This patch gathers all values to be loaded to shader uniforms into a new
struct gl_shader_config along with texture target and filter
information. Struct gl_shader becomes opaque outside of gl-shaders.c.
Everything that used or open-coded these are converted.
The aim is to make gl-renderer.c easier to read. Previously, uniform
values were loaded up in various places, texture units were set up in
one place, textures were bound into units in different places. Stuff was
all over the place.
Now, shader requirements and associated uniform data is stored in a
single struct. The data is loaded into a shader program in one function
only.
That makes it easy for things like maybe_censor_override() to replace
the whole config rather than poke only the shader requirements. This may
not look like much right now, but when color management adds more
uniforms and even hardcoded color need to go through the proper color
pipeline, doing things the old way would become intractable.
Similar simplification can be seen in draw_view(), where the RGBA->RGBX
override becomes more contained. There is no longer a need to "pre-load"
the shader used by triangle fan debug. Triangle fan debug no longer
needs to play tricks with saving and restoring the current shader.
The real benefit of this change will probably come when almost all
shader operations need to take color spaces into account. That means
filling in gl_shader_config parts based on a color transformation.
This is based on an idea Sebastian already used in his Weston color
management work.
Co-authored-by: Sebastian Wick <sebastian@sebastianwick.net>
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Avoid looking up 'gr' from view->compositor by passing it explicitly
into the functions needing it.
Also fixes the whitespace in repaint_region() signature.
Clarifies code by removing local variables, but also future changes will
need 'gr' more.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
A future change will call this function from draw_view(), so move it
upwards to avoid adding a function declaration.
No functional or even cosmetic change.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
These functions are related to shaders, so they are more at home in
gl-shaders.c. gl-renderer.c is too long already.
This allows making a couple functions static while the moved functions
become non-static. Future changes turn some of these functions into
static again, with the ultimate goal of making struct gl_shader opaque.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
This adds a heuristic for freeing shader programs that have not been
needed for a while. The intention is to stop Weston accumulating shader
programs indefinitely, especially in the future when color management
will explode the number of possible different shader programs.
Shader programs that have not been used in the past minute are freed,
except always keep the ten most recently used shader programs anyway.
The former rule is to ensure we keep shader programs that are actively
used regardless of how many. The latter rule is to prevent freeing too
many shader programs after Weston has been idle for a long time and then
repaints just a small area. Many of the shader programs could still be
relevant even though not needed in the first repaint after idle.
The numbers ten and one minute in the above are arbitrary and not based
on anything.
These heuristics are simpler to implement than e.g. views taking
references on shader programs. Expiry by time allows shader programs to
survive a while even after their last user is gone, with the hope of
being re-used soon. Tracking actual use instead of references also
adapts to what is actually visible rather than what merely exists.
Keeping the shader list in most recently used order might also make
gl_renderer_get_program() more efficient on average.
last_repaint_start time is used for shader timestamp to avoid calling
clock_gettime() more often. Adding that variable is an ABI break, but
libweston major has already been bumped to 10 since last release.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
This is useful for seeing that the shader program garbage collection
works in a future patch.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
One more thing is coming to need this, so add the compositor pointer and
migrate existing places to use it where it simplifies things.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
I have verified that the conversion here follows ITU-R BT.601 except for
the offsets 16/256 and 128/256 which should be 16/255 and 128/255
respectively.
I used to following octave script to verify this:
rf = 0.299;
gf = 0.587;
bf = 0.114;
crdiv = 1.402;
cbdiv = 1.772;
M = [ rf, gf, bf ;
-rf / cbdiv, -gf / cbdiv, (1 - bf) / cbdiv;
(1 - rf) / crdiv, -gf / crdiv, -bf / crdiv ];
YCbCr = [ 'Y'; 'Cb'; 'Cr' ];
RGB = [ 'R'; 'G'; 'B' ];
eq = [ ' '; '='; ' ' ];
l = [ ' [ '; ' [ '; ' [ ' ];
r = [ ' ] '; ' ] '; ' ] ' ];
mat = [
sprintf('%9f %9f %9f', M(1,:));
sprintf('%9f %9f %9f', M(2,:));
sprintf('%9f %9f %9f', M(3,:));
];
[ l YCbCr r eq l mat r l RGB r ]
R = inv(M);
mat = [
sprintf('%9f %9f %9f', R(1,:));
sprintf('%9f %9f %9f', R(2,:));
sprintf('%9f %9f %9f', R(3,:));
];
[ l RGB r eq l mat r l YCbCr r ]
[ R(:,1), R(:,2:3) .* (255/224) ]
The final matrix printed is what the shader uses down to +/- one digit,
so at least 7 correct decimals.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Sampling input texture has nothing to do with view alpha. This clarifies
the code structure.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Reading the input texture is just one part of the future color pipeline,
so separate it into a function of its own. This makes it easier to add
more steps to the pipeline, and shows the green tint is separate as
well.
Making use of early returns, reducing the if-else ladder should help
with readability. Sharing the call to yuva2rgba() likewise.
Setting yuva.w = alpha is not shared though, in case support for AYUV
format might be added in the future.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Do not call texture2D() in the shader when we already have the result.
Simpler code, maybe even a little bit faster?
Suggested-by: Harish Krupo <harishkrupo@gmail.com>
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
These same magic constants were used in all cases, so move them into a
common place.
While we are touching all these lines, also change from the four floats
into a vec4. This allows further clean-up in the next patch.
This makes the code easier to read.
Behavior and results are unchanged.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Mathematically the result is the same, while multiplying RGB with alpha
is easier to understand as correct than the earlier form.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
A more unique name is easier to grep for. Using 'color' as a local
variable might be useful in the future.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
This support is added so that the XYUV shader variant can be tested with
wl_shm from the test suite.
Libwayland version requirement is bumped to get WL_SHM_FORMAT_XYUV8888.
Libwayland is bumped to 1.18 too in the CI image. libwayland-dev package
is dropped, because we build wayland anyway.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
This header is for sharing fallback definitions for drm_fourcc.h. A new
test in tests/yuv-buffer-test.c is going to be needing XYUV8888 format,
and more new formats will be expected with HDR supports.
Share these fallback definitions in one place instead of copying them
all over.
All users of drm_fourcc.h are converted to include weston-drm-fourcc.h
instead for consistency: have the same definitions available everywhere.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
These were introduced in libdrm 2.4.68, commit
268ae7cae5afd76462c3ef14ed9021a2d40c2e57. Weston unconditionally
requires libdrm >= 2.4.95, so these fallback definitions are
unnecessary now.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Extend the existing output-damage test to test
blit_shadow_to_output() specifically. This function had problems
originally, so make sure they can't reappear.
The added quirk is explained in the test.
An additional check of the quirk in gl_renderer_output_create() ensures
that the shadow framebuffer is really used. The test could false-pass if
the shadow is not used.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Proper color management will need blending done with linear light pixel
values, that is, EOTF applied before blending, and then inverse-EOTF
applied for scanout after blending. The simplest way to set that up is
to use an intemediate framebuffer a.k.a shadow buffer containing the
composited image in linear light values, then blit from that to the
actual framebuffer.
This patch implements the shadow buffer, but the linear light
blending is left for another patch. This allows GL-renderer to turn
WESTON_CAP_COLOR_OPS on.
Half-float is chosen as the buffer format because linear light values
require more bits to encode with sufficient precision than the usual
non-linear pixel values.
v2: Use /* */ instead of // (Pekka)
Rename fbo and tex to shadow_{fbo,tex} (Pekka)
Check for OpenGLES capabilities before creating
shadow_{tex,fbo} (Pekka)
Signed-off-by: Harish Krupo <harishkrupo@gmail.com>
v3: Rebased.
Simplified GL version checks (Sebastian)
Apply changes from "libweston: add color ops cap and bool renderer
shadow buffer"
Renamed supports_half_float_texture to has_gl_half_float to
follow the existing naming pattern.
Introduce gl_renderer_create_shadow_16f().
Undo moving of glViewport() call.
Replace half_float_texture_enabled with shadow_exists().
Introduce struct gl_output_state_shadow.
Assert no resizing with shadow.
Fix triangle fan debug.
Rename repaint_from_texture() to blit_shadow_to_output().
Rewrite commit message because linear light blending is not
implemented in this patch.
Fix blit_shadow_to_output() for scaled/transformed outputs and
remove redundant code.
Fix has_gl_half_float determination.
v4: Disable blending in blit_shadow. (Daniel)
Port to gl_renderer_get_program().
Make a generic fbo-texture struct with parameterized format. (Daniel)
Change has_gl_half_float into gl_half_float_type.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
This helps accounting how many shaders live in the cache, what the
shader source code is, and when shaders are compiled.
Signed-off-by: Harish Krupo <harishkrupo@gmail.com>
v2: Resolved rebase conflicts.
Put shader_scope in struct gl_renderer, remove struct
gl_shader_generator.
Wrote commit message.
Rebased for "gl-renderer: rewrite fragment shaders" which completely
changed how shader sources are generated.
Added cache statistics to debug output on subscribe.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Various functions leave the current active texture as whatever. The
functions touched in this commit forgot to reset the active texture to
slot 0 before binding their textures. If not explicitly unbound, this
could leave textures lingering in unused slots, perhaps. Not sure if
that could cause any harm, but for consistency's sake, always use slot 0
when not multitexturing.
Found by code inspection.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
If shader compiling on demand fails, then rather than using whatever
random shader happens to be current, use an explicit fallback shader
painting stuff brown.
The color is chosen dim enough to hopefully not cause problems even in
a HDR setting as it will be written verbatim into the fb/shadow.
This also prevents NULL dereference on shader->key.variant in
draw_view().
One way to test this shader is to hack fragment.glsl:
#if DEF_VARIANT == SHADER_VARIANT_EXTERNAL
#extension GL_OES_EGL_image_external : require
+#error haa haa
#endif
and then run e.g. weston-simple-dmabuf-v4l -f YUYV
with vivid kernel module loaded. This worked on Intel.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
If we are trying to use a NULL shader, it is likely that the shader
compilation failed for some reason. Since we are trying this for a view,
the failure was probably triggered by a client. If there is a client,
get rid of it by sending it a protocol error. Hopefully the compositor
can then continue operation after a glitch on screen.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
This patch modifies the shader generation code so that the shaders are
stitched together based on the requirement instead of creating them
during initialization. This is necessary for HDR use cases where each
surface would have different properties based on which different
de-gamma or tone mapping or gamma shaders are stitched together.
v2: Use /* */ instead of // (Pekka)
Move shader strings to gl-shaders.c file (Pekka)
Remove Makefile.am changes (Pekka)
Use a struct instead of uint32_t for storing requirements (Pekka)
Clean up shader list on destroy (Pekka)
Rename shader_release -> shader_destroy (Pekka)
Move shader creation/deletion into gl-shaders.c (Pekka)
Use create_shaders's multi string capbility instead of
concatenating (Pekka)
v3: Add length check when adding shader string (Pekka)
Signed-off-by: Harish Krupo <harishkrupo@gmail.com>
v4: Rebased, PROTECTION_MODE_ENFORCED converted.
Dropped unnecessary { }.
Ported setup_censor_overrides().
Split out moving code into gl-shaders.c.
Changed to follow "gl-renderer: rewrite fragment shaders",
no more shader source stitching.
Added SHADER_VARIANT_XYUV.
Const'fy function arguments.
Added gl_shader_requirements_cmp() and moved the early return in
use_gl_program().
Moved use_gl_program() before first use in file.
Split solid shader requirements by use case: requirements_censor and
requirements_triangle_fan.
Simplified fragment_debug_binding() since no need to force anything.
Ensure struct gl_shader_requirements has no padding. This allows us
to use normal C syntax instead of memset() and memcpy() when
initializing or assigning. See also:
https://gitlab.freedesktop.org/mesa/mesa/-/issues/2071
Make it also a bitfield to squeeze the size.
v5: Move wl_list_insert() into gl_shader_create() (Daniel)
Compare variant to explicit value. (Daniel)
Change functions to gl_renderer_get_program,
gl_renderer_use_program, and
gl_renderer_use_program_with_view_uniforms.
Use local variable instead of gr->current_shader. (Daniel)
Simplified gl_renderer_get_program.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Do not change in setup_censor_overrides() and then put back gs->shader
in draw_view() when the shader needs to be something else than what the
surface content calls for.
This makes the logic simpler, and makes following changes simpler as
well.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
To help debugging shader compilation errors, print the shader source the
way it was given to the GLSL compiler and with line numbers that match
the compiler error messages.
This is necessary because some snippets are added at runtime to the
beginning, the source is not only what is in the respective .glsl file.
I did look into using #line directives, but you cannot put source file
names to it, only "source string numbers" which must be an integer
expression. If we used #line, the reader would need to know that string
number 0 is the version, string 1 is the config and string number 2 is
fragment.glsl. I think that would have been too cumbersome.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
The main goal of this patch is to improve the readability of how and
what fragment shaders are generated.
Instead of having C code that assembles each shader variant from literal
string snippets, create one big fragment shader source that has
everything in it. This relies on a GLSL compiler to optimize statically
false conditions and unused uniforms away.
Having all the fragment shader code in one file, uncluttered by C string
literal syntax, improves readability significantly. A disadvantage is
that the code is more verbose, but it allows comments much better.
The actual shader code is kept unchanged except:
- FRAGMENT_CONVERT_YUV macro is now a proper function
- GLSL version is explicitly set to 1.00 ES
- RGBA and EXTERNAL use the same path, the difference is how the sampler
is declared
Further shader code consolidation is possible, but is left for another
time.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
This patch adds the tooling for incorporating files as C data, so that
files can be built into the binaries. The tool is in Python to avoid
adding extra dependencies like xxd.
xxd.py is copied from Mesa as-is, from commit
b729cd58d76f97f3fc04a67569535ee5ef2f5278 (master branch on 2021-01-26),
a.k.a 21.0-branchpoint-635-gb729cd58d76.
Moving the GLSL vertex shader into a separate file is not that
interesting, the purpose of this commit is to provide a simple
demonstration of the tooling. The real benefits come in a following
patch where the fragment shaders are re-written and externalized.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
This is purely moving code as is with no changes other than making the
three functions non-static.
Originally this was part of "gl-renderer: Requirement based shader
generation" by Harish Krupo, but that patch made also big changes to the
code at the same time. Patches are easier to review when code movement
is separate from behavioral changes, therefore I introduced this patch.
Cc: Harish Krupo <harishkrupo@gmail.com>
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
The main point here is to print "GL ES %d.%d" instead of "GL ES 2"
because GL-renderer can and will use GL ES 3 features when present.
Saying it's GL ES 2 renderer is not quite true.
To print that, I need to extract major, minor from gr->gl_version and
those didn't have ready made macros yet. While writing the extraction,
make all these trivial functions, so that the compiler might warn us if
one passes e.g. negative literal numbers to gr_gl_version(). Explicit
types help keeping the bit operations safe too.
The only purpose for GR_GL_VERSION_INVALID was to fall back to version
2.0. Moving the fallback and logging into get_gl_version() makes that
macro unnecessary.
Finally, just in case GL version string contained garbage, reject
negative version numbers.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
This test ensures that client submitted damage goes to the screen
correctly, regardless of output scale or transform.
The added quirk is explained in the test that uses it.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
For a pbuffer EGLSurface, assume that EGL swap behavior is "preserved"
which means buffer age is always 1 (after the very first
eglSwapBuffers()).
EGL pbuffers are always single-buffered.
Mesa EGL Surfaceless platform does not seem to expose EGL_EXT_buffer_age
that could have told us the same. Hence all repaints to pbuffer surfaces
used to need to repaint the whole output always. This patch makes
repainting only the latest damage possible.
Repainting only the latest damage is required for a future test on
output damage regions: "tests: add output damage test".
Technically, setting buffer_age to 1 is not correct before the very
first eglSwapBuffers(), but to keep the code simpler I chose to rely on
a newly enabled output always having full damage anyway.
Checking that EGL_SWAP_BEHAVIOR is EGL_BUFFER_PRESERVED would do too.
Unfortunately, Mesa seems to return EGL_BUFFER_DESTROYED, so I cannot
fail the headless-backend in that check. Even so, the output damage test
actually succeeds.
See also: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4278
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Require GL_EXT_unpack_subimage unconditionally in GL-renderer. Without
this extension, it would take considerable effort in GL-renderer to
handle correctly images that contain row padding, either as a temporary
copy to remove padding or doing SubImage updates row by row.
I would guess that this path has gone long completely untested, and if
it was exercised, the rows never had padding thanks to 32-bit pixel
formats. Instead of writing tests to poke the corner cases and fixing
it, remove it.
This will make it easier to fix other problems in GL-renderer in this
area in the future - one less path to consider and many restrictions in
GL API gone.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
When there's neither configless nor surfaceless EGL extension
(i.e. not a Mesa driver), Weston falls back to a dummy pbuffer surface.
Weston attempts to find for that surface an EGL config but uses a NULL
array of pixel formats. This fails with the following messages:
EGL_KHR_surfaceless_context unavailable. Trying PbufferSurface
Found an EGLConfig matching { pbf; } but it is not usable because
neither EGL_KHR_no_config_context nor EGL_MESA_configless_context
are supported by EGL.
failed to choose EGL config for PbufferSurface
EGL error state: EGL_SUCCESS (0x3000)
Failed to initialise the GL renderer;
Signed-off-by: Tomek Bury <tomek.bury@broadcom.com>
Using the number of planes to determine if GL_TEXTURE_EXTERNAL_OES should be
used is incorrect with some modifiers: For example RGBA with a
I915_FORMAT_MOD_Y_TILED_CCS modifier has two planes.
Use eglQueryDmaBufModifiersEXT() to query if the current format/modifier only
supports GL_TEXTURE_EXTERNAL_OES.
Use the current code as fallback of modifiers are not supported.
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
gl_rendererer's output_pbuffer_create has a lot of arguments now. Add a
structure for the options to make it more clear what is what.
This is in preparation for adding bare-integer arguments which are ripe
for confusion when passing positional arguments.
Signed-off-by: Daniel Stone <daniels@collabora.com>
gl_rendererer's output_window_create has a lot of arguments now. Add a
structure for the options to make it more clear what is what.
This is in preparation for adding bare-integer arguments which are ripe
for confusion when passing positional arguments.
Signed-off-by: Daniel Stone <daniels@collabora.com>
gl_rendererer's output_create has a lot of arguments now. Add a
structure for the options to make it more clear what is what.
This is in preparation for adding bare-integer arguments which are ripe
for confusion when passing positional arguments.
Signed-off-by: Daniel Stone <daniels@collabora.com>
This is to put more of the EGL client extension handling in the same
place. This also adds a boolean to check if EGL_EXT_platform_base is
supported, similar to other extensions we check.
Signed-off-by: Scott Anderson <scott.anderson@collabora.com>
EGL client extensions are not tied to the EGLDisplay we create, and have
an effect on how we create the EGLDisplay. Since we're using this to
look for EGL_EXT_platform_base, it makes more sense for this to be near
the start of the GL renderer initialization.
Signed-off-by: Scott Anderson <scott.anderson@collabora.com>
If gl-renderer fails its initialisation, we return to compositor
teardown, which will try to free the renderer if ec->renderer was set.
This is unfortunate when we've already torn it down whilst failing
gl-renderer init, so just clear the renderer member so we don't try to
tear down twice.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Reported-by: Emil Velikov <emil.velikov@collabora.com>
Commit adaf8c7410 ("renderer: change frame_signal emission to pass
previous_damage as data argument") missed updating all frame_signal
emissions. Later commit 2619bfe420 ("move frame_signal emission to
weston_output_repaint()") fixed this deficency along with moving the
location of the emission. Due to an issue of the location change, this
commit had to be reverted again.
This makes sure that the pixman as well as the GL renderer now also
emits the damage region instead of the Weston output.
Fixes: adaf8c7410 ("renderer: change frame_signal emission to pass previous_damage as data argument")
Signed-off-by: Stefan Agner <stefan@agner.ch>
Pekka Paalanen
Harish Krupo
Michael Olbrich
Tomek Bury
Daniel Stone
Scott Anderson
Stefan Agner