os/os_threads.h includes c11/threads.h. We can use c11 threads
directly.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Yiwei Zhang <zzyiwei@chromium.org>
Reviewed-by: Ryan Neph <ryanneph@google.com>
Acked-by: Gert Wollny <gert.wollny@collabora.com>
Also, if no VBO is bound bail out of draw_vbo, to avoid trying to
dereference the pointer when filling the shader key.
v2: rename pointer to subcontext to owning_sub (Chia-I)
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: John Bates <jbates@chromium.org>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Some paths don't emit rasterizer state or framebuffer state before calling
do_readpixels, and because of that the clamping is not done properly.
Fixes 0d84775435
Fixes several piglit snorm tests.
Signed-off-by: Italo Nicola <italonicola@collabora.com>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
If the guest is creating texture and the memory
comes from buffer object by creating GL_TEXTURE_BUFFER,
then host creates GL_TEXTURE_BUFFER too.
No texture parameters can be set for GL_TEXTURE_BUFFER.
If there is mismatch between the guest texture format and
the host texture format, for example GL_ALPHA8(guest) and
GL_R8(host), then we can't apply swizzling for such textures.
In such case, add manually swizzling in GLSL shader generation
step.
The logic of this patch:
1. Add additional fields in shader key struct
2. During draw_vbo call check if "manual swizzling" is needed
3. If yes, the add fields in key struct and generate shader again
4. During generation of for example texelFetch instruction
in GLSL put additional instruction for swizzling
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Telegram (Android App) shows high cpu usage under
virgl. It turns out when the guest send blend_mode
via alpha_src_factor, BLEND_SOFTLIGHT (0x9) would be
treated as PIPE_BLENDFACTOR_SRC1_COLOR (0x9). The
dual_src didn't match dual_src_linked and virgl keeps
linking the same program.
Fix the logic to calculate dual_src to make it as close
as possible to calculate dual_src_linked.
Signed-off-by: Lepton Wu <lepton@chromium.org>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Tipically mesa handles CLAMP_READ_COLOR in gallium frontend, and the
default value for CLAMP_READ_COLOR as defined by the GL spec is
GL_FIXED_ONLY.
As CLAMP_READ_COLOR is really only used for reading pixel color values
from a buffer, there's no problem always returning unclamped colors and
letting the guest mesa clamp it if it wants.
On the other hand, if we clamp it on the host, as we do now, then the
guest mesa only has clamped values to work with, which needless to say,
can't be "unclamped".
Unfortunaly, this fix doesn't work for the GL ES backend, because in
GL ES the colors read with glReadPixels are always clamped.
Fixes piglit's fbo-blending-snorm tests.
Signed-off-by: Italo Nicola <italonicola@collabora.com>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
It doesn't end up getting used because the sub_ctx->prog check happens
first, but let's not return a surprising truthy value from one fail
path of a bool function.
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
This crashed for me all the time with nir_to_tgsi triggering compile
failures. The glsl_strings get freed at shader destroy time, no need
to do so here.
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Refactor only. There are a lot of variables named like *_id/*_index
already.
Signed-off-by: Ryan Neph <ryanneph@google.com>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Refactor only. There are a lot of variables named like *_id/*_index
already.
Signed-off-by: Ryan Neph <ryanneph@google.com>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Let the value found in the properties takes preference.
This fixes all compilation and link errors with
KHR-GL43.cull_distance.functional
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Rohan Garg <rohan.garg@collabora.com>
To make the TGSI-GLSL conversion easier to track, keep the original TGSI
text in place when it is to be logged, and print it right before the shader
conversion.
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Rohan Garg <rohan.garg@collabora.com>
Commit 603699f1 ("vrend: Cleanup and fix bug in next_sampler_id
tracking") fixed a rendering issue in Factorio but inadvertently changed
the indexing behavior for shadow samplers in some related code cleaning.
shadow_samp_*_locs arrays are currently indexed according to the bit
position of a sampler in samplers_used_mask and shadow_samp_mask, so
sampler_index must be incremented for each sampler, not just for each
shadow sampler.
Fixes: 603699f1 ("vrend: Cleanup and fix bug in next_sampler_id tracking")
Signed-off-by: Ryan Neph <ryanneph@google.com>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
A debug message is emitted whenever sampler states are applied and a
NULL state pointer is encountered. NULL pointers indicate that the
default state should be used instead, so there's no need to spam with
error messages approx. 1+ times per draw-call.
commit 405a50d ("vrend: cleanup vrend_bind_sampler_states") adds a more
useful debug message at sampler state bind-time when handle is non-zero
and corresponding state object cannot be located.
Fixes: 32c733f ("src/: replace all instances of "fprintf to stderr" with "vrend_printf"")
Signed-off-by: Ryan Neph <ryanneph@google.com>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Discovered while investigating broken floor tile rendering in Factorio
(game). Skipping increment of next_sampler_id, regardless of dirty
status, results in collision with caller's use of next_sampler_id for
additional samplers.
Closes: https://gitlab.freedesktop.org/virgl/virglrenderer/-/issues/227
Fixes: c08c8419 ("vrend: use helper pointers in draw_bind_samplers_shader")
Signed-off-by: Ryan Neph <ryanneph@google.com>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Cleaner to modify ctx->sub->sampler_views_dirty directly.
Added error message for failed sampler state binds.
Signed-off-by: Ryan Neph <ryanneph@google.com>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
refactor vrend_shader_info use to constant and reduce some variable
scope.
Signed-off-by: Ryan Neph <ryanneph@google.com>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Up until now sme of the info stored in sinfo was changed
based on the shader key, and since this info is used for
all shader variants when evaluating the key for shaders
to be linked, this may result in the wrong shader
variant being picked up.
Instead track the shader info that can be changed like this
in the shader variant itself.
Closes: #239
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Tested-by: Dave Airlie <airlied@redhat.com>
Fixes e5fabecbe0
vrend: pass texture levels per shader on GLES as uniform
v2: - only emit texlod uniform when textureQueryLevels is called
- initialize uniform location to -1 if the mask shows that
the texture lod levels arenot needed and ony test this when
uploading the data
(all suggested in a way by Chia-I Wu)
Closes: #237
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Fixes a missed internalformat fixup when glTextureView() is used to create
vrend_surfaces for resources backed by shared EGL images without an
alpha channel (e.g. RGBX, BGRX). Such resources are registered in host
Mesa with internalformat GL_RGB8, but virglrenderer prefers to create
32bpp textures and do it's own swizzling.
fixes: commit ebb2cf3 "vrend: convert linear color to srgb for 24bpp
imported EGL resources".
Signed-off-by: Ryan Neph <ryanneph@google.com>
Although anisotropic filtering uses a float value, most hw doesn't
care below integer levels, so just transmit 1-16 for the sampler value
not the full float.
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Resources imported from dma-bufs as EGL images are given internal format
GL_RGB8 (24bpp) by host mesa. When placing srgb texture views on such
resources, the appropriate view format would be GL_SRGB8, but this is
not color-renderable for GL or GLES and results in an incomplete
framebuffer. Instead, GL_RGB8 must be used as the attachment format and
linear->srgb colorspace conversions must be applied manually in the frag
shader and for calls to glClearColor() involving these surfaces.
This resolves a
dEQP-EGL.functional.wide_color.window_888_colorspace_srgb test failure
for GBM-enabled hosts with newer host mesa, which imports
B8G8R8X8_UNORM and R8G8B8X8_UNORM as 24bpp instead of the more
compatible 32bpp.
Signed-off-by: Ryan Neph <ryanneph@google.com>
Reviewed-by: Lepton Wu <lepton@chromium.org>
Reviewed-by: John Bates <jbates@chromium.org>
For fragment ARB programs, the fog co-ordinate input
can be requested independently of any other part of the
graphics pipeline. Through the use of an OPTION declaration.
This can lead to a case where a fragment shader expects
a fog co-ordinate input, but it is not being supplied
by the preceding shader stage.
This patch will fix it by forcing it to 0 when it detects
the case, which should neutralize the fog color.
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Compute shader don't have in- or outputs, so don't try to
sync the interfaces with the shaders that are still bound.
This fixes a state leak where non-compute shaders are still
bound when a compute shader is compiled.
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Lepton Wu <lepton@chromium.org>
As suggested by Lepton
Fixes 57a4e9498a
vrend: get rid of glShaderStorageBlockBinding
Closes: #229
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Lepton Wu <lepton@chromium.org>
It indicates that fences should be retired directly from the
sync thread when enabled.
Signed-off-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Ryan Neph <ryanneph@google.com>
Once we retire fences directly in the sync thread, we will also need
to check queries in the same loop as virgl_renderer_poll() will no
longer be called.
In order to do that safely, we need to lock around operations on
the query waiting list. We also need to keep track of the current
context for the sync thread given we need to update it before
checking query status.
Signed-off-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Ryan Neph <ryanneph@google.com>
Since GLES doesn't support QueryTextureLevels in shaders we have
to collect this info and pass it as uniform to the shader.
v2: use pipe_shader_to_prefix
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Rohan Garg <rohan.garg@collabora.com>
Seperate more elements into the shader stages that they are used in
and evaluate them only for the stage they are relevant for.
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Rohan Garg <rohan.garg@collabora.com>
There is no reason why this value should be obtained using a return
parameter. Also fix the version to return zero when the version string
doesn't follow the standard.
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Rohan Garg <rohan.garg@collabora.com>
Mesa clamps the number of cbufs to eight, so we can use uint8_t bitmask for
cbufs states. Be save for the future by only reporting support for
at most eight cbufs.
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Rohan Garg <rohan.garg@collabora.com>
Compress the structure so that it fits into a 64 bit value.
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Rohan Garg <rohan.garg@collabora.com>
In addition combine values into a bitfield so that the compiler can use a
64 bit move instead of individual moves.
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Rohan Garg <rohan.garg@collabora.com>
Keeping all the information in one place might come in handy when we
want to further refactor this. On the way also compress the structure.
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Rohan Garg <rohan.garg@collabora.com>
Use a bitfield to declare vrend_interp_info and make it a fixed size array.
One one hand this avoids all the hassles with allocating and freeing memory,
and it will make it possible to shrink the size of the data that is passed
from the sinfo to the shader key.
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Rohan Garg <rohan.garg@collabora.com>
- don't write zeros, the memory is initialized to zero anyway
- reorder evaluation to check whether it is a FS only once.
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Rohan Garg <rohan.garg@collabora.com>
The property doesn't change. so it doesn't make sense to set it in the
shader key.
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Rohan Garg <rohan.garg@collabora.com>
Caught by fuzzer. surf[0]->texture cannot be assumed to be a valid
pointer.
Signed-off-by: Ryan Neph <ryanneph@google.com>
Reviewed-by: Yiwei Zhang <zzyiwei@chromium.org>
On GLES it is not supported, and on GL it seems to be irrelevant,
because the guest already handles this.
Update the test expectation accordingly.
Fixes#201
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Italo Nicola <italonicola@collabora.com>
Chia-I Wu
Gert Wollny
Italo Nicola
mwezdeck
Lepton Wu
Emma Anholt
Ryan Neph
Dave Airlie
Isaac Bosompem
Louis-Francis Ratté-Boulianne