views belonging to a layer
to an arbitrary rect, in the global space. The parts that don't fit
in that rect will be clipped away.
Supported by the gl and pixman renderer only for now.
If the client attaches a new SHM buffer with a different format from a
previous one then we ought to trigger a full upload so that GL can
allocate a new texture. Otherwise Weston would technically be doing
invalid operations because it would call glTexSubImage2D with a
different format from the one specified in glTexImage2D. Presumably it
would also mean GL would have to convert the buffer as it copies the
sub-region in which isn't ideal.
This patch makes it decide the GL format when the buffer is attached
instead of when processing the damage and it will set
needs_full_upload if it is different from what it used before.
Previously when uploading SHM data we would initialise the texture
with glTexImage2D and NULL data when the buffer is attached. Then if
the GL_EXT_unpack_subimage extension is available we would always use
glTexSubImage2D to upload the data. The problem with that is that the
first glTexImage2D was always setting the internal format to
GL_BGRA_EXT and then if a 16-bit texture is used we would later call
glTexSubImage2D with a data format of GL_RGBA. Under GLES2 the
internal format must always match the data format so this is
technically invalid.
This patch makes it so that it always calls glTexImage2D when flushing
the damage for the first time. That way it will use the right internal
format and we don't need to call glTexImage2D with NULL data.
https://bugs.freedesktop.org/show_bug.cgi?id=75251
Queueing in the Presentation extension requires splitting the viewport
state into buffer state and surface state. To conveniently allow
assigning only one, the other, or both, reorganize the
weston_buffer_viewport structure.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
The gbm-format configuration option can now be specified per-output as
well as in the core config section. If it is not specified it will
default to the format specified in the core section. The
EGL_MESA_configless_context extension is required for this to work. If
this extension is available it will create a context without an
EGLConfig and then it will potentially use a different EGLConfig for
each output.
The gl-renderer interface has been changed so that it takes the EGL
attributes and visual ID in the create_output function as well as in
the create function.
Part of the gl_renderer_setup function only deals with checking EGL
extensions and doesn't need to have a current context. This patch
moves these checks so that they are done during gl_renderer_create
instead of waiting until we have an output. We will need this in a
later patch because some of the EGL extensions will affect how we
create the EGLSurface.
eglCreateContext fails with every EGLConfig that
nvidia blob 334.16 provides causing NULL pointer
dereference in gl_renderer_destroy when destroying
fragment and fan bindings.
https://bugs.freedesktop.org/show_bug.cgi?id=74699
Signed-off-by: Mariusz Ceier <mceier+wayland@gmail.com>
This has a couple of additional implications for the internal weston API:
1) weston_view_configure no longer exists. Use weston_view_set_position
instead.
2) The weston_surface.configure callback no longer takes a width and
height. If you need these, surface.width/height are set before
configure is called. If you need to know when the width/height
changes, you must track that yourself.
Gather the variables affecting the coordinate transformations between
buffer and local coordinates into a new struct weston_buffer_viewport.
This will be more useful later, when the crop & scale extension is
implemented.
When a renderer switch happens, it is possible that when the surface
state is created, a buffer for the given surface is already available.
In that case, run the attach routine so that the pixel contents are
properly set. Otherwise, it would only be set when a new attach request
is made for that surface.
Also, change the drm backend so that it keeps the buffer reference in
the weston_surface when running with the pixman renderer. The pixman
renderer keeps a reference to it anyway, so it is never released
early.
This makes the renderer transition seamless, without leaving a black
screen as before.
When running with the pixman renderer, the debug binding 'W'
(mod-shift-space W) will cause the compositor to load gl-renderer.so
and start using it instead of the pixman renderer.
This wraps all accesses to an SHM buffer between wl_shm_buffer_begin
and end so that wayland-shm can install a handler for SIGBUS and catch
attempts to pass the compositor a buffer that is too small.
The only user for this was the wayland backend with the GL renderer. It is
not needed in the Pixman renderer because you can easily create subimages.
All of the fancy output matrix calculations can be replaced by a single
glViewport call. Also, it didn't work with outputs located anywhere but
(0, 0) and I'm pretty sure output transformed outputs would break it too.
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
