The recommended usage model for rendering pipelines with minor changes
is to make a copy of a base pipeline just before rendering and then
modify that. The new pipeline can then be used as the base pipeline
for the next paint. Currently this has a known problem when modifying
uniform values in that Cogl won't prune the redundant ancestry and
instead it will end up with an ever-growing chain of pipelines. This
is particularly bad for something like CoglGST where it could also end
up leaking textures for the video frames if the pipelines are used to
render video.
The patch adds a test case for that situation so that we won't forget
about the problem. The test is maked as a known failure. Additionally
the patch adds a similar test for setting the blend constant to
constrast the test with some state that does work correctly.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
(cherry picked from commit 2550181543389d6e9e1cb9618d17cd352a0cf9b6)
Previously on GLES2 where there is no builtin point size uniform then
we would always add a line to the vertex shader to write to the
builtin point size output because when generating the shader it is not
possible to determine if the pipeline will be used to draw points or
not. This patch changes it so that the default point size is 0.0f
which is documented to have undefined results when drawing points.
That way we can avoid adding the point size code to the shader in that
case. The assumption is that any application that is drawing points
will probably have explicitly set the point size on the pipeline
anyway so it is not a big deal to change the default size from 1.0f.
This adds a new pipeline state flag to track whether the point size is
non-zero. This needs to be its own state because altering it needs to
cause a different shader to be added to the pipeline cache. The state
flags that affect the vertex shader have been changed from a constant
to a runtime function because they will be different depending on
whether there is a builtin point size uniform.
There is also a unit test to ensure that changing the point size does
or doesn't generate a new shader depending on the values.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
(cherry picked from commit b2eba06e16b587acbf5c57944a70ceccecb4f175)
Conflicts:
cogl/cogl-pipeline-private.h
cogl/cogl-pipeline-state-private.h
cogl/cogl-pipeline-state.c
cogl/cogl-pipeline.c
Previously CoglPipelineSnippetList was using the BSD embedded list
type with a mini struct to combine the list node with a pointer to the
snippet. This is effectively equivalent to just using a GList so we
might as well do that. This will help if we eventually want to get rid
of cogl-queue.h
Reviewed-by: Robert Bragg <robert@linux.intel.com>
(cherry picked from commit 54a168f3c7829c427d54ab517533bb9f7384d022)
This adds a new function to enable per-vertex point size on a
pipeline. This can be set with
cogl_pipeline_set_per_vertex_point_size(). Once enabled the point size
can be set either by drawing with an attribute named
'cogl_point_size_in' or by writing to the 'cogl_point_size_out'
builtin from a snippet.
There is a feature flag which must be checked for before using
per-vertex point sizes. This will only be set on GL >= 2.0 or on GLES
2.0. GL will only let you set a per-vertex point size from GLSL by
writing to gl_PointSize. This is only available in GL2 and not in the
older GLSL extensions.
The per-vertex point size has its own pipeline state flag so that it
can be part of the state that affects vertex shader generation.
Having to enable the per vertex point size with a separate function is
a bit awkward. Ideally it would work like the color attribute where
you can just set it for every vertex in your primitive with
cogl_pipeline_set_color or set it per-vertex by just using the
attribute. This is harder to get working with the point size because
we need to generate a different vertex shader depending on what
attributes are bound. I think if we wanted to make this work
transparently we would still want to internally have a pipeline
property describing whether the shader was generated with per-vertex
support so that it would work with the shader cache correctly.
Potentially we could make the per-vertex property internal and
automatically make a weak pipeline whenever the attribute is bound.
However we would then also need to automatically detect when an
application is writing to cogl_point_size_out from a snippet.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
(cherry picked from commit 8495d9c1c15ce389885a9356d965eabd97758115)
Conflicts:
cogl/cogl-context.c
cogl/cogl-pipeline-private.h
cogl/cogl-pipeline.c
cogl/cogl-private.h
cogl/driver/gl/cogl-pipeline-progend-fixed.c
cogl/driver/gl/gl/cogl-pipeline-progend-fixed-arbfp.c
Since _cogl_pipeline_update_blend_enable() can sometimes show up quite
high in profiles; instead of calling
_cogl_pipeline_update_blend_enable() whenever we change pipeline state
that may affect blending we now just set a dirty flag and when we flush
a pipeline we check this dirty flag and lazily calculate whether blender
really needs to be enabled if it's set.
Since it turns out we were too optimistic in assuming most GL drivers
would recognize blending with ADD(src,0) is equivalent to disabling
GL_BLEND we now check this case ourselves so we can always explicitly
disable GL_BLEND if we know we don't need blending.
This introduces the idea of an 'unknown_color_alpha' boolean to the
pipeline flush code which is set whenever we can't guarantee that the
color attribute is opaque. For example this is set whenever a user
specifies a color attribute with 4 components when drawing a primitive.
This boolean needs to be cached along with every pipeline because
pipeline::real_blend_enabled depends on this and so we need to also call
_cogl_pipeline_update_blend_enable() if the status of this changes.
Incidentally with this patch we now no longer ever use
_cogl_pipeline_set_blend_enable() internally. For now the internal api
hasn't been removed though since we might want to consider re-purposing
it as a public api since it will now not conflict with our own internal
state tracking and could provide a more convenient way to disable
blending than setting a blend string.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit ab2ae18f3207514c91fa6fd9f2d3f2ed93a86497)
This removes the various checks for != COGL_DRIVER_GLES1 when tracking
blend state that was trying to avoid checking the equation or alpha
component factors when they are known to be fixed on gles1. Now we just
rely on the opengl driver to do the right thing for the different
drivers and ignore the differences in the general pipeline state
tracking.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f67c7eaf23e1e2088e9956cb2b66dfdc9abc8b3b)
This adds a new CoglDriver for GL 3 called COGL_DRIVER_GL3. When
requested, the GLX, EGL and SDL2 winsyss will set the necessary
attributes to request a forward-compatible core profile 3.1 context.
That means it will have no deprecated features.
To simplify the explosion of checks for specific combinations of
context->driver, many of these conditionals have now been replaced
with private feature flags that are checked instead. The GL and GLES
drivers now initialise these private feature flags depending on which
driver is used.
The fixed function backends now explicitly check whether the fixed
function private feature is available which means the GL3 driver will
fall back to always using the GLSL progend. Since Rob's latest patches
the GLSL progend no longer uses any fixed function API anyway so it
should just work.
The driver is currently lower priority than COGL_DRIVER_GL so it will
not be used unless it is specificly requested. We may want to change
this priority at some point because apparently Mesa can make some
memory savings if a core profile context is used.
In GL 3, getting the combined extensions string with glGetString is
deprecated so this patch changes it to use glGetStringi to build up an
array of extensions instead. _cogl_context_get_gl_extensions now
returns this array instead of trying to return a const string. The
caller is expected to free the array.
Some issues with this patch:
• GL 3 does not support GL_ALPHA format textures. We should probably
make this a feature flag or something. Cogl uses this to render text
which currently just throws a GL error and breaks so it's pretty
important to do something about this before considering the GL3
driver to be stable.
• GL 3 doesn't support client side vertex buffers. This probably
doesn't matter because CoglBuffer won't normally use malloc'd
buffers if VBOs are available, but it might but worth making
malloc'd buffers a private feature and forcing it not to use them.
• GL 3 doesn't support the default vertex array object. This patch
just makes it create and bind a single non-default vertex array
object which gets used just like the normal default object. Ideally
it would be good to use vertex array objects properly and attach
them to a CoglPrimitive to cache the state.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
(cherry picked from commit 66c9db993595b3a22e63f4c201ea468bc9b88cb6)
This adds a new "fixed-arbfp" progend so we now have 3 distinct ways of
setting up the state of a pipeline:
» fixed; where the vertex and fragment processing are implemented
using fixed function opengl apis.
» fixed-arbfp; where vertex processing is implemented using fixed
function opengl apis but fragment processing is implemented
using the ARB Fragment Processing language.
» glsl; there vertex and fragment processing are both implemented
using glsl.
This means we avoid unusual, combinations such as glsl for vertex
processing and arbfp for fragment processing, and also avoid pairing
fixed-function vertex processing with glsl fragment processing which we
happen to know hits some awkward code paths in Mesa that lead to poor
performance.
As part of this change, the progend now implies specific vertend and
fragend choices so instead of associating a vertend and fragend with a
pipeline we now just associate a progend choice.
When flushing a pipeline and choosing what progend to use, we now call a
progend->start() method that is able to determine if the vertend and
fragend together will be able to handle the given pipeline so the
vertend and fragend ->start() methods no longer need to return a boolean
status.
Since we now don't need to support glsl used in conjunction with fixed
function this will allow us to avoid ever using OpenGL builtin attribute
names, though this patch doesn't change that yet.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit cec381f50c7a2f2186bd4a8c5f38fecd5f099075)
Although we use GLib internally in Cogl we would rather not leak GLib
api through Cogl's own api, except through explicitly namespaced
cogl_glib_ / cogl_gtype_ feature apis.
One of the benefits we see to not leaking GLib through Cogl's public API
is that documentation for Cogl won't need to first introduce the Glib
API to newcomers, thus hopefully lowering the barrier to learning Cogl.
This patch provides a Cogl specific typedef for reporting runtime errors
which by no coincidence matches the typedef for GError exactly. If Cogl
is built with --enable-glib (default) then developers can even safely
assume that a CoglError is a GError under the hood.
This patch also enforces a consistent policy for when NULL is passed as
an error argument and an error is thrown. In this case we log the error
and abort the application, instead of silently ignoring it. In common
cases where nothing has been implemented to handle a particular error
and/or where applications are just printing the error and aborting
themselves then this saves some typing. This also seems more consistent
with language based exceptions which usually cause a program to abort if
they are not explicitly caught (which passing a non-NULL error signifies
in this case)
Since this policy for NULL error pointers is stricter than the standard
GError convention, there is a clear note in the documentation to warn
developers that are used to using the GError api.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit b068d5ea09ab32c37e8c965fc8582c85d1b2db46)
Note: Since we can't change the Cogl 1.x api the patch was changed to
not rename _error_quark() functions to be _error_domain() functions and
although it's a bit ugly, instead of providing our own CoglError type
that's compatible with GError we simply #define CoglError to GError
unless Cogl is built with glib disabled.
Note: this patch does technically introduce an API break since it drops
the cogl_error_get_type() symbol generated by glib-mkenum (Since the
CoglError enum was replaced by a CoglSystemError enum) but for now we
are assuming that this will not affect anyone currently using the Cogl
API. If this does turn out to be a problem in practice then we would be
able to fix this my manually copying an implementation of
cogl_error_get_type() generated by glib-mkenum into a compatibility
source file and we could also define the original COGL_ERROR_ enums for
compatibility too.
Note: another minor concern with cherry-picking this patch to the 1.14
branch is that an api scanner would be lead to believe that some APIs
have changed, and for example the gobject-introspection parser which
understands the semantics of GError will not understand the semantics of
CoglError. We expect most people that have tried to use
gobject-introspection with Cogl already understand though that it is not
well suited to generating bindings of the Cogl api anyway and we aren't
aware or anyone depending on such bindings for apis involving GErrors.
(GnomeShell only makes very-very minimal use of Cogl via the gjs
bindings for the cogl_rectangle and cogl_color apis.)
The main reason we have cherry-picked this patch to the 1.14 branch
even given the above concerns is that without it it would become very
awkward for us to cherry-pick other beneficial patches from master.
The coding style has for a long time said to avoid using redundant glib
data types such as gint or gchar etc because we feel that they make the
code look unnecessarily foreign to developers coming from outside of the
Gnome developer community.
Note: When we tried to find the historical rationale for the types we
just found that they were apparently only added for consistent syntax
highlighting which didn't seem that compelling.
Up until now we have been continuing to use some of the platform
specific type such as gint{8,16,32,64} and gsize but this patch switches
us over to using the standard c99 equivalents instead so we can further
ensure that our code looks familiar to the widest range of C developers
who might potentially contribute to Cogl.
So instead of using the gint{8,16,32,64} and guint{8,16,32,64} types this
switches all Cogl code to instead use the int{8,16,32,64}_t and
uint{8,16,32,64}_t c99 types instead.
Instead of gsize we now use size_t
For now we are not going to use the c99 _Bool type and instead we have
introduced a new CoglBool type to use instead of gboolean.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit 5967dad2400d32ca6319cef6cb572e81bf2c15f0)
This option to GCC makes it give a warning whenever a global function
is defined without a declaration. This should catch cases were we've
defined a function but forgot to put it in a header. In that case it
is either only used within one file so we should make it static or we
should declare it in a header.
The following changes where made to fix problems:
• Some functions were made static
• cogl-path.h (the one containing the 1.0 API) was split into two
files, one defining the functions and one defining the enums so that
cogl-path.c can include the enum and function declarations from the
2.0 API as well as the function declarations from the 1.0 API.
• cogl2-clip-state has been removed. This only had one experimental
function called cogl_clip_push_from_path but as this is unstable we
might as well remove it favour of the equivalent cogl_framebuffer_*
API.
• The GLX, SDL and WGL winsys's now have a private header to define
their get_vtable function instead of directly declaring in the C
file where it is called.
• All places that were calling COGL_OBJECT_DEFINE need to have the
cogl_is_whatever function declared so these have been added either
as a public function or in a private header.
• Some files that were not including the header containing their
function declarations have been fixed to do so.
• Any unused error quark functions have been removed. If we later want
them we should add them back one by one and add a declaration for
them in a header.
• _cogl_is_framebuffer has been renamed to cogl_is_framebuffer and
made a public function with a declaration in cogl-framebuffer.h
• Similarly for CoglOnscreen.
• cogl_vdraw_indexed_attributes is called
cogl_framebuffer_vdraw_indexed_attributes in the header. The
definition has been changed to match the header.
• cogl_index_buffer_allocate has been removed. This had no declaration
and I'm not sure what it's supposed to do.
• CoglJournal has been changed to use the internal CoglObject macro so
that it won't define an exported cogl_is_journal symbol.
• The _cogl_blah_pointer_from_handle functions have been removed.
CoglHandle isn't used much anymore anyway and in the few places
where it is used I think it's safe to just use the implicit cast
from void* to the right type.
• The test-utils.h header for the conformance tests explicitly
disables the -Wmissing-declaration option using a pragma because all
of the tests declare their main function without a header. Any
mistakes relating to missing declarations aren't really important
for the tests.
• cogl_quaternion_init_from_quaternion and init_from_matrix have been
given declarations in cogl-quaternion.h
Reviewed-by: Robert Bragg <robert@linux.intel.com>
The get_all_uniform_values function tries to walk the parent hierarchy
of pipelines to find pipelines overriding the uniforms state and then
grabs the values from the override. However it was accessing data
inside the ‘big state’ even if the pipeline didn't override the
uniforms state so it would crash if it encountered a parent pipeline
with no big state.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
If we need to realloc the uniforms overrides array for a pipeline to
insert a new override then we copy the old state into the new allocation
for the entries surrounding the inserted entry.
This patch fixes a mistake in how we copied the old entries that follow
the inserted entry since we were actually copying to begining of the new
allocation and potentially reading from beyond the extents of the old
allocation.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
This adds a hook called COGL_SNIPPET_HOOK_TEXTURE_COORD_TRANSFORM.
This can be used to alter the application of the layer user matrix to
a texture coordinate or it can bypass it altogether.
This is the first per-layer hook that affects the vertex shader state
so the patch includes the boilerplate needed to get that to work.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
Instead of specifying the hook point when adding to the pipeline using
a separate function for each hook, the hook is now a property of the
snippet. The hook is set on construction and is then read-only.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
Whenever snippets are enabled we can't determine whether the final
color will be fully opaque so we just have to assume blending should
be enabled.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
This adds a per-layer snippet hook for the texure lookup. Here the
snippet can modify the texture coordinates used for the lookup or
modify the texel resulting from the lookup. This is the first
per-layer hook so this also adds the
COGL_PIPELINE_LAYER_STATE_FRAGMENT_SNIPPETS state and all of the
boilerplate needed to make that work.
Most of the functions used by the pipeline state to manage the snippet
list has been moved into cogl-pipeline-snippet.c so that it can be
shared with the layer state.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
This adds two new public experimental functions for attaching
CoglSnippets to two hook points on a CoglPipeline:
void cogl_pipeline_add_vertex_hook (CoglPipeline *, CoglSnippet *)
void cogl_pipeline_add_fragment_hook (CoglPipeline *, CoglSnippet *)
The hooks are intended to be around the entire vertex or fragment
processing. That means the pre string in the snippet will be inserted
at the very top of the main function and the post function will be
inserted at the very end. The declarations get inserted in the global
scope.
The snippets are stored in two separate linked lists with a structure
containing an enum representing the hook point and a pointer to the
snippet. The lists are meant to be for hooks that affect the vertex
shader and fragment shader respectively. Although there are currently
only two hooks and the names match these two lists, the intention is
*not* that each new hook will be in a separate list. The separation of
the lists is just to make it easier to determine which shader needs to
be regenerated when a new snippet is added.
When a pipeline becomes the authority for either the vertex or
fragment snipper state, it simply copies the entire list from the
previous authority (although of course the shader snippet objects are
referenced instead of copied so it doesn't duplicate the source
strings).
Each string is inserted into its own block in the shader. This means
that each string has its own scope so it doesn't need to worry about
name collisions with variables in other snippets. However it does mean
that the pre and post strings can't share variables. It could be
possible to wrap both parts in one block and then wrap the actual
inner hook code in another block, however this would mean that any
further snippets within the outer snippet would be able to see those
variables. Perhaps something to consider would be to put each snippet
into its own function which calls another function between the pre and
post strings to do further processing.
The pipeline cache for generated programs was previously shared with
the fragment shader cache because the state that affects vertex
shaders was a subset of the state that affects fragment shaders. This
is no longer the case because there is a separate state mask for
vertex snippets so the program cache now has its own hash table.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
When comparing uniform values, it was not correctly handling the case
where pipeline0 has the value set but pipeline1 does not (only the
other way around) so it would crash.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
Previously the uniform overrides were stored in a linked list. Now
they are stored in a g_malloc'd array. The values are still tightly
packed so that there is only a value for each uniform that has a
corresponding bit in override_mask. The allocated size of the array
always exactly corresponds to the number of bits set in the
override_mask. This means that when a new uniform value is set on a
pipeline it will have to grow the array and copy the old values
in. The assumption is that setting a value for a new uniform is much
less frequent then setting a value for an existing uniform so it makes
more sense to optimise the latter.
The advantage of using an array is that we can quickly jump to right
boxed value given a uniform location by doing a population count in
the bitmask for the number of bits less than the given uniform
location. This can be done in O(1) time whereas the old approach using
a list would scale by the number of bits set.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
This adds the following new public experimental functions to set
uniform values on a CoglPipeline:
void
cogl_pipeline_set_uniform_1f (CoglPipeline *pipeline,
int uniform_location,
float value);
void
cogl_pipeline_set_uniform_1i (CoglPipeline *pipeline,
int uniform_location,
int value);
void
cogl_pipeline_set_uniform_float (CoglPipeline *pipeline,
int uniform_location,
int n_components,
int count,
const float *value);
void
cogl_pipeline_set_uniform_int (CoglPipeline *pipeline,
int uniform_location,
int n_components,
int count,
const int *value);
void
cogl_pipeline_set_uniform_matrix (CoglPipeline *pipeline,
int uniform_location,
int dimensions,
int count,
gboolean transpose,
const float *value);
These are similar to the old functions used to set uniforms on a
CoglProgram. To get a value to pass in as the uniform_location there
is also:
int
cogl_pipeline_get_uniform_location (CoglPipeline *pipeline,
const char *uniform_name);
Conceptually the uniform locations are tied to the pipeline so that
whenever setting a value for a new pipeline the application is
expected to call this function. However in practice the uniform
locations are global to the CoglContext. The names are stored in a
linked list where the position in the list is the uniform location.
The global indices are used so that each pipeline can store a mask of
which uniforms it overrides. That way it is quicker to detect which
uniforms are different from the last pipeline that used the same
CoglProgramState so it can avoid flushing uniforms that haven't
changed. Currently the values are not actually compared which means
that it will only avoid flushing a uniform if there is a common
ancestor that sets the value (or if the same pipeline is being flushed
again - in which case the pipeline and its common ancestor are the
same thing).
The uniform values are stored in the big state of the pipeline as a
sparse linked list. A bitmask stores which values have been overridden
and only overridden values are stored in the linked list.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
This adds two new experimental public functions to replace the old
internal _cogl_pipeline_set_cull_face_state function:
void
cogl_pipeline_set_cull_face_mode (CoglPipeline *pipeline,
CoglPipelineCullFaceMode cull_face_mode);
void
cogl_pipeline_set_front_face_winding (CoglPipeline *pipeline,
CoglWinding front_winding);
There are also the corresponding getters.
https://bugzilla.gnome.org/show_bug.cgi?id=663628
Reviewed-by: Robert Bragg <robert@linux.intel.com>
Since we've had several developers from admirable projects say they
would like to use Cogl but would really prefer not to pull in
gobject,gmodule and glib as extra dependencies we are investigating if
we can get to the point where glib is only an optional dependency.
Actually we feel like we only make minimal use of glib anyway, so it may
well be quite straightforward to achieve this.
This adds a --disable-glib configure option that can be used to disable
features that depend on glib.
Actually --disable-glib doesn't strictly disable glib at this point
because it's more helpful if cogl continues to build as we make
incremental progress towards this.
The first use of glib that this patch tackles is the use of
g_return_val_if_fail and g_return_if_fail which have been replaced with
equivalent _COGL_RETURN_VAL_IF_FAIL and _COGL_RETURN_IF_FAIL macros.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
This adds an internal function to set the backface culling state on a
pipeline. This includes properties to set the culling mode (front,
back or both) and also to set which face is considered the front
(COGL_WINDING_CLOCKWISE or COGL_WINDING_COUNTER_CLOCKWISE). The actual
front face flushed to GL depends on whether we are rendering to an
offscreen buffer or not. This means that when changing between on- and
off- screen framebuffers it now checks whether the last flushed
pipeline has backface culling enabled and forces a reflush of the cull
face state if so.
The backface culling is now set on a pipeline as part of the legacy
state. This is important because some code in Cogl assumes it can
flush a temporary pipeline to revert to a known state, but previously
this wouldn't disable backface culling so things such as flushing the
clip stack could get confused.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
Since cogl-pipeline.c has become very unwieldy this make a start at
trying to shape this code back into a manageable state. This patche
moves all the API relating to core pipeline state into
cogl-pipeline-state.c. This doesn't move code relating to layer state
out nor does it move any of the code supporting the core design
of CoglPipeline itself.
This change alone factors out 2k lines of code from cogl-pipeline.c
which is obviously a good start. The next step will be to factor
out the layer state and then probably look at breaking all of this
state code down into state-groups.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
