Commit Graph

2774 Commits

Author SHA1 Message Date
Neil Roberts
3f28168f1e attribute: Add the buffer and normalized property APIs to the header
The header was missing declarations for
cogl_attribute_{get,set}_{buffer,normalized}.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-12-06 19:02:06 +00:00
Neil Roberts
f005f517fe Flush matrices in the progend and flip with a vector
Previously flushing the matrices was performed as part of the
framebuffer state. When on GLES2 this matrix flushing is actually
diverted so that it only keeps a reference to the intended matrix
stack. This is necessary because on GLES2 there are no builtin
uniforms so it can't actually flush the matrices until the program for
the pipeline is generated. When the matrices are flushed it would
store the age of modifications on the matrix stack so that it could
detect when the matrix hasn't changed and avoid flushing it.

This patch changes it so that the pipeline is responsible for flushing
the matrices even when we are using the GL builtins. The same
mechanism for detecting unmodified matrix stacks is used in all
cases. There is a new CoglMatrixStackCache type which is used to store
a reference to the intended matrix stack along with its last flushed
age. There are now two of these attached to the CoglContext to track
the flushed state for the global matrix builtins and also two for each
glsl progend program state to track the flushed state for a
program. The framebuffer matrix flush now just updates the intended
matrix stacks without actually trying to flush.

When a vertex snippet is attached to the pipeline, the GLSL vertend
will now avoid using the projection matrix to flip the rendering. This
is necessary because any vertex snippet may cause the projection
matrix not to be used. Instead the flip is done as a forced final step
by multiplying cogl_position_out by a vec4 uniform. This uniform is
updated as part of the progend pre_paint depending on whether the
framebuffer is offscreen or not.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-12-06 19:02:06 +00:00
Neil Roberts
7590fc3ec3 snippet: Add a hook for the vertex transform
This adds a hook to wrap or replace the vertex transform stage.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-12-06 19:02:06 +00:00
Neil Roberts
68b3643b25 snippet: Add a hook for the layer texture coordinate transformation
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>
2011-12-06 19:02:06 +00:00
Neil Roberts
b7e15929b6 snippet: Add a hook for the layer fragment processing
This adds a hook to replace or wrap the fragment processing for a
particular layer.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-12-06 19:02:06 +00:00
Neil Roberts
4581ce5f15 fragend-glsl: Don't generate the default texture lookup if replaced
Previously the function containing the default texture lookup is
always generated regardless of whether there is a snippet with a
replace string which would cause it not be used. Now this snippets are
all scanned to check for replace strings before generating the texture
lookup.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-12-06 19:02:06 +00:00
Neil Roberts
272c6567fa glsl-fragend: Store the cached variables in global variables
The variables caching the result of texture lookups and layer
calculations are now stored in global variables so that when a hook
for the layer processing is added the variables can still be accessed
even if the generated code is within a separate function.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-12-06 19:02:06 +00:00
Neil Roberts
a0903e0cfb snippet: Don't generate code for snippets that are replaced
The loop that generates code for a list of snippets now starts from
the first snippet that has a replace string. Any snippets before that
would be ignored so there's no point in generating code for them.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-12-06 19:02:06 +00:00
Neil Roberts
cdc4bd6361 snippet: Fix detecting the last snippet
The function for generating the GLSL for a list of snippets was trying
to detect the last snippet so that it could use a different function
name. However this wouldn't work if the last snippet has a different
hook. To fix this it now just counts the snippets that have the same
hook beforehand and detects the last one using the count.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-12-06 19:02:06 +00:00
Neil Roberts
09c2e4abe7 snippet: Move the hook to be a property of the snippet
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>
2011-12-06 19:02:06 +00:00
Neil Roberts
5be5a03343 pipeline: Assume blend enabled when there are snippets
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>
2011-12-06 19:02:06 +00:00
Neil Roberts
df0f9a862f pipeline: Add a snippet hook for the texture lookup
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>
2011-12-06 19:02:06 +00:00
Neil Roberts
4cdf66f89b pipeline: Move the snippet source loop into a common function
The two loops that generate the functions for the snippets in the
fragend and vertend are very similar so to avoid code duplication this
patch moves the logic to its own function in a new
cogl-pipeline-snippet.c file.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-12-06 19:02:06 +00:00
Neil Roberts
717e96b304 snippet: Add a 'replace' string
If present, the 'replace' string will be used instead of whatever code
would normally be invoked for that hook point. It will also replace
any previous snippets.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-12-06 19:02:05 +00:00
Neil Roberts
0fdd5690c6 pipeline: Wrap each snippet in its own function
Each snippet is now given its own function which contains the pre and
post strings. Between these strings the function will chain on to
another function. The generated cogl source is now stored in a
function called cogl_generated_source() which the last snippet will
chain on to. This should make it so that each snippet has its own
namespace for local variables and it can share variables declared in
the pre string in the post string. Hopefully the GLSL compiler will
just inline all of the functions so it shouldn't make much difference
to the compiled output.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-12-06 19:02:05 +00:00
Neil Roberts
2f66cd7036 Add a conformance test for CoglSnippets
This adds a conformance test testing various operations on
CoglSnippets.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-12-06 19:02:05 +00:00
Neil Roberts
d38ae0284b cogl-pipeline: Add two hook points for adding shader snippets
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>
2011-12-06 19:02:05 +00:00
Neil Roberts
f37738453e Add a CoglSnippet type
This adds a CoglObject called CoglSnippet which will be used to store
strings used as GLSL snippets to be attached at hook points to a
CoglPipeline. The snippets can currently contain three strings:

declarations - This will be placed in the global scope and is intended
               to be used to declare uniforms, attributes and
               functions.

pre           - This will be inserted before the hook point.

post          - This will be inserted after the hook point.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-12-06 19:02:05 +00:00
Neil Roberts
0c82c296bf framebuffer: Handle a NULL previous framebuffer when flushing
_cogl_framebuffer_flush_state needs to handle the case where
ctx->current_draw_buffer is NULL because this will be set in the
destructor for CoglFramebuffer if the framebuffer being destroyed is
the current framebuffer. This patch just makes it assume all state has
changed in that case.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-12-06 19:02:05 +00:00
Robert Bragg
7283e0a49c rework enabling of attributes, removing _cogl_enable()
This removes the limited caching of enabled attributes done by
_cogl_enable() and replaces it with a more generalized set of bitmasks
associated with the context that allow us to efficiently compare the set
of attribute locations that are currently enabled vs the new locations
that need enabling so we only have to inform OpenGL of the changes in
which locations are enabled/disabled.

This also adds a per-context hash table for mapping attribute names to
global name-state structs which includes a unique name-index for any
name as well as pre-validated information about builtin "cogl_"
attribute names including whether the attribute is normalized and what
texture unit a texture attribute corresponds too.

The name-state hash table means that cogl_attribute_new() now only needs
to validate names the first time they are seen.

CoglAttributes now reference a name-state structure instead of just the
attribute name, so now we can efficiently get the name-index for any
attribute and we can use that to index into a per-glsl-program cache
that maps name indices to real GL attribute locations so when we get
asked to draw a set of attributes we can very quickly determine what GL
attributes need to be setup and enabled. If we don't have a cached
location though we can still quickly access the string name so we can
query OpenGL.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:57 +00:00
Robert Bragg
2112af0bc5 framebuffer: deal with winding state like other state
We should not be deciding whether we need to really update the GL face
winding state at the point where a new framebuffer has been pushed, we
should be waiting until we have really been asked to flush some
framebuffer state otherwise we may do redundant work if multiple
framebuffers are pushed/popped before something is really drawn.

This integrates the face winding state tracking with the design we have
for handling most of the other framebuffer state so we benefit from the
optimizations for minimizing the cost of _cogl_framebuffer_flush_state()

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:57 +00:00
Robert Bragg
cf4b228fc0 framebuffer: track fb color mask like other fb state
We should not be deciding whether we need to really update the GL color
mask state at the point where a new framebuffer has been pushed, we
should be waiting until we have really been asked to flush some
framebuffer state otherwise we may do redundant work if multiple
framebuffers are pushed/popped before something is really drawn.

This integrates the color mask state tracking with the design we have
for handling most of the other framebuffer state so we benefit from the
optimizations for minimizing the cost of _cogl_framebuffer_flush_state()

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:57 +00:00
Robert Bragg
042dc7c8cd framebuffer: Optimize _cogl_framebuffer_flush_state()
Previously the cost of _cogl_framebuffer_state_flush() would always
scale by the total amount of state tracked by CoglFramebuffer even in
cases where we knew up-front that we only wanted to flush a subset of
the state or in cases where we requested to flush the same framebuffer
multiple times with no changes being made to the framebuffer.

We now track a set of state changed flags with each framebuffer and
track the current read/draw buffers as part of the CoglContext so that
we can quickly bail out when asked to flush the same framebuffer
multiple times with no changes.

_cogl_framebuffer_flush_state() now takes a mask of the state that we
want to flush and the implementation has been redesigned so that the
cost of checking what needs to be flushed and flushing those changes
now scales by how much state we actually plan to update.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:57 +00:00
Robert Bragg
6ab371d587 clip-state: remove redundant _cogl_clip_state_flush() api
There was only one place where we called _cogl_clip_state_flush() in
_cogl_framebuffer_flush_state() and we can just as well use
_cogl_clip_state_get_stack() and _cogl_clip_stack_flush() directly
instead.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:57 +00:00
Robert Bragg
6d1bf9a0d4 remove redundant _cogl_clip_stack_dirty() api
The only place we were calling _cogl_clip_stack_dirty() was when
changing the current draw_buffer which also implies a change in
the current clip stack. _cogl_clip_stack_flush() would already
be able to quickly determine that the clip stack has changed by
checking ctx->current_clip_stack so there isn't really any need
to explicitly mark the clip_stack state as dirty.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:57 +00:00
Robert Bragg
104a6be84c cogl-matrix-stack: remove flush avoidance checks
The aim is to make cogl-framebuffer.c responsible for avoiding redundant
flushing of its matrix stacks so this removes the checks done directly
within cogl-matrix-stack.c.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:57 +00:00
Robert Bragg
ff4921cef9 docs: updates cogl-matrix.h docs to be consistent
This update some of the cogl-matrix.h documentation to be consistent
with the corresponding documentation for framebuffer matrix-stack
methods in cogl-framebuffer.h

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:57 +00:00
Robert Bragg
39a7bcff67 matrix-stack: Avoid referencing the default CoglContext
This removes the use of _COGL_GET_CONTEXT() from cogl-matrix-stack.c
as part of the ongoing effort to evolve cogl to get rid of the need for
a default context.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:56 +00:00
Robert Bragg
5156365eae path: use cogl_framebuffer_push_path_clip in cogl2-path.c
_cogl_path_fill_nodes_with_clipped_rectangle() sometimes falls back to
pushing a framebuffer clip region and filling the region using
cogl_rectangle(). Since we aim to eventually deprecate
cogl_clip_push_from_path() as it relies on the default CoglContext we
would rather this internal code update a framebuffer's clip-state using
the cogl_framebuffer clip stack api instead.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:56 +00:00
Robert Bragg
36575502e8 docs: make cogl2-clip-state.h docs consistent
This updates some of the cogl2-clip-state.h cogl_clip_ API documentation
to make it consistent with the documentation for corresponding
framebuffer clip stack API in cogl-framebuffer.h

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:56 +00:00
Robert Bragg
de04a58f30 docs: make cogl.h clip stack docs consistent
This updates some of the cogl.h cogl_clip_ API documentation to make it
consistent with the documentation for corresponding framebuffer clip
stack API in cogl-framebuffer.h

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:56 +00:00
Robert Bragg
6bb409e996 framebuffer: Add CoglFramebuffer clip stack methods
This adds CoglFramebuffer methods for accessing the clip stack. We plan
on making some optimizations to how framebuffer state is flushed which
will require us to track when a framebuffer's clip state has changed.

This api also ties in to the longer term goal of removing the need for a
default global CoglContext since these methods are all implicitly
related to a specific context via their framebuffer argument.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:56 +00:00
Robert Bragg
652c8c31cf avoid touching fb matrix stack directly
This ensures we don't touch a framebuffer's matrix stack directly if we
are also relying on _cogl_framebuffer_flush_state(). We want to get to
the point where we can set dirty flags against framebuffer state at the
point it changes but that means we can't allow direct access to the
matrix stack. _cogl_texture_draw_and_read() has now been changed so it
uses cogl_framebuffer_ methods to update the matrix stacks including
adding new internal _cogl_framebuffer_push/pop_projection() functions
that allow us to set transient projections.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:56 +00:00
Robert Bragg
5ca78668af Adds cogl_framebuffer_identity_matrix
This adds a cogl_framebuffer_identity_matrix() method that can be used
to reset the current modelview matrix to the identity matrix.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:56 +00:00
Robert Bragg
a81458f5b9 docs: make cogl.h matrix stack documentation consistent
This updates the cogl.h matrix stack documentation consistent with the
corresponding documentation in cogl-framebuffer.h for the framebuffer
matrix stack methods.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:56 +00:00
Robert Bragg
ccde191191 framebuffer: Add CoglFramebuffer matrix stack methods
This adds cogl_framebuffer_ methods to update the modelview and
projection matrix stacks to replace functions like cogl_translate(),
cogl_rotate() and cogl_scale() etc.

This is part of the on-going effort to get rid of the global CoglContext
pointer since the existing methods don't take an explicit pointer to a
CoglContext. All the methods are now related to a context via the
framebuffer.

We added framebuffer methods instead of direct context methods because
the matrix stacks are per-framebuffer and as well as removing the global
CoglContext we would rather aim for a more direct state access API
design than, say, cairo or OpenGL, so we'd like to avoid needing the
cogl_push/pop_framebuffer(). We anticipate that Cogl will mostly be
consumed by middleware graphics layers such as toolkits or game engines
and feel that a more stateless model will avoid impedance mismatches if
higher levels want to expose a stateless model to their developers and
statefullness can still be added by higher levels if really desired.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:56 +00:00
Robert Bragg
595311ba55 cogl-matrix-stack: Adds _cogl_matrix_stack_equal
Adds a function for comparing the top matrix entries of two matrix
stacks.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:56 +00:00
Robert Bragg
a4f3d0d18b matrix: Add cogl_matrix_orthographic
This adds an experimental cogl_matrix_orthographic() function that is
more consistent with other Cogl api by taking x_1, y_1, x_2, y_2
arguments to define the top-left and bottom-right coordinates of the
orthographic coordinates instead of OpenGL style left, right, bottom and
top values.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:56 +00:00
Robert Bragg
ba02f70961 texture: assert set_region size <= src size
This improves the documentation for cogl_texture_set_region() and
cogl_texture_set_region_from_bitmap() to explain that the region can't
be larger than the source data and also adds runtime assertions that
such a request isn't made.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:56 +00:00
Robert Bragg
aa59dc5e8d docs: Fix a s/called/culled/ typo
The documentation for CoglPipelineCullFaceMode had a repeated typo with
"called" being used instead of "culled" which this fixes.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-12-06 18:51:56 +00:00
Tomeu Vizoso
1de6575ca2 build: Add missing include in cogl-winsys-egl-private.h
We were missing APIENTRY.
2011-12-06 19:49:00 +01:00
Rob Bradford
7f2a896351 kms: Avoid using struct member before it is set
When saving the CRTC we were trying to use a struct member for the encoder
that wasn't valid at that point in time - instead use the local variable.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-11-30 15:37:09 +00:00
Neil Roberts
da154b9a3b pipeline: Add a missing break for the uniforms state
_cogl_pipeline_init_multi_property_sparse_state was missing a break in
the case statement handling uniforms. This doesn't yet matter because
it is the last one handled anyway but it will bite someone later.

Reviewed-by: Robert Bragg <robert@linux.intel.com>
2011-11-29 13:03:25 +00:00
Robert Bragg
be40cbb8c3 build: include missing headers in cogl-winsys-egl-private.h
We were missing various platform header includes in
cogl-winsys-egl-private.h when building support for non KMS egl
platforms.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-11-28 23:19:45 +00:00
Robert Bragg
c2729e4ea8 build: don't inc egl headers for non EGL builds
For example when building on windows we don't want to require EGL
headers when compiling cogl-renderer.c or cogl-texture-2d.c so we make
sure not to include cogl-winsys-egl-private.h if we aren't supporting
EGL.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-11-28 23:19:35 +00:00
Robert Bragg
40b14c6084 kms: avoid using redundant g* and GL* types
A small, pedantic change to remove the use of redundant gint and GLuint
types instead of int and unsigned int.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-11-28 22:43:15 +00:00
Robert Bragg
056daad850 build: only include cogl-winsys-kms.h if KMS enabled
Only include cogl-winsys-kms.h in cogl-winsys-egl-private.h if KMS
support has been enabled.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-11-28 22:39:19 +00:00
Robert Bragg
aa6433d8c5 kms: Check for EGL_KHR_surfaceless_opengl extension
This adds a check for the EGL_KHR_surfaceless_opengl extension which we
depend on for being able to MakeCurrent (NO_SURFACE) as well as create a
context without and EGLConfig.

Reviewed-by: Rob Bradford <rob@linux.intel.com>
2011-11-28 17:38:17 +00:00
Robert Bragg
7a91c91994 kms: flatten setup_kms into _cogl_winsys_kms_display_setup
Since _cogl_winsys_kms_display_setup was basically just calling
setup_kms() it made sense to fold the code of setup_kms() back into the
_cogl_winsys_kms_display_setup() function.

Reviewed-by: Rob Bradford <rob@linux.intel.com>
2011-11-28 17:38:15 +00:00
Robert Bragg
607bfec48d kms: move Cogl*EGL typedefs to cogl-winsys-egl-private.h
So that the various internal Cogl*EGL typedefs can be available to
cogl-winsys-kms.c this moves them into cogl-winsys-egl-private.h

Reviewed-by: Rob Bradford <rob@linux.intel.com>
2011-11-28 17:37:31 +00:00