CoglOffscreen is effectively a CoglFramebuffer, but it isn't being marked as
such by the GType machinery. This makes it impossible for introspection to
correctly set this class up.
Fix that by adding a COGL_GTYPE_IMPLEMENT_INTERFACE() code into the declaration
of CoglOffscreen. This does not have any functional changes though.
Meson uses the 'dependencies' field to determine and
parallelize build steps, but that isn't entirely true
with 'link_with'; this might cause a race condition
when generating header files while trying to build
them.
Fix that by only using 'dependencies' instead of 'link_with'.
This commit adds meson build support to mutter. It takes a step away
from the three separate code bases with three different autotools setups
into a single meson build system. There are still places that can be
unified better, for example by removing various "config.h" style files
from cogl and clutter, centralizing debug C flags and other configurable
macros, and similar artifacts that are there only because they were once
separate code bases.
There are some differences between the autotools setup and the new
meson. Here are a few:
The meson setup doesn't generate wrapper scripts for various cogl and
clutter test cases. What these tests did was more or less generate a
tiny script that called an executable with a test name as the argument.
To run particular tests, just run the test executable with the name of
the test as the argument.
The meson setup doesn't install test files anymore. The autotools test
suite was designed towards working with installed tests, but it didn't
really still, and now with meson, it doesn't install anything at all,
but instead makes sure that everything runs with the uninstalled input
files, binaries and libraries when running the test suite. Installable
tests may come later.
Tests from cogl, clutter and mutter are run on 'meson test'. In
autotools, only cogl and clutter tests were run on 'make check'.
Install include files in
$prefix/include/mutter-$apiversion/[clutter,cogl,...,meta]/, and
datafiles in /usr/share/mutter-$apiversion/.... We still would conflict
e.g. given that our gettext name is "mutter", and how keybindings are
installed, but it's a step in the right direction.
There are different unit-tests file generated containing lists of tests
the test-runner.sh should run. Running run-tests.sh read the unit-tests
in the current directory, which is inconvenient to do when using meson.
The docs previously suggested that `cogl_frame_info_get_frame_counter`
returned a timestamp of an unknown clock ID. That's not correct. The
cogl source code shows that it does and must use the same clock as
`cogl_get_clock_time`.
Related to https://gitlab.gnome.org/GNOME/mutter/issues/131
Before we just set it to "none", but this was not enough since various
calls will depend on not just the context being active, but the main
rendering surface.
Fixes https://gitlab.gnome.org/GNOME/mutter/issues/21
By the looks of it, commit 95e9fa10ef was taping over an Intel DRI bug
that would make it return post-swizzling pixel data on glReadPixels().
There's been reports over time of that commit resulting in wrong colors
on other drivers, and lately Mesa >17.3 started showing the same symptoms
on Intel.
But texture swizzling works by changing parameters before fragment shaders
and reading pixels from an already drawn FBO/texture doesn't involve those.
This should thus use pixel_format_to_gl_with_target(), which will result in
correctly requesting the same pixel format than the underlying texture,
while still considering it BGRA for the upper layers in the swizzling case.
https://gitlab.gnome.org/GNOME/mutter/issues/72Closes: #72
We just arbitrarily chose the first EGL config matching the passed
attributes, but we then assumed we always got GBM_FORMAT_XRGB8888. That
was not a correct assumption. Instead, make sure we always pick the
format we expect.
Closes: https://gitlab.gnome.org/GNOME/mutter/issues/2
On drivers that do not support glGetTexImage2D (i.e. on GLES),
cogl_texture_get_data() has a "feature" that allows it to download
texture data by rendering the texture on an intermediate framebuffer
object and then reading back the data from there. However, this
feature requires the user to have previously set an "active"
framebuffer object in the context, which makes this very tricky
because it is not clear to the developer that he needs to do that
in order for some code to work on GLES (of course it works on
desktop GL, so nobody notices...) and additionally the code actually
crashes if an active fbo is not set!
This patch basically removes this feature in order to prevent
the crash and is in line with how this code has evolved in cogl-2.0:
https://git.gnome.org/browse/cogl/commit/?id=6d6a277b8e9a63a8268046e5258877ba94a1da5bhttps://bugzilla.gnome.org/show_bug.cgi?id=789961
When creating a renderer with a custom winsys (which is always how
mutter uses cogl) make it possible to pass a user data with the winsys.
Still unused.
https://bugzilla.gnome.org/show_bug.cgi?id=785381
The GL_BGRA definition is not available for GLES2 contexts, which use
the EXT_texture_format_BGRA8888 instead, causing a build failure when
trying to use it in those contexts.
Fortunately, this hack is only relevant for GL, so let's guard it to
prevent the failure in GLES2, where that extension is used instead.
https://bugzilla.gnome.org/show_bug.cgi?id=786568
Those are cached and reused across runs, which doesn't qualify to mesa
as "static" indeed. Properly marking those as dynamic is more true, and
brings in slight performance benefits just by avoiding the resulting
(and later silenced) mesa warning.
https://bugzilla.gnome.org/show_bug.cgi?id=782344
Fixes cogl_texture_get_data() resorting to the wrong conversions when
extracting the texture data. This notably resulted in RGB/RGBA buffers
copied as-is into BGRA buffers, for instance for the fullscreen animation,
or single-window screenshots of such buffers.
https://bugzilla.gnome.org/show_bug.cgi?id=779234
The ARB_robustness extension defined the following tokens as
returned by GetGraphicsResetStatusARB (see spec at [1]):
GUILTY_CONTEXT_RESET_ARB 0x8253
INNOCENT_CONTEXT_RESET_ARB 0x8254
UNKNOWN_CONTEXT_RESET_ARB 0x8255
These tokens might not be defined in some GL implementations,
such as Mesa 13's implementation of GLES 2.0, so we need to
define them ourselves not to break those builds.
[1] https://www.khronos.org/registry/OpenGL/extensions/ARB/ARB_robustness.txthttps://bugzilla.gnome.org/show_bug.cgi?id=781398
We already do have a texture with an internal format in these paths,
so we should check the required format according to it.
This fixes CoglAtlasTexture (and CoglPangoRenderer indirectly), as
it forces a RGBA format on its texture, but pixel_format_to_gl()
anyway assumed swizzling is performed on the texture, while it is
not the case.
https://bugzilla.gnome.org/show_bug.cgi?id=779234
This is used by the GL driver in order to determine whether swizzling
actually applies given the bitmap and target texture internal format.
If both agree that they store BGRA, then swizzling may apply.
https://bugzilla.gnome.org/show_bug.cgi?id=779234
If the GL implementation/hw supports the GL_*_texture_swizzle extension,
pretend that BGRA textures shall contain RGBA data, and let the flipping
happen when the texture will be used in the rendering pipeline.
This avoids rather expensive format conversions when forcing BGRA buffers
into RGBA textures, which happens rather often with WL_SHM_FORMAT_ARGB8888
buffers (like gtk+ uses) in little-endian machines.
In intel/mesa/wayland, the performance improvement is rather noticeable,
CPU% as seen by top decreases from 45-50% to 25-30% when running
gtk+/tests/scrolling-performance with a cairo renderer.
https://bugzilla.gnome.org/show_bug.cgi?id=779234
Because the threaded-swap-wait functionality requires XInitThreads(),
and because it isn't clear that it is a win for all applications,
add a API function to conditionally enable it.
Fix the cogl-crate example not to just have a hard-coded dependency
on libX11.
https://bugzilla.gnome.org/show_bug.cgi?id=779039
It's a good guess that the buffer swap will occur at the next vblank,
so use glXWaitVideoSync in a separate thread to deliver a sync event
rather than just letting the client block when frame drawing, which
can signficantly change app logic as compared to the INTEL_swap_event
case.
https://bugzilla.gnome.org/show_bug.cgi?id=779039
When we don't have GLX_OML_sync_control, we still can set the
frame presentation time, but we always use the system monotonic time,
so return that from get_clock_time().
https://bugzilla.gnome.org/show_bug.cgi?id=779039
As previously commented in the code, SGI_video_sync is per-display, rather
than per-renderer. The is_direct flag for the renderer was tested before
it was initialized (per-display) and that resulted in SGI_video_sync
never being used.
https://bugzilla.gnome.org/show_bug.cgi?id=779039
In order to minimize the amount of breakage, while at the same time
making it easier to make backward incompatible changes needed to
continue turning libmutter into a capable Wayland compositor, make the
libmutter and friends (libmutter-clutter, libmutter-cogl*) parallel
installable by adding a version number to the name. This changes
various filenames, for example what previously was libmutter.so is now
libmutter-0.so (assuming the version for now is 0), and
libmutter-clutter-1.0.so is now libmutter-clutter-0.so. The pkg-config
filenames and GObject introspection has been renamed to reflect this as
well.
This enables a downstream compositor rely on a specific version of the
libmutter API, while gracefully handling API/ABI changes by having to
update to the new version at their own pace.
https://bugzilla.gnome.org/show_bug.cgi?id=777317
Different libEGL will do different things for eglGetDisplay since it has
to guess what kind of display it's been handed. Better to just use the
API that makes it explicit.
Signed-off-by: Adam Jackson <ajax@redhat.com>
https://bugzilla.gnome.org/show_bug.cgi?id=772422
We need a GLES2 extension macro in cogl-texture-2d-gl.c, but we can't
include GLES2/gl2ext.h because it will conflict with things in
GL/glext.h. We can't rely on cogl including anything for us since it'd
only include GLES2/gl2ext.h if OpenGL support was explicitly disabled.
https://bugzilla.gnome.org/show_bug.cgi?id=774891
EGLDevice requires a define from GLES2, even when GL is used instead.
As type definitions may conflict between the two, we shouldn't include
both at the same time. Instead, provide the missing define explicitly
when not using GLES2.
https://bugzilla.gnome.org/show_bug.cgi?id=774891
Add API to enable the caller to have a custom method for allocating an
external texture. This will enable the possibility for mutter to
generate a texture from for example an EGLStream without having to add
support for that in Cogl.
https://bugzilla.gnome.org/show_bug.cgi?id=773629
Don't rely on the Cogl layer having Wayland specific paths by
determining the buffer type and creating the EGLImage ourself, while
using the newly exposed CoglTexture from EGLImage API. This changes the
API used by MetaWaylandSurface to make the MetaWaylandBuffer API be
aware when the buffer is being attached. For SHM and EGL buffers, only
the first time it is attached will result in a new texture being
allocated, but later for EGLStream's, more logic on every attach is
needed.
https://bugzilla.gnome.org/show_bug.cgi?id=773629
The WL_bind_wayland_display spec says that EGL images should be created
using EGL_WAYLAND_BUFFER_WL as the target and a NULL context. Mesa
seems to be lenient and accept any context, however some other stacks
aren't so forgiving and fail if anything apart from EGL_NO_CONTEXT is
used.
Signed-off-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
https://bugzilla.gnome.org/show_bug.cgi?id=769731
Mutter (and libmutter users) are the only users of this version of
cogl, and will more or less only use the cogl-1.0, cogl-2.0 and cogl
experimental API variants, and having the possibility of having
different API versions of the same API depending on what file includes
it is error prone and confusing. Lets just remove the possibility of
having different versions of the same API.
https://bugzilla.gnome.org/show_bug.cgi?id=768977
We bypass our build configuration to fetch API from a version which
isn't the one we actually use. Stop bypassing and just admit that the
1.0 API is still there, but still deprecated.
https://bugzilla.gnome.org/show_bug.cgi?id=768977
CoglFrameInfo is a frame info container associated with a single
onscreen framebuffer. The clutter stage will eventually support drawing
a stage frame with multiple onscreen framebuffers, thus needs its own
frame info container.
This patch introduces a new stage signal 'presented' and a accompaning
ClutterFrameInfo and adapts the stage windows and past onscreen frame
callbacks users to use the signal and new info container.
https://bugzilla.gnome.org/show_bug.cgi?id=768976
Move the KMS interaction from cogl into mutter, where most of the other
KMS interaction already takes place. This also removes dead code which
were only excercised when non-mutter callers used the cogl KMS backend.
The cogl KMS API was updated to pass via MetaRendererNative instead of
via the different cogl objects.
https://bugzilla.gnome.org/show_bug.cgi?id=768976
In cogl use cogl-config.h and in clutter use clutter-build-config.h. We
can't use clutter-config.h in clutter because its already used and
installed.
https://bugzilla.gnome.org/show_bug.cgi?id=768976
If you include a file that might define __INSIDE_COGL_H__, don't
undefine it if it wasn't defined in that file. This makes it possible
to include for example cogl-gles2.h from some other file which defines
__INSIDE_COGL_H__.
https://bugzilla.gnome.org/show_bug.cgi?id=768976
When using a context with robustness, glGetError() may return
GL_CONTEXT_LOST at any time and this error doesn't get cleared until
the application calls glGetGraphicsResetStatus() . This means that our
error checking can't call glGetError() in a loop without checking for
that return value and returning in that case.
https://bugzilla.gnome.org/show_bug.cgi?id=739178
commit 188752158 changed cogl to stop needlessly creating its own
monitor output configuration when mutter would just soon overwrite
it anyway.
Unfortunately, that commit is causing a crash in some cases because
cogl will now create and later draw to a 0x0 egl surface until mutter
sets the monitor layout.
This commit changes cogl to avoid creating and using a surface, before
it knows how big of a surface to create.
https://bugzilla.gnome.org/show_bug.cgi?id=758073
If we get EACCES from drmPageFlip we're not going to get
a flip event and shouldn't wait for one.
This commit changes the EACCES path to silently ignore the
failed flip request and just clean up the fb.
https://bugzilla.gnome.org/show_bug.cgi?id=756926
ES3 provides glMapBufferRange as core, with the added bonus that it also
supports read mappings. Use this where possible.
Signed-off-by: Daniel Stone <daniels@collabora.com>
https://bugzilla.gnome.org/show_bug.cgi?id=728355
GLES2 does not support passing READ to glMapBuffer; attempting to call
cogl_buffer_map for read on ES2 will bring it down with an assert. Make
sure COGL_DEBUG=journal doesn't do this when it's not possible.
Signed-off-by: Daniel Stone <daniels@collabora.com>
https://bugzilla.gnome.org/show_bug.cgi?id=728355
On Wayland deinit() of an onscreen buffer is going to destroy the
associated native window of an EGLSurface. So we should destroy the
EGLSurface as well otherwise we might end up confusing the GL driver.
We also currently guard against setting a EGL_NO_SURFACE as current
EGLSurface, but this shouldn't be a problem if we have a surfaceless
context. So we allow surface destruction under that condition.
https://bugzilla.gnome.org/show_bug.cgi?id=754667
If the user switches VTs in the middle of a page flip, the
page flip operation may fail with EACCES. page flipping will
work next time the VT becomes active, so we shouldn't disable
page flipping in that case.
https://bugzilla.gnome.org/show_bug.cgi?id=754540
If cogl fails to open the drm device, initialize gbm, or open the
egl display, then it closes the drm fd, uninitializes gbm, closes the
display and then calls _cogl_winsys_renderer_disconnect which does
most of those things again, on the, now deinitialized, members.
This commit removes the explicit failure handling in renderer_connect and
defers cleanup to disconnect.
https://bugzilla.gnome.org/show_bug.cgi?id=754540
gbm confusingly has two different format types, and cogl
is using the wrong one in some of its calls to gbm_surface_create
This commit fixes the calls that are wrong.
https://bugzilla.gnome.org/show_bug.cgi?id=754540
glGetIntegerv (GL_DEPTH_BITS, ...) and friends are deprecated in GL3; we
have to use glGetFramebufferAttachmentParameteriv() instead, like we do
for offscreen framebuffers.
Based on a patch by: Adel Gadllah <adel.gadllah@gmail.com>
Signed-off-by: Emmanuele Bassi <ebassi@gnome.org>
https://bugzilla.gnome.org/show_bug.cgi?id=753295
We want to be able to retrieve the XVisualInfo used when creating the
GL context under GLX and EGL-X11, so that we can use the visual before
we have an onscreen frame buffer.
Initialize variables; GCC does not always catch all cases where the
variables are used after being initialized, especially when it comes to
out parameters.
Some drivers ( like mgag200 ) don't yet support drmModePageFlip.
This commit forgoes waiting for vblank and flips right away
in those cases. That prevents the hardware from freezing up the screen,
but does mean there will be some visible tearing.
https://bugzilla.gnome.org/show_bug.cgi?id=746042
We can't just destroy and replace the EGL and gbm surfaces while
they are still in use i.e. while there is a pending flip. In fact, in
that case, we were calling gbm_surface_destroy() on a surface that
still had the front buffer locked and then, on the flip handler,
gbm_surface_release_buffer() for a buffer that didn't belong to the
new surface.
Instead, we still allocate new surfaces when requested but they only
replace the old ones on the next swap buffers when we're sure that the
previous flip has been handled and buffers properly released.
Currently the code queries the current msc then tries to approximate the
value of the next msc satisfing the modulus 2 for when to wait. This
introduces some instability as the msc may tick over during the
roundtrip leading to a 32ms wait instead of a 16ms wait. This happens
often enough to cause jerky animations, and affect gnome-shell-perf-tool.
A simpler solution is just use a single roundtrip by using WaitForMsc to
ask the driver to compute the next vblank itself.
Cc: Owen W. Taylor <otaylor@fishsoup.net>
Cc: Robert Bragg <robert@linux.intel.com>
Reviewed-by: Robert Bragg <robert@sixbynine.org>
