If we are asked to fill a path using any sliced textures then we need
to fallback to adding a mask of the path to the stencil buffer
and then drawing a bounding rectangle with the source textures instead.
Previously we were sharing some of the clip-stack code for adding
the path mask to the stencil buffer and being careful to check
that the current clip stack has been flushed already. We then made
sure to dirty the clip stack to be sure the path mask would be cleared
from the stencil buffer later.
This patch aims to simplify how this fallback is dealt with by just
using the public clipping API instead of relying on more fiddly tricks
to modify the stencil buffer directly without conflicting with the clip
stack.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
This adds a new experimental function, cogl_framebuffer_finish(), which
can be used to explicitly synchronize the CPU with the GPU. It's rare
that this level of explicit synchronization is desirable but for example
it can be useful during performance analysys to make sure measurements
reflect the working time of the GPU not just the time to queue commands.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
This adds a very basic test of onscreen and offscreen multisample
rendering with 4 samples per pixel. The test simply draws two triangles;
the one on the left is rendered directly to the onscreen framebuffer and
the other is first rendered offscreen before copying it onscreen.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
This adds support for multisample rendering to offscreen framebuffers.
After an offscreen framebuffer is first instantiated using
cogl_offscreen_new_to_texture() it is then possible to use
cogl_framebuffer_set_samples_per_pixel() to request multisampling before
the framebuffer is allocated. This also adds
cogl_framebuffer_resolve_samples() for explicitly resolving point
samples into pixels. Even though we currently only support the
IMG_multisampled_render_to_texture extension which doesn't require an
explicit resolve, the plan is to also support the
EXT_framebuffer_multisample extension which uses the framebuffer_blit
extension to issue an explicit resolve.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
This adds support for multisample based rendering of onscreen windows
whereby multiple point samples per pixel can be requested and if the
hardware supports that it results in reduced aliasing (especially
considering the jagged edges of polygons)
Reviewed-by: Neil Roberts <neil@linux.intel.com>
When creating new onscreen framebuffers we need to take the
configuration in cogl terms and translate that into a configuration
applicable to any given winsys, e.g. an EGLConfig or a GLXFBConfig
or a PIXELFORMATDESCRIPTOR.
Also when we first create a context we typically have to do a very
similar thing because most OpenGL winsys APIs also associate a
framebuffer config with the context and all future configs need to be
compatible with that.
This patch introduces an internal CoglFramebufferConfig to wrap up some
of the configuration parameters that are common to CoglOnscreenTemplate
and to CoglFramebuffer so we aim to re-use code when dealing with the
above two problems.
This patch also aims to rework the winsys code so it can be more
naturally extended as we start adding more configureability to how
onscreen framebuffers are created.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
This adds cogl_framebuffer_discard_buffers API that allows applications
to explicitly discard depth and stencil buffers which really helps when
using a tile based GPU architexture by potentially avoiding the need to
save the results of depth and stencil buffer changes to system memory
between frames since these can usually be handled directly with on-chip
memory instead.
The semantics for cogl_framebuffer_swap_buffers and
cogl_framebuffer_swap_region are now documented to include an implicit
discard of all buffers, including the color buffer.
We now recommend that all rendering to a CoglOffscreen framebuffer
should be followed by a call like:
cogl_framebuffer_discard_buffers (fb,
COGL_BUFFER_BIT_DEPTH|
COGL_BUFFER_BIT_STENCIL);
Reviewed-by: Neil Roberts <neil@linux.intel.com>
Most of the conformance tests read a pixel value and assert that it
matches a known value. However they were all doing this with slightly
different methods. This adds a common test_utils_check_pixel function
which they now all use. The function takes an x and y coordinate and a
32-bit value representing the color. It is assumed that writing a
known color is most convenient as an 8 digit hex sequence which this
function allows. There is also a test_utils_check_pixel_rgb function
wrapper which takes the components as separate arguments. This is more
convenient when the expected color is also calculated by the test.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
Both the GLSL and the ARBfp pipeline backends were using a variable
called last_used_for_pipeline to keep track of the last pipeline that
the shader or program state was used for. If this address is the same
as last time when the pipeline state is flushed then Cogl will only
flush the uniforms that have been modified, otherwise it will flush
all of them. The problem with this is that there was nothing to keep
that address alive so it could be destroyed and reused for a different
pipeline by the time the shader state is reused. This is quite likely
to happen in an application using legacy state because in that case
the shader state will always be used with a one-shot pipeline that
will likely be recycled in the next frame.
There is already a destroy callback to unref the shader state when the
pipeline is destroyed so this patch just makes that callback also
clear the last_used_for_pipeline pointer if it matches the pipeline
being destroyed.
https://bugzilla.gnome.org/show_bug.cgi?id=662542
Reviewed-by: Robert Bragg <robert@linux.intel.com>
This adds a description of the replacement Cogl names for the GLSL
builtins. It also updates some of the deprecation tags and corrects
some misinformation.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
Previously the EGL backend was directly prodding the width/height
members of the framebuffer structure when a configure notify event is
received. However this doesn't set the dirty flag for the viewport so
Cogl will continue using the wrong viewport y offset. The GLX backend
is already using an abstraction for updating the size which does set
the flag. This patch just makes the EGL backend also use that
abstraction.
Now that we have reasonably good Windows support, we'd like to make
sure each release still compiles there.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
Previously the crate example incremented the angle of rotation of the
cube every frame so depending on the framerate the cube might rotate
too fast to see. This just changes it to calculate the rotation based
on the elapsed time using a GTimer. The rate that frames are drawn is
unaffected.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
This namespaces all of the examples and marks them for installation
if --enable-examples-install has been passed to ./configure. This
simplifies packaging the examples which can be quite convenient
for smoke testing Cogl on various platform.
https://bugzilla.gnome.org/show_bug.cgi?id=656755
Reviewed-by: Neil Roberts <neil@linux.intel.com>
When freeing a pipeline in _cogl_pipeline_free we weren't making sure to
free the layers_cache which was leading to a memory leak.
Thanks to Sunjin Yang for finding this.
https://bugzilla.gnome.org/show_bug.cgi?id=660986
Reviewed-by: Neil Roberts <neil@linux.intel.com>
If we failed to create a native texture from pixmap via EGL or GLX then
we shouldn't call the winsys's texture_pixmap_x11_damage_notify
function. By doing the validation in cogl-texture-pixmap-x11.c the
winsys code can continue to assume that it doesn't need to verify there
is a valid tex_pixmap->winsys pointer.
Thanks to Damien Leone <dleone@nvidia.com> for catching this issue.
https://bugzilla.gnome.org/show_bug.cgi?id=660184
The previous detection was based on color depth only to determine the
texture format to use in GLX. If that worked fine at depths 24 (RGB8)
and 32 (ARGB8), that would fail at depth 30 (BGR10) and fallback to
software instead of using the TFP extension.
This commit uses an efficient population count implementation to
compare the number of 1-bits in color masks against the color depth
requested by the X client. If they are not equal this means that an
alpha channel has been requested.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
This removes some redundant initializing of the modelview matrix since
we can assume the initial state is already the identity matrix. The
explicit initialization was only really necessary when running as a
clutter test because there the default matrix isn't the identity matrix.
Also some calls to cogl_orth to change the projection matrix have been
moved into the test entry point functions instead of calling this in the
paint function. Again the previous style probably came about because
with clutter we always had to re-assert the projection but now we are in
full control of the projection and can assume it doesn't need
re-asserting once set.
Acked-by: Luca Bruno <lucabru@src.gnome.org>
Uninitialized textures could contain random bits. That makes the test
fail as glColorMask is used to let only one of the RGB pass through.
https://bugzilla.gnome.org/show_bug.cgi?id=660387
Reviewed-by: Neil Roberts <neil@linux.intel.com>
G_DISABLE_DEPRECATED is only intended for developers of Cogl and it
sometimes breaks the build for people just trying to build a
release. This patch adds an option to enable deprecated Glib
features. By default it is enabled for non-git versions of Cogl.
The patch is based on similar code in Clutter except it adds the flags
to COGL_EXTRA_CFLAGS instead of having a separate variable.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
It seems that cogl-context-private.h needs to be included before including
any of the pipeline-related stuff to avoid build errors on C89 compilers.
This is due to the recent cogl-pipeline decoupling, seems like.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
During arbfp codegen we weren't checking for NULL textures and so we
would crash when trying to query a NULL texture's GL texture target.
Since NULL texture targets result in ctx->default_gl_texture_2d_tex
being bound we can assume that a NULL texture corresponds to a
GL_TEXTURE_2D target.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
This optimizes the layer and pipeline _compare_differences functions so
neither of them use the GArray api since building up the list of
ancestors by appending to a shared GArray was showing quite high on
profiles due to how frequently pipeline comparisons are made. Instead
we now build up a transient, singly linked list by allocating GList
nodes via alloca to build up the parallel lists of ancestors.
This tweaked approach actually ends up being a bit more concise than
before, we avoid the overhead of the GArray api and now avoid making any
function calls while comparing (assuming the _get_parent() calls always
inline), we avoiding needing to get the default cogl context.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
We only need a ctx pointer if we need to refer to the default_layer_x
layers to copy as templates so only call _cogl_context_get_default()
once we need to copy a template. _cogl_context_get_default() was
starting to show up in profiles and this was the main cause.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
This splits out the core CoglPipelineLayer support code from
cogl-pipeline.c into cogl-pipeline-layer.c; it splits out the debugging
code for dumping a pipeline to a .dot file into cogl-pipeline-debug.c
and it splits the CoglPipelineNode support which is shared between
CoglPipeline and CoglPipelineLayer into cogl-node.c.
Note: cogl-pipeline-layer.c only contains the layer code directly
relating to CoglPipelineLayer objects; it does not contain any
_cogl_pipeline API relating to how CoglPipeline tracks and manipulates
layers.
This avoids using sscanf to determine the texture_unit that a
tex_coord%d_in attribute name corresponds to since that was showing high
on profiles. Instead once we know we have a "tex_coord" name then we
can simply use strtoul which is much cheaper and use the returned endptr
we get to verify we have a "_in" suffix after the number.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
We've started seeing cases where we want to allocate lots of one-shot
primitives per-frame and the cost of allocating primitives becomes
important in this case since it can start being noticeable in profiles.
The main cost for allocating primitives was the GArray allocation
and appending the attributes to the array. This updates the code to
simply over allocate the primitive storage so we can embed the list
of attributes directly in that allocation.
If the user later sets new attributes and there isn't enough embedded
space then a separate slice allocation for the new attributes is made
but still this should be far less costly than using a GArray as before.
Most of the time we would expect when setting new attributes there will
still be the same number of attributes, so the embedded space can simple
be reused.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
Calling g_strdup for attribute names was starting to show up in profiles
due to calling malloc for new string storage so frequently. This avoids
calling g_strdup and calls g_intern_string() instead. For the really
common case names we even avoid the cost of g_intern_string since we
can trivially relate our internal name_id to a static string.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
As part of the on going, incremental effort to purge the non type safe
CoglHandle type from the Cogl API this patch tackles most of the
CoglHandle uses relating to textures.
We'd postponed making this change for quite a while because we wanted to
have a clearer understanding of how we wanted to evolve the texture APIs
towards Cogl 2.0 before exposing type safety here which would be
difficult to change later since it would imply breaking APIs.
The basic idea that we are steering towards now is that CoglTexture
can be considered to be the most primitive interface we have for any
object representing a texture. The texture interface would provide
roughly these methods:
cogl_texture_get_width
cogl_texture_get_height
cogl_texture_can_repeat
cogl_texture_can_mipmap
cogl_texture_generate_mipmap;
cogl_texture_get_format
cogl_texture_set_region
cogl_texture_get_region
Besides the texture interface we will then start to expose types
corresponding to specific texture types: CoglTexture2D,
CoglTexture3D, CoglTexture2DSliced, CoglSubTexture, CoglAtlasTexture and
CoglTexturePixmapX11.
We will then also expose an interface for the high-level texture types
we have (such as CoglTexture2DSlice, CoglSubTexture and
CoglAtlasTexture) called CoglMetaTexture. CoglMetaTexture is an
additional interface that lets you iterate a virtual region of a meta
texture and get mappings of primitive textures to sub-regions of that
virtual region. Internally we already have this kind of abstraction for
dealing with sliced texture, sub-textures and atlas textures in a
consistent way, so this will just make that abstraction public. The aim
here is to clarify that there is a difference between primitive textures
(CoglTexture2D/3D) and some of the other high-level textures, and also
enable developers to implement primitives that can support meta textures
since they can only be used with the cogl_rectangle API currently.
The thing that's not so clean-cut with this are the texture constructors
we have currently; such as cogl_texture_new_from_file which no longer
make sense when CoglTexture is considered to be an interface. These
will basically just become convenient factory functions and it's just a
bit unusual that they are within the cogl_texture namespace. It's worth
noting here that all the texture type APIs will also have their own type
specific constructors so these functions will only be used for the
convenience of being able to create a texture without really wanting to
know the details of what type of texture you need. Longer term for 2.0
we may come up with replacement names for these factory functions or the
other thing we are considering is designing some asynchronous factory
functions instead since it's so often detrimental to application
performance to be blocked waiting for a texture to be uploaded to the
GPU.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
