OSDN Git Service
Lars Hamre [Thu, 26 May 2016 22:30:24 +0000 (18:30 -0400)]
tgsi: use truncf in micro_trunc
Switches to using truncf in micro_trunc.
Fixes the following piglit tests (for softpipe):
/spec/glsl-1.30/execution/built-in-functions/...
fs-trunc-float
fs-trunc-vec2
fs-trunc-vec3
fs-trunc-vec4
vs-trunc-float
vs-trunc-vec2
vs-trunc-vec3
vs-trunc-vec4
/spec/glsl-1.50/execution/built-in-functions/...
gs-trunc-float
gs-trunc-vec2
gs-trunc-vec3
gs-trunc-vec4
Signed-off-by: Lars Hamre <chemecse@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Samuel Iglesias Gonsálvez [Fri, 27 May 2016 09:59:48 +0000 (11:59 +0200)]
i965/gs/scalar: Fix load input for doubles
Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Samuel Iglesias Gonsálvez [Thu, 26 May 2016 05:56:37 +0000 (07:56 +0200)]
i965/fs: fix offset when loading double vector input varyings
When we are not packing a double input varying, we might need to
read its data in a non-aligned to 64-bit offset, so we read
the wrong data. This is happening when using explicit locations
in varyings because Mesa disables packing varying for that case.
const_index is in 32-bit size units but offset() is multiplying
it by destination type size units. When operating with double
input varyings, const_index value could be not aligned to 64 bits.
To fix it, we load the double vector as if it was a float based vector
with twice the number of components.
Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Samuel Iglesias Gonsálvez [Thu, 26 May 2016 05:56:38 +0000 (07:56 +0200)]
i965/fs: fix FS_OPCODE_CINTERP for unpacked double input varyings
Data starts at suboffet 3 in 32-bit units (12 bytes), so it is not
64-bit aligned and the current implementation fails to read the data
properly. Instead, when there is is a double input varying, read it as
vector of floats with twice the number of components.
Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Dave Airlie [Fri, 3 Jun 2016 00:45:07 +0000 (10:45 +1000)]
glsl: geom shader max_vertices layout must match.
From GLSL 4.5 spec, "4.4.2.3 Geometry Outputs".
"all geometry shader output vertex count declarations in a
program must declare the same count."
Fixes:
GL45-CTS.geometry_shader.output.conflicted_output_vertices_max
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Cc: "11.2 12.0" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Jason Ekstrand [Sat, 4 Jun 2016 22:10:22 +0000 (15:10 -0700)]
anv/pipeline: Add support for caching the push constant map
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
Dave Airlie [Wed, 11 May 2016 00:49:19 +0000 (10:49 +1000)]
glsl: use enum glsl_interface_packing in more places. (v2)
Although the glsl_types.h stores this in a bitfield,
we should hide that from everyone else. Hide the cast
in an accessor method and use the enum everywhere.
This makes things a bit nicer in gdb, and improves type
safety.
v2: fix a few pieces of interface I missed that caused some
piglit regressions.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Dave Airlie [Fri, 3 Jun 2016 01:36:38 +0000 (11:36 +1000)]
i965: don't use NumLayers for 3D textures.
For 3D textures we shouldn't be using NumLayers, we need
to get it from the depth.
This fixes:
GL45-CTS.geometry_shader.layered_framebuffer.clear_call_support
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Mon, 6 Jun 2016 00:33:51 +0000 (10:33 +1000)]
glsl: for anonymous struct matching use without_array() (v3)
With tessellation shaders we can have cases where we have
arrays of anon structs, so make sure we match using without_array().
Fixes:
GL45-CTS.tessellation_shader.tessellation_control_to_tessellation_evaluation.gl_in
v2:
test lengths match as well (Ilia)
v3:
descend array lengths to check for matches as well (Ilia)
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Tue, 3 May 2016 04:39:06 +0000 (14:39 +1000)]
glsl/ast: don't crash when func_name is NULL
This fixes a crash in
GL43-CTS.shader_subroutine.subroutines_not_allowed_as_variables_constructors_and_argument_or_return_types
If we can't find the func_name in one of these paths,
we have emitted an earlier error so just return here.
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
Cc: "11.2 12.0" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Tue, 3 May 2016 07:16:27 +0000 (17:16 +1000)]
glsl: handle ast_aggregate in has_sequence_subexpression. (v2)
GL43-CTS.compute_shader.work-group-size does
uniform uint g_uniform[gl_WorkGroupSize.z + 20] = { 1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24 };
The initializer triggers the GLSL 4.30/GLES3 tests
for constant sequence subexpressions, so it doesn't
happen unless you are using those, so just return
false as this path is now reachable.
v2: update commit msg with diagnosis
Acked-by: Timothy Arceri <timothy.arceri@collabora.com>
Cc: "11.2 12.0" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Kenneth Graunke [Sun, 5 Jun 2016 23:31:11 +0000 (16:31 -0700)]
mesa: Try to unbreak the MSVC build.
PATH_MAX is apparently not a thing on Windows. Borrow the hack from
pipe_loader.c to try and make this work.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Kenneth Graunke [Sun, 7 Sep 2014 03:26:51 +0000 (20:26 -0700)]
mesa: Add MESA_SHADER_CAPTURE_PATH for writing .shader_test files.
This writes linked shader programs to .shader_test files to
$MESA_SHADER_CAPTURE_PATH in the format used by shader-db
(http://cgit.freedesktop.org/mesa/shader-db).
It supports both GLSL shaders and ARB programs. All stages that
are linked together are written in a single .shader_test file.
This eliminates the need for shader-db's split-to-files.py, as Mesa
produces the desired format directly. It's much more reliable than
parsing stdout/stderr, as those may contain extraneous messages, or
simply be closed by the application and unavailable.
We have many similar features already, but this is a bit different:
- MESA_GLSL=dump writes to stdout, not files.
- MESA_GLSL=log writes each stage to separate files (rather than
all linked shaders in one file), at draw time (not link time),
with uniform data and state flag info.
- Tapani's shader replacement mechanism (MESA_SHADER_DUMP_PATH and
MESA_SHADER_READ_PATH) also uses separate files per shader stage,
but allows reading in files to replace an app's shader code.
v2: Dump ARB programs too, not just GLSL.
v3: Don't dump bogus 0.shader_test file.
v4: Add "GL_ARB_separate_shader_objects" to the [require] block.
v5: Print "GLSL 4.00" instead of "GLSL 4.0" in the [require] block.
v6: Don't hardcode /tmp/mesa.
v7: Fix memoization of getenv().
v8: Also print "SSO ENABLED" (suggested by Timothy).
v9: Also handle ES shaders (suggested by Ilia).
v10: Guard against MESA_SHADER_CAPTURE_PATH being too long; add
_mesa_warning calls on error handling (suggested by Ben).
v11: Fix crash when variable is unset introduced in v10.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Ilia Mirkin [Sun, 5 Jun 2016 19:00:36 +0000 (15:00 -0400)]
nv50,nvc0: fix BGR10_A2UI vertex format
This is mostly academic as this is not reachable from GL, which only has
the packed RGB10_A2UI vertex format.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Samuel Pitoiset [Sun, 5 Jun 2016 16:53:26 +0000 (18:53 +0200)]
nvc0: do not clear surfaces bins in the validate function
We should not call nouveau_bufctx_reset() inside a validate function.
This only affects Fermi where images are aliased between 3D and CP.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Samuel Pitoiset [Sun, 5 Jun 2016 16:01:19 +0000 (18:01 +0200)]
nvc0: re-validate images after launching a grid on Fermi
Images invalidation is a bit weird on Fermi and there is already a hack
which forces invalidating all images when launching a computer shader
to help in fixing 3D<->CP interaction.
However, we need to re-validate images for compute because
nvc0_compute_invalidate_surfaces() will destroy the previous binding.
This is not really good for performance purposes but this might be
improved later.
This fixes the following piglits:
- spec/arb_compute_shader/execution/basic-uniform-access
- spec/arb_compute_shader/execution/mutiple-texture-reading
- spec/arb_compute_shader/execution/multiple-workgroups
- spec/glsl-4.30/execution/built-in-functions/cs-* (207 tests)
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Marek Olšák [Fri, 3 Jun 2016 17:17:46 +0000 (19:17 +0200)]
radeonsi: fix images with level > 0
This should fix spec@arb_shader_image_load_store@level.
Broken by:
Commit:
95c5bbae66af3ca1f805d94f6fe8d8e4ba2c9c43
radeonsi: set some image descriptor fields at bind time
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Ilia Mirkin [Sat, 4 Jun 2016 18:13:38 +0000 (14:13 -0400)]
nvc0: reduce overhead from always marking images dirty
We would revalidate images when anything was touched at all. Which is
unfortunate, since the state tracker does not use CSO's to reduce the
workload. So instead implement a protocol to ensure that something has
changed before revalidating all the images.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Ilia Mirkin [Sat, 4 Jun 2016 17:50:21 +0000 (13:50 -0400)]
nvc0: reduce overhead from always marking buffers dirty
We would revalidate buffers when anything was touched at all. Which is
unfortunate, since the state tracker does not use CSO's to reduce the
workload. So instead implement a protocol to ensure that something has
changed before revalidating all the SSBOs.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Ilia Mirkin [Fri, 3 Jun 2016 01:36:04 +0000 (21:36 -0400)]
nvc0: fix memory barrier flag handling
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Ilia Mirkin [Fri, 3 Jun 2016 01:42:14 +0000 (21:42 -0400)]
nvc0: mark bound buffer range valid
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Dave Airlie [Sat, 4 Jun 2016 20:49:42 +0000 (06:49 +1000)]
anv/entrypoints: don't go using wayland/xcb unless they are configured
The fix in:
anv: let anv_entrypoints_gen.py generate proper Wayland/Xcb guards
breaks things if wayland headers aren't installed.
Separate things out properly to avoid that problem.
[airlied: fixed up to put in pre-existing sections].
Reported-by: Arjan van de Ven
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Marek Olšák [Thu, 26 May 2016 16:20:42 +0000 (18:20 +0200)]
gallium/radeon: don't use the DMA ring for pipelined buffer uploads
Submitting a DMA IB flushes the GFX IB and all GPU caches.
Vedran Miletić said:
"On Tonga 380X, this improves The Talos Principle from 8.3 fps to 28.3 fps
(all graphics settings Ultra, 4xAA, 1080p resolution with downsampling
from 1200p)."
Some anonymous dude said:
R9 390 results:
Tomb Raider (normal settings): 80 -> 88 FPS
Talos Principle (custom settings): 23 -> 56 FPS
Metro Last Light Redux (default benchmark settings): 39 -> 40 FPS
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Tested-by: Vedran Miletić <vedran@miletic.net>
Tested-by: Grazvydas Ignotas <notasas@gmail.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Thu, 26 May 2016 16:14:27 +0000 (18:14 +0200)]
r600g: don't flush caches when binding shader resources
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Tested-by: Grazvydas Ignotas <notasas@gmail.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Thu, 26 May 2016 15:25:46 +0000 (17:25 +0200)]
r600g: only do necessary cache flushes in cp_dma_copy_buffer
The main impact is that {upload, draw, upload, draw, ..} doesn't flush
framebuffer caches before every upload.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Tested-by: Grazvydas Ignotas <notasas@gmail.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Thu, 26 May 2016 15:18:13 +0000 (17:18 +0200)]
r600g: only do necessary cache flushes in cp_dma_clear_buffer
The main impact is that fast color clear doesn't flush TC, CONST, DB.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Tested-by: Grazvydas Ignotas <notasas@gmail.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Wed, 1 Jun 2016 16:39:53 +0000 (18:39 +0200)]
r600g: remove a CP DMA workaround that's not needed anymore
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Tested-by: Grazvydas Ignotas <notasas@gmail.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Thu, 26 May 2016 20:00:03 +0000 (22:00 +0200)]
r600g: fix CP DMA hazard with index buffer fetches (v3)
v3: use PFP_SYNC_ME on EG-CM only when supported by the kernel,
otherwise use MEM_WRITE + WAIT_REG_MEM to emulate that
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Tested-by: Grazvydas Ignotas <notasas@gmail.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Wed, 1 Jun 2016 16:35:33 +0000 (18:35 +0200)]
r600g: properly sync CP with CP DMA on R6xx
This will allow removing useless cache & IB flushes.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Tested-by: Grazvydas Ignotas <notasas@gmail.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Tue, 31 May 2016 21:07:15 +0000 (23:07 +0200)]
r600g: write WAIT_UNTIL in the correct place
This has been wrong all along. Fixing this will allow removing useless
cache flushes.
Cc: 11.1 11.2 12.0 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Tested-by: Grazvydas Ignotas <notasas@gmail.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Tue, 31 May 2016 17:11:54 +0000 (19:11 +0200)]
gallium/radeon: rename allocator_so_filled_size -> allocator_zeroed_memory
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Tested-by: Grazvydas Ignotas <notasas@gmail.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Tue, 31 May 2016 17:06:45 +0000 (19:06 +0200)]
gallium/u_suballoc: allow different alignment for each allocation
Just move the alignment parameter from u_suballocator_create
to u_suballocator_alloc.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Tested-by: Grazvydas Ignotas <notasas@gmail.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Jason Ekstrand [Thu, 2 Jun 2016 23:34:11 +0000 (16:34 -0700)]
anv/blit: Use CLAMP_TO_EDGE for scaled blits
When upscaling you can end up interpolating between the edge pixel and one
past the edge. Using CLAMP_TO_EDGE seems like the most reasonable thing to
do in this case. This fixes two of the new Vulkan CTS tests in
dEQP-VK.api.copy_and_blit.blit_image.*
Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Thu, 2 Jun 2016 23:25:44 +0000 (16:25 -0700)]
anv/copy: Account for the anv_surface.offset when creating a blit2d_surf
This was causing problems if the user tried to copy to/from the stencil
portion of a combined depth/stencil image.
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Thu, 2 Jun 2016 21:36:58 +0000 (14:36 -0700)]
nir/spirv: Make a decoration switch complete
Getting rid of the default case makes the compiler warn if we are missing
cases. While we're here, we also add the one missing case.
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Thu, 2 Jun 2016 21:34:15 +0000 (14:34 -0700)]
nir/spirv: Make unhandled decorations and capabilities non-fatal
glslang frequently throw bogus decorations into shaders. While we are free
to assert-fail, it's a bit nicer to the application to just warn.
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Thu, 2 Jun 2016 21:32:56 +0000 (14:32 -0700)]
nir/spirv: Add a way to print non-fatal warnings
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Thu, 2 Jun 2016 21:06:30 +0000 (14:06 -0700)]
nir/spirv: Add string lookup tables for a couple of SPIR-V enums
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Thu, 2 Jun 2016 20:43:19 +0000 (13:43 -0700)]
nir/spirv: Complete the list of capabilities
Previously we supported a subset of capabilities and just left a default
case for the others. It's time to stop being lazy and actually audit the
capabilities. This should bring them up-to-date with reality.
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Wed, 1 Jun 2016 03:16:01 +0000 (20:16 -0700)]
anv/pipeline: Add support for early depth stencil
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Thu, 2 Jun 2016 01:53:32 +0000 (18:53 -0700)]
mesa: Get rid of _mesa_active_fragment_shader_has_side_effects
It is no longer used.
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Jason Ekstrand [Thu, 2 Jun 2016 01:55:35 +0000 (18:55 -0700)]
i965/ps_state: Use wm_prog_data.has_side_effects
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Jason Ekstrand [Thu, 2 Jun 2016 01:46:30 +0000 (18:46 -0700)]
i965/fs Add a wm_prog_data bit for has_side_effects
This is more accurate than calling
_mesa_active_fragment_shader_has_side_effects because it looks at whether
or not the SSBOs, images, or atomic buffers are actually written rather
than just existing in the program.
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Thu, 2 Jun 2016 01:29:09 +0000 (18:29 -0700)]
nir/info: Get rid of uses_interp_var_at_offset
We were using this briefly in the i965 driver to trigger recompiles but we
haven't been using it since we switched to the NIR y-transform lowering
pass.
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Jason Ekstrand [Wed, 1 Jun 2016 05:23:18 +0000 (22:23 -0700)]
anv/pipeline: Silently pass tests if depth or stencil is missing
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Wed, 1 Jun 2016 05:19:53 +0000 (22:19 -0700)]
anv/pipeline: Unify gen7/8 emit_ds_state
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Wed, 1 Jun 2016 05:15:38 +0000 (22:15 -0700)]
genxml/gen6,7,75: s/BackFace/Backface
This is more consistent with gen8+
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Wed, 1 Jun 2016 18:20:22 +0000 (11:20 -0700)]
nir/spirv: Handle the WorkgroupSize builtin decoration
This fixes the 7 dEQP-VK.pipeline.spec_constant.compute.local_size.* tests
in the latest dev version of the Vulkan CTS.
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Wed, 1 Jun 2016 17:34:04 +0000 (10:34 -0700)]
nir/spirv: Use breaks instead of returns in constant handling
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Tue, 31 May 2016 23:27:19 +0000 (16:27 -0700)]
anv/pipeline: Refactor specialization constant handling a bit
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Tue, 31 May 2016 22:02:10 +0000 (15:02 -0700)]
nir/lower_indirect_derefs: Use the direct array deref for recursion
This fixes about 100 of the new Vulkan CTS tests.
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Tue, 31 May 2016 18:26:06 +0000 (11:26 -0700)]
anv/clear: Handle ClearImage on 3-D images
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Francisco Jerez [Fri, 3 Jun 2016 19:32:15 +0000 (12:32 -0700)]
Revert "i965/fs: Allow scalar source regions on SNB math instructions."
This reverts commit
c1107cec44ab030c7fcc97c67baa12df1cc9d7b5.
Apparently the hardware spec text I quoted in the commit message was
outright lying about scalar source math being supported on SNB, the
hardware seems to load 32 contiguous bits of data for each channel
regardless of the regioning mode. Fixes regressions in the following
CTS tests (which we didn't catch early due to CTS being temporarily
disabled in our CI system):
es2-cts.gtf.gl.atan.atan_vec3_frag_xvary
es2-cts.gtf.gl.cos.cos_vec2_frag_xvary
es2-cts.gtf.gl.atan.atan_vec2_frag_xvary
es2-cts.gtf.gl.pow.pow_vec2_frag_xvary_yconsthalf
es2-cts.gtf.gl.cos.cos_float_frag_xvary
es2-cts.gtf.gl.pow.pow_float_frag_xvary_yconsthalf
es2-cts.gtf.gl.atan.atan_vec3_frag_xvaryyvary
es2-cts.gtf.gl.pow.pow_vec3_frag_xvary_yconsthalf
es2-cts.gtf.gl.cos.cos_vec3_frag_xvary
es2-cts.gtf.gl.atan.atan_vec2_frag_xvaryyvary
Cc: mesa-stable@lists.freedesktop.org
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=96346
Reported-by: Mark Janes <mark.a.janes@intel.com>
Acked-by: Matt Turner <mattst88@gmail.com>
Francisco Jerez [Wed, 1 Jun 2016 23:27:52 +0000 (16:27 -0700)]
i965/vec4: Fix cmod propagation not to propagate non-identity cmod into CMP(N).
The conditional mod of these instructions determines the semantics of
the comparison itself (rather than being evaluated based on the result
of the instruction as is usually the case for most other instructions
that allow conditional mods), so it's in general not legal to
propagate a conditional mod into a CMP instruction. This prevents
cmod propagation from (mis)optimizing:
cmp.z.f0 tmp, ...
mov.z.f0 null, tmp
into:
cmp.z.f0 tmp, ...
which gives the negation of the flag result of the original sequence.
I originally noticed this while working on SIMD32 in the scalar
back-end, but the same scenario is likely to be possible in vec4
programs so this commit ports the bugfix with the same name from the
scalar back-end to the vec4 cmod propagation pass.
Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Emil Velikov [Fri, 3 Jun 2016 23:20:53 +0000 (00:20 +0100)]
anv: add the X related and Wayland CFLAGS to VULKAN_ENTRYPOINT_CPPFLAGS
Otherwise we will fail to find the headers in some scenarios.
Cc: <mesa-stable@lists.freedesktop.org>
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reported-by: Tobias Klausmann <tobias.johannes.klausmann@mni.thm.de>
Tested-by: Tobias Klausmann <tobias.johannes.klausmann@mni.thm.de>
Reviewed-by: Tobias Klausmann <tobias.johannes.klausmann@mni.thm.de>
Emil Velikov [Fri, 3 Jun 2016 23:18:40 +0000 (00:18 +0100)]
nir: automake: add nir_search_helpers.h to the sources list(s)
Fixes:
dfbae7d64f4 ("nir/algebraic: support for power-of-two
optimizations")
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Rob Clark [Mon, 9 May 2016 16:41:00 +0000 (12:41 -0400)]
freedreno/ir3: do idiv lowering after main opt loop
Give algebraic-opt pass a chance to catch udiv by const power-of-two,
before running lower-idiv pass.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Sat, 7 May 2016 17:01:24 +0000 (13:01 -0400)]
nir/algebraic: support for power-of-two optimizations
Some optimizations, like converting integer multiply/divide into left/
right shifts, have additional constraints on the search expression.
Like requiring that a variable is a constant power of two. Support
these cases by allowing a fxn name to be appended to the search var
expression (ie. "a#32(is_power_of_two)").
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Nicolai Hähnle [Thu, 2 Jun 2016 20:17:40 +0000 (22:17 +0200)]
radeonsi: mark buffer texture range valid for shader images
When a shader image view into a buffer texture can be written to, the buffer's
valid range must be updated, or subsequent transfers may incorrectly skip
synchronization.
This fixes a bug that was exposed in Xephyr by PBO acceleration for glReadPixels,
reported by Michel Dänzer.
Cc: Michel Dänzer <michel.daenzer@amd.com>
Cc: 12.0 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Marek Olšák [Fri, 3 Jun 2016 09:25:19 +0000 (11:25 +0200)]
Revert "egl: Check if API is supported when using eglBindAPI."
This reverts commit
e8b38ca202fbe8c281aeb81a4b64256983f185e0.
It broke Glamor for Gallium at least.
Alejandro Piñeiro [Fri, 6 May 2016 14:13:26 +0000 (16:13 +0200)]
mesa/formatquery: expand NUM_SAMPLE_COUNTS OpenGL ES comment
For ES 3.0 NUM_SAMPLE_COUNTS spec points that some formats will be
always zero. But on ES 3.1 can be different to zero.
The current code is correctly checking exactly against version 3.0,
but the comment only mentions 3.0 spec. It is clearer mentioning both.
v2: better wording on the comment (Ian Romanick)
Acked-by: Eduardo Lima <elima@igalia.com>
Acked-by: Antia Puentes <apuentes@igalia.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Dave Airlie [Fri, 3 Jun 2016 02:26:05 +0000 (12:26 +1000)]
mesa/get: return correct value for layer provoking vertex.
This fixes:
GL45-CTS.geometry_shader.layered_rendering.layered_rendering
on Skylake.
Reviewed-by: Chris Forbes <chrisforbes@google.com>
Cc: "11.2 12.0" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Plamena Manolova [Wed, 1 Jun 2016 16:31:29 +0000 (17:31 +0100)]
egl: Account for default values of texture target and format
When validating attributes during surface creation we should account
for the default values of texture target and format (EGL_NO_TEXTURE)
since the user is not obligated to explicitly set both via the
attribute list passed to eglCreatePbufferSurface.
Signed-off-by: Plamena Manolova <plamena.manolova@intel.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Samuel Pitoiset [Thu, 2 Jun 2016 22:00:27 +0000 (00:00 +0200)]
nvc0: mark buffer texture range valid for shader images
Loosely based on radeonsi (Thanks to Nicolai).
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: 12.0 <mesa-stable@lists.freedesktop.org>
Mauro Rossi [Thu, 2 Jun 2016 19:15:35 +0000 (21:15 +0200)]
isl: add support for Android libmesa_isl static library
isl library is needed to build i965, libmesa_isl static library is added
to fix related Android building errors.
Any attempt to build libmesa_genxml as phony package module failed to deliver
gen{7,75,8,9}_pack.h generated headers, needed for libmesa_isl_gen{7,75,8,9}
Due to constraints in Android Build System, libmesa_genxml is built as static,
at least one source is needed, so dummy.c is autogenerated for this scope,
libmesa_genxml dependency is declared using LOCAL_WHOLE_STATIC_LIBRARIES,
to avoid building errors due to missing genxml/gen{7,75,8,9}_pack.h headers.
Cc: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Mauro Rossi [Mon, 30 May 2016 22:20:28 +0000 (00:20 +0200)]
android: libmesa_glsl: add a dependency on libmesa_nir static
Fixes the following building error:
target C++: libmesa_glsl <= external/mesa/src/compiler/glsl/glsl_to_nir.cpp
In file included from external/mesa/src/compiler/glsl/glsl_to_nir.h:28:0,
from external/mesa/src/compiler/glsl/glsl_to_nir.cpp:28:
external/mesa/src/compiler/nir/nir.h:42:25: fatal error: nir_opcodes.h: No such file or directory
compilation terminated.
build/core/binary.mk:432: recipe for target 'out/target/product/x86/obj/STATIC_LIBRARIES/libmesa_glsl_intermediates/glsl/glsl_to_nir.o' failed
make: *** [out/target/product/x86/obj/STATIC_LIBRARIES/libmesa_glsl_intermediates/glsl/glsl_to_nir.o] Error 1
make: *** Waiting for unfinished jobs....
Cc: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Emil Velikov [Tue, 31 May 2016 15:59:39 +0000 (16:59 +0100)]
isl: automake: don't include isl_format_layout.c in two lists.
Including the file in both ISL_FILES and ISL_GENERATED_FILES makes
the actual dependency list less obvious.
v2: Drop unrelated vulkan hunk (Jason).
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Emil Velikov [Tue, 31 May 2016 13:46:19 +0000 (14:46 +0100)]
automake: bring back the .PHONY git_sha1.h.tmp rule
With earlier commit
3689ef32afd ("automake: rework the git_sha1.h rule,
include in tarball") we/I erroneously removed the PHONY rule and the
temporary file.
The former is used to ensure that the header is regenerated when on each
make invocation, while the latter helps us avoid the unneeded rebuild(s)
when the SHA1 hasn't changed.
Reported-by: Grazvydas Ignotas <notasas@gmail.com>
Tested-by: Grazvydas Ignotas <notasas@gmail.com>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Kenneth Graunke [Thu, 2 Jun 2016 00:32:55 +0000 (17:32 -0700)]
i965: Add _NEW_POINT to a couple of comments.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Charmaine Lee [Tue, 31 May 2016 23:33:52 +0000 (16:33 -0700)]
svga: allow copy box in svga_transfer_dma_band()
Instead of just allow copy of a rectangle in svga_transfer_dma_band(),
this patch allows it to copy a box, hence allows copy a 3d texture
in one transfer.
Fixes black screen in running Heaven after commit
fb9fe35. (Bug
1663282)
Tested with Heaven, glretrace, piglit.
Reviewed-by: Sinclair Yeh <syeh@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Rob Clark [Thu, 2 Jun 2016 20:23:36 +0000 (16:23 -0400)]
freedreno: fix bad bitshift warnings
Coverity doesn't realize idx will never be negative. Throw in some
assert()s to help it out.
(Hopefully assert() isn't getting compiled out for coverity build.. but
there seems to be just one way to find out. We might have to change
these to assume())
Fixes CID
1362442,
1362443
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Thu, 2 Jun 2016 20:17:16 +0000 (16:17 -0400)]
freedreno: assume builtin shaders do compile
Maybe we should switch to ureg to build the builtin shaders. But at any
rate, if they fail to compile it is because someone messed them up (or
changed TGSI syntax?).
CID
1362444
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Francisco Jerez [Fri, 27 May 2016 08:02:19 +0000 (01:02 -0700)]
i965/fs: Reindent emit_zip().
Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Francisco Jerez [Fri, 27 May 2016 07:45:04 +0000 (00:45 -0700)]
i965/fs: Skip SIMD lowering destination zipping if possible.
Skipping the temporary allocation and copy instructions is easy (just
return dst), but the conditions used to find out whether the copy can
be optimized out safely without breaking the program are rather
complex: The destination must be exactly one component of at most the
execution width of the lowered instruction, and all source regions of
the instruction must be either fully disjoint from the destination or
be aligned with it group by group.
v2: Don't handle partial source-destination overlap for simplicity
(Jason). No instruction count regressions with respect to v1 in
either shader-db or the few FP64 shader_runner test-cases with
partial overlap I've checked manually.
Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Anuj Phogat [Thu, 2 Jun 2016 18:05:44 +0000 (11:05 -0700)]
blorp: Fix 16x multisample scaled blits
Piglit test ext_framebuffer_multisample_blit_scaled-blit-scaled
(with added 16x sample support) now passes with this patch.
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Anuj Phogat [Tue, 31 May 2016 17:57:03 +0000 (10:57 -0700)]
meta: Fix indentation in shader code
Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Acked-by: Matt Turner <mattst88@gmail.com>
Dave Airlie [Thu, 2 Jun 2016 04:13:18 +0000 (14:13 +1000)]
mesa/copyimage: report INVALID_VALUE for missing cube face
The specs says INVALID_VALUE for exceeding dimensions,
which is really what is happening here.
This fixes:
GL45-CTS.copy_image.non_existent_mipmap
Cc: "11.2 12.0" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Antia Puentes <apuentes@igalia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Thu, 2 Jun 2016 03:41:28 +0000 (13:41 +1000)]
mesa/copyimage: fix num samples check to handle renderbuffers.
This test was only happening for textures, but there is
nothing in the spec to say this, so test it for all cases.
This fixes:
GL45-CTS.copy_image.invalid_target
Cc: "11.2 12.0" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Rob Clark [Thu, 2 Jun 2016 15:47:11 +0000 (11:47 -0400)]
freedreno/a4xx: silence coverity warning
CID
1362451
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Thu, 2 Jun 2016 15:42:25 +0000 (11:42 -0400)]
freedreno/a3xx+a4xx: fix potential null ptr deref
Coverity spotted the a3xx case (not sure why not the a4xx).
CID
1362452
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Thu, 2 Jun 2016 15:19:43 +0000 (11:19 -0400)]
freedreno/ir3: fix coverity warning
CID
1362453
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Thu, 2 Jun 2016 15:13:26 +0000 (11:13 -0400)]
freedreno/ir3: use nir_shader_get_entrypoint() helper
Should also fix coverity warning: CID
1362454
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Thu, 2 Jun 2016 14:49:18 +0000 (10:49 -0400)]
freedreno/a4xx: fix incorrect enum type
a4xx has it's own enum, different from a2xx/a3xx.
Spotted by coverity: CID
1362458,
1362459
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Thu, 2 Jun 2016 14:36:23 +0000 (10:36 -0400)]
freedreno: fix coverity negative array index warning
Never can happen, since query would not have been created in the first
place if pidx(query_type) return negative. Lets let coverity realize
this.
CID
1362460
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Thu, 2 Jun 2016 14:33:08 +0000 (10:33 -0400)]
freedreno: fix dereference before null check
ptr can actually never be null so just drop the check.
CID
1362464 (#1 of 1): Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking ptr suggests that it may be null,
but it has already been dereferenced on all paths leading to the check.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Sun, 29 May 2016 16:25:32 +0000 (12:25 -0400)]
gallium/util: remove u_staging
Unused, and fixes a couple of coverity warnings: CID
1362171,
1362170
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Acked-by: Marek Olšák <marek.olsak@amd.com>
Rob Clark [Wed, 1 Jun 2016 17:40:53 +0000 (13:40 -0400)]
freedreno/a3xx: only update/emit bordercolor state when needed
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Wed, 1 Jun 2016 16:23:58 +0000 (12:23 -0400)]
freedreno/a4xx: only update/emit bordercolor state when needed
I noticed in stk that it was contributing to a lot of overhead.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Matt Turner [Tue, 24 May 2016 22:10:25 +0000 (15:10 -0700)]
i965: Add missing types to type_sz().
Coverity warns in multiple places about the potential for division by
zero, caused by this function's default case.
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Nanley Chery [Tue, 24 May 2016 21:27:26 +0000 (14:27 -0700)]
mesa/extensions: Fix ES1 extension reporting
Commit
eda15abd84af575d3bde432e2163e30d743a7c87 , unintentionally
advertised these extensions in ES1 contexts. Undo this error.
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Plamena Manolova [Tue, 31 May 2016 16:32:38 +0000 (17:32 +0100)]
egl: Check if API is supported when using eglBindAPI.
According to the EGL specifications before binding an API
we must check whether it's supported first. If not eglBindAPI
should return EGL_FALSE and generate a EGL_BAD_PARAMETER error.
Signed-off-by: Plamena Manolova <plamena.manolova@intel.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Engestrom [Tue, 31 May 2016 01:26:00 +0000 (19:26 -0600)]
st/osmesa: remove double-write (overwriting)
These two lines have been here since the file was created.
I'm guessing the second one was just for testing during dev, so it's the
one that's going away.
CoverityID:
1296205
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Brian Paul <brianp@vmware.com>
Nayan Deshmukh [Thu, 2 Jun 2016 06:41:58 +0000 (12:11 +0530)]
st/vdpau: check for null pointer in get/put bits.
Check for null pointer before accessing arrays in get/put bits
native/YCbCr/Indexed in VdpOutputSurface and VdpVideoSurface.
Signed-off-by: Nayan Deshmukh <nayan26deshmukh@gmail.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Christian König [Wed, 25 May 2016 14:55:48 +0000 (16:55 +0200)]
radeon/uvd: fix the H264 level for Tonga v2
We support 5.2 for a while now.
v2: we even support 5.2 for H264, 5.1 is for HEVC.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Cc: <mesa-stable@lists.freedesktop.org>
Alejandro Piñeiro [Thu, 5 May 2016 09:27:05 +0000 (11:27 +0200)]
mesa/formatquery: add a comment to clarify INTERNALFORMAT_PREFERRED
The comment clarifies that the driver is called only to try to get
a preferred internalformat, and that it was already checked if the
format is supported or not.
Acked-by: Eduardo Lima <elima@igalia.com>
Acked-by: Antia Puentes <apuentes@igalia.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Alejandro Piñeiro [Thu, 5 May 2016 09:28:37 +0000 (11:28 +0200)]
i965/formatquery: remove INTERNALFORMAT_PREFERRED implementation
Right now the implementation only checks if the internalformat is
supported or not. But that implementation is wrong, returning
unsupported for some internalformats. Additionally, checking if
the internalformat is supported or not is already done at mesa/main
before calling the driver hook, so this new check is not needed.
Acked-by: Eduardo Lima <elima@igalia.com>
Acked-by: Antia Puentes <apuentes@igalia.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Alejandro Piñeiro [Wed, 1 Jun 2016 16:49:29 +0000 (18:49 +0200)]
i965/eu: use simd8 when exec_size != EXECUTE_16
Among other thigs, fix a gpu hang when using INTEL_DEBUG=shader_time
for any shader.
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Signed-off-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Jordan Justen [Mon, 23 May 2016 05:31:06 +0000 (22:31 -0700)]
i965: Remove old CS local ID handling
The old method pushed data for each channels uvec3 data of
gl_LocalInvocationID.
The new method pushes 1 dword of data that is a 'thread local ID'
value. Based on that value, we can generate gl_LocalInvocationIndex
and gl_LocalInvocationID with some calculations.
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Jordan Justen [Tue, 31 May 2016 22:45:24 +0000 (15:45 -0700)]
i965: Enable cross-thread constants and compact local IDs for hsw+
The cross thread constant support appears on Haswell. It allows us to
upload a set of uniform data for all threads without duplicating it
per thread.
One complication is that cross-thread constants are loaded into
registers before per-thread constants. Previously, our local IDs were
loaded before the uniform data and treated as 'payload' data, even
though they were actually pushed into the registers like the other
uniform data.
Therefore, in this patch we simultaneously enable a newer layout where
each thread now uses a single uniform slot for a unique local ID for
the thread. This uniform is handled specially to make sure it is added
last into the uniform push constant registers. This minimizes our
usage of push constant registers, and maximizes our ability to use
cross-thread constants for registers.
To swap from the old to the new layout, we also need to flip some
lowering pass switches to let our driver handle the lowering instead.
We also no longer force thread_local_id_index to -1.
v4:
* Minimize size of patch that switches from the old local ID layout
to the new layout (Jason)
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Jordan Justen [Fri, 27 May 2016 07:53:27 +0000 (00:53 -0700)]
anv: Support new local ID generation & cross-thread constants
The cross thread constant support appears on Haswell. It allows us to
upload a set of uniform data for all threads without duplicating it
per thread.
We also support per-thread data which allows us to store a per-thread
ID in one of the uniforms that can be used to calculate the
gl_LocalInvocationIndex and gl_LocalInvocationID variables.
v4:
* Support the old local ID push constant layout as well (Jason)
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>