OSDN Git Service
Roland Scheidegger [Thu, 23 Aug 2018 17:07:05 +0000 (19:07 +0200)]
gallivm: don't use saturated unsigned add/sub intrinsics for llvm 8.0
These have been removed. Unfortunately auto-upgrade doesn't work for
jit. (Worse, it seems we don't get a compilation error anymore when
compiling the shader, rather llvm will just do a call to a null
function in the jitted shaders making it difficult to detect when
intrinsics vanish.)
Luckily the signed ones are still there, I helped convincing llvm
removing them is a bad idea for now, since while the unsigned ones have
sort of agreed-upon simplest patterns to replace them with, this is not
the case for the signed ones, and they require _significantly_ more
complex patterns - to the point that the recognition is IMHO probably
unlikely to ever work reliably in practice (due to other optimizations
interfering). (Even for the relatively trivial unsigned patterns, llvm
already added test cases where recognition doesn't work, unsaturated
add followed by saturated add may produce atrocious code.)
Nevertheless, it seems there's a serious quest to squash all
cpu-specific intrinsics going on, so I'd expect patches to nuke them as
well to resurface.
Adapt the existing fallback code to match the simple patterns llvm uses
and hope for the best. I've verified with lp_test_blend that it does
produce the expected saturated assembly instructions. Though our
cmp/select build helpers don't use boolean masks, but it doesn't seem
to interfere with llvm's ability to recognize the pattern.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=106231
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Marek Olšák [Mon, 6 Aug 2018 02:16:48 +0000 (22:16 -0400)]
st/mesa: expose KHR_texture_compression_astc_sliced_3d
This is ASTC 2D LDR allowing texture arrays and 3D, compressing each
slice as a separate 2D image. Tested by piglit. Trivial.
Marek Olšák [Mon, 6 Aug 2018 01:41:11 +0000 (21:41 -0400)]
st/mesa: expose EXT_disjoint_timer_query
same cap as ARB_timer_query, no changes needed, tested by piglit
Marek Olšák [Mon, 6 Aug 2018 06:48:12 +0000 (02:48 -0400)]
mesa: expose EXT_vertex_attrib_64bit
because the closed driver exposes it.
It's the same as the ARB extension.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Marek Olšák [Mon, 6 Aug 2018 06:26:09 +0000 (02:26 -0400)]
mesa: expose AMD_query_buffer_object
it's a subset of the ARB extension.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Marek Olšák [Mon, 6 Aug 2018 05:55:59 +0000 (01:55 -0400)]
mesa: expose AMD_multi_draw_indirect
because the closed driver exposes it.
This is equivalent to the ARB extension.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Marek Olšák [Mon, 6 Aug 2018 04:56:35 +0000 (00:56 -0400)]
mesa: expose AMD_gpu_shader_int64
because the closed driver exposes it.
It's equivalent to ARB_gpu_shader_int64.
In this patch, I did everything the same as we do for ARB_gpu_shader_int64.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Marek Olšák [Mon, 6 Aug 2018 05:25:51 +0000 (01:25 -0400)]
mesa: expose ARB_post_depth_coverage in the Compatibility profile
It only contains GLSL changes.
v2: allow the layout qualifier on GLSL <= 1.30
Jason Ekstrand [Tue, 24 Jul 2018 05:20:41 +0000 (22:20 -0700)]
intel/nir: Enable nir_opt_find_array_copies
We have to be a bit careful with this one because we want it to run in
the optimization loop but only in the first brw_nir_optimize call.
Later calls assume that we've lowered away copy_deref instructions and
we don't want to introduce any more.
Shader-db results on Kaby Lake:
total instructions in shared programs:
15176942 ->
15176942 (0.00%)
instructions in affected programs: 0 -> 0
helped: 0
HURT: 0
In spite of the lack of any shader-db improvement, this patch completely
eliminates spilling in the Batman: Arkham City tessellation shaders.
This is because we are now able to detect that the temporary array
created by DXVK for storing TCS inputs is a copy of the input arrays and
use indirect URB reads instead of making a copy of 4.5 KiB of input data
and then indirecting on it with if-ladders.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Jason Ekstrand [Tue, 24 Jul 2018 02:16:56 +0000 (19:16 -0700)]
nir: Add an array copy optimization
This peephole optimization looks for a series of load/store_deref or
copy_deref instructions that copy an array from one variable to another
and turns it into a copy_deref that copies the entire array. The
pattern it looks for is extremely specific but it's good enough to pick
up on the input array copies in DXVK and should also be able to pick up
the sequence generated by spirv_to_nir for a OpLoad of a large composite
followed by OpStore. It can always be improved later if needed.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Jason Ekstrand [Wed, 25 Jul 2018 15:54:09 +0000 (08:54 -0700)]
intel/nir: Use nir_shrink_vec_array_vars
Shader-db results on Kaby Lake:
total instructions in shared programs:
15177605 ->
15176765 (<.01%)
instructions in affected programs: 4259 -> 3419 (-19.72%)
helped: 1
HURT: 0
total spills in shared programs: 10954 -> 10855 (-0.90%)
spills in affected programs: 295 -> 196 (-33.56%)
helped: 1
HURT: 0
total fills in shared programs: 22222 -> 22117 (-0.47%)
fills in affected programs: 417 -> 312 (-25.18%)
helped: 1
HURT: 0
The helped shader is from the OglCSDof synmark test. On my Kaby Lake
laptop, the actual framerate of the benchmark didn't appear to improve
beyond the noise.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Jason Ekstrand [Wed, 25 Jul 2018 02:32:27 +0000 (19:32 -0700)]
nir: Add a array-of-vector variable shrinking pass
This pass looks for variables with vector or array-of-vector types and
narrows the type to only the components used.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Jason Ekstrand [Tue, 24 Jul 2018 17:08:20 +0000 (10:08 -0700)]
intel/nir: Use the new structure and array splitting passes
We call structure splitting once because it is guaranteed to split all
the structures in the entire shader in one go. We call array splitting
in the loop in case future optimizations turn indirects into direct
dereferences and we can split more arrays.
Shader-db results on Kaby Lake:
total instructions in shared programs:
15177605 ->
15177605 (0.00%)
instructions in affected programs: 0 -> 0
helped: 0
HURT: 0
This is unsurprising because nir_lower_vars_to_ssa already effectively
does structure and array splitting internally. It doesn't actually
split the variables but it's ability to reason about aliasing in the
presence of arrays and structures and pick out scalars or vectors to be
lowered to SSA values is fairly advanced.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Jason Ekstrand [Tue, 24 Jul 2018 19:33:46 +0000 (12:33 -0700)]
nir: Add an array splitting pass
This pass looks for array variables where at least one level of the
array is never indirected and splits it into multiple smaller variables.
This pass doesn't really do much now because nir_lower_vars_to_ssa can
already see through arrays of arrays and can detect indirects on just
one level or even see that arr[i][0][5] does not alias arr[i][1][j].
This pass exists to help other passes more easily see through arrays of
arrays. If a back-end does implement arrays using scratch or indirects
on registers, having more smaller arrays is likely to have better memory
efficiency.
v2 (Jason Ekstrand):
- Better comments and naming (some from Caio)
- Rework to use one hash map instead of two
v2.1 (Jason Ekstrand):
- Fix a couple of bugs that were added in the rework including one
which basically prevented it from running
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Jason Ekstrand [Tue, 24 Jul 2018 17:08:06 +0000 (10:08 -0700)]
nir: Add a structure splitting pass
This pass doesn't really do much now because nir_lower_vars_to_ssa can
already see through structures and considers them to be "split". This
pass exists to help other passes more easily see through structure
variables. If a back-end does implement arrays using scratch or
indirects on registers, having more smaller arrays is likely to have
better memory efficiency.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Jason Ekstrand [Wed, 25 Jul 2018 15:53:58 +0000 (08:53 -0700)]
nir/types: Add array_or_matrix helpers
Reviewed-by: Thomas Helland<thomashelland90@gmail.com>
Kenneth Graunke [Fri, 24 Aug 2018 00:24:19 +0000 (17:24 -0700)]
i965: don't include compute resources in "Combined" limits
The combined limits should only include shader stages that can be active
at the same time. We don't need to include compute.
See also
cff290df4c09547cd2cb3b129ec59bdebdadba90 for st/mesa.
Unbreaks i965 from assert failing on driver load since Marek's
45f87a48f94148b484961f18a4f1ccf86f066b1c, which dropped the core
Mesa capabilities before adjusting driver limits down to match.
Marek Olšák [Wed, 8 Aug 2018 19:37:21 +0000 (15:37 -0400)]
radeonsi: increase the maximum UBO size to 2 GB
Same as the closed driver.
This causes a failure in GL45-CTS.compute_shader.max, which has a trivial
bug.
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Mon, 6 Aug 2018 12:09:52 +0000 (08:09 -0400)]
radeonsi: bump MAX_GS_INVOCATIONS
same as the closed driver
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Mon, 6 Aug 2018 12:38:54 +0000 (08:38 -0400)]
gallium: add PIPE_CAP_MAX_SHADER_BUFFER_SIZE
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Mon, 6 Aug 2018 12:07:25 +0000 (08:07 -0400)]
gallium: add PIPE_CAP_MAX_GS_INVOCATIONS
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Wed, 8 Aug 2018 19:07:51 +0000 (15:07 -0400)]
tgsi/ureg: don't call tgsi_sanity when it's too slow
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Wed, 8 Aug 2018 19:17:26 +0000 (15:17 -0400)]
st/mesa: fix up uniform limits to be able to expose large UBOs
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Wed, 8 Aug 2018 19:21:05 +0000 (15:21 -0400)]
st/mesa: don't include compute resources in "Combined" limits
The combined limits should only include shader stages that can be active
at the same time.
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Mon, 6 Aug 2018 08:25:15 +0000 (04:25 -0400)]
st/mesa: set ctx->Const.SubPixelBits
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Wed, 8 Aug 2018 19:53:10 +0000 (15:53 -0400)]
glsl: fix error checking against MAX_UNIFORM_LOCATIONS
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Wed, 8 Aug 2018 20:53:03 +0000 (16:53 -0400)]
mesa: make MaxCombinedUniformComponents 64-bit to allow large UBOs
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Mon, 6 Aug 2018 12:09:12 +0000 (08:09 -0400)]
mesa: add ctx->Const.MaxGeometryShaderInvocations
radeonsi wants to report a different value
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Mon, 6 Aug 2018 12:32:23 +0000 (08:32 -0400)]
mesa: don't include compute resources in MAX_COMBINED_* limits
5 is the maximum number of shader stages that can be used by 1 execution
call at the same time (e.g. a draw call). The limit ensures that each
stage can use all of its binding points.
Compute is separate and doesn't need the 5x multiplier.
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Mon, 6 Aug 2018 12:13:35 +0000 (08:13 -0400)]
mesa: bump GL_MAX_ELEMENTS_INDICES and GL_MAX_ELEMENTS_VERTICES
same number as our closed GL driver
v2: don't use MaxArrayLockSize
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Mon, 6 Aug 2018 01:39:00 +0000 (21:39 -0400)]
mesa: remove incorrect change for EXT_disjoint_timer_query
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Marek Olšák [Wed, 22 Aug 2018 21:28:26 +0000 (17:28 -0400)]
glapi: actually implement GL_EXT_robustness for GLES
The extension was exposed but not the functions.
This fixes:
dEQP-GLES31.functional.debug.negative_coverage.get_error.buffer.readn_pixels
dEQP-GLES31.functional.debug.negative_coverage.get_error.state.get_nuniformfv
dEQP-GLES31.functional.debug.negative_coverage.get_error.state.get_nuniformiv
Cc: 18.1 18.2 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Kenneth Graunke [Thu, 23 Aug 2018 08:06:53 +0000 (01:06 -0700)]
intel/decoder: Decode SFIXED values.
This lets us example SAMPLER_STATE's LOD Bias field, among other things.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Emil Velikov [Thu, 16 Aug 2018 12:01:05 +0000 (13:01 +0100)]
travis: use python3 for the autoconf builds
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Emil Velikov [Thu, 16 Aug 2018 12:59:38 +0000 (13:59 +0100)]
configure: allow building with python3
Pretty much all of the scripts are python2+3 compatible.
Check and allow using python3, while adjusting the PYTHON2 refs.
Note:
- python3.4 is used as it's the earliest supported version
- python3 chosen prior to python2
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Acked-by: Eric Engestrom <eric.engestrom@intel.com>
Emil Velikov [Thu, 23 Aug 2018 13:44:19 +0000 (14:44 +0100)]
bin/git_sha1_gen.py: remove execute bit/shebang
The script is executed explicitly via the build system, that uses
PYTHON/prog_python and equivalent.
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Eric Engestrom [Tue, 17 Jul 2018 12:13:40 +0000 (13:13 +0100)]
vk/wsi: avoid reading uninitialised memory
It will be ignored by x11_swapchain_result() anyway (because reaching
the `fail` label without setting `result` means the swapchain status was
already a hard error), but the compiler still complains about reading
uninitialised memory.
While at it, drop the unused assignment right before returning.
Signed-off-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Eric Engestrom [Mon, 9 Jul 2018 20:15:36 +0000 (21:15 +0100)]
egl: drop unused _EGL_BUILT_IN_DRIVER_DRI2
Unused since
b174a1ae720cb404738c "egl: Simplify the "driver" interface".
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Eric Anholt <eric@anholt.net>
Samuel Pitoiset [Fri, 27 Jul 2018 14:14:28 +0000 (16:14 +0200)]
radv/gfx9: implement coherent shaders for VK_ACCESS_SHADER_READ_BIT
Single-sample color and single-sample depth (not stencil)
are coherent with shaders.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl
Mathieu Bridon [Wed, 22 Aug 2018 16:58:26 +0000 (18:58 +0200)]
bin/install_megadrivers.py: Remove shebang and executable bit
Since the script is never executed directly, but launched by Meson as an
argument to the Python interpreter, those are not needed any more.
In addition, they are the reason this script was missed when I moved the
Meson buildsystem to Python 3, so removing them helps avoiding future
confusion.
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Mathieu Bridon [Wed, 22 Aug 2018 12:09:49 +0000 (14:09 +0200)]
meson: Run the install script with Python 3
The script was being run directly as an executable, and it has a
Python 2 shebang.
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Emil Velikov [Fri, 17 Aug 2018 10:41:25 +0000 (11:41 +0100)]
glsl: remove execute bit and shebang from python tests
Just like the rest of the tree - these should be run either as part of
the build system check target, or at the very least with an explicitly
versioned python executable.
Fixes:
db8cd8e3677 ("glcpp/tests: Convert shell scripts to a python script")
Fixes:
97c28cb0823 ("glsl/tests: Convert optimization-test.sh to pure python")
Fixes:
3b52d292273 ("glsl/tests: reimplement warnings-test in python")
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Emil Velikov [Thu, 16 Aug 2018 16:57:10 +0000 (17:57 +0100)]
docs: update required mako version
The requirement was bumped a while back, but we forgot to update the
docs.
Fixes:
ed871af91c2 ("configure.ac: raise Mako required version to
0.8.0")
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Emil Velikov [Thu, 16 Aug 2018 12:36:05 +0000 (13:36 +0100)]
configure: use distutils in ax_check_python_mako_module
Handling the version comparison by hand is a bad idea. Python has a handy
module distutils for that - use it.
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Emil Velikov [Thu, 16 Aug 2018 12:40:21 +0000 (13:40 +0100)]
configure: enforce python 2.7 with AM_PATH_PYTHON
Currently we use AC_CHECK_PROGS looking for python2.7, python2 and
finally python. That is due to the varying names used across the
different OS.
Use the handy AM_PATH_PYTHON which finds the correct name and checks for
the version.
Note: python2.7 has been an unofficial requirement for quite some time.
Update the docs to reflect that.
Cc: Dylan Baker <dylan@pnwbakers.com>
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Ian Romanick [Wed, 18 Apr 2018 21:08:48 +0000 (14:08 -0700)]
i965: Enable INTEL_shader_atomic_float_minmax on Gen9+
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Ian Romanick [Wed, 18 Apr 2018 21:06:27 +0000 (14:06 -0700)]
i965: Sort Gen9+ extension enables
This is a strictly alphabetic sort, as is done in extensions_table.h
There are other options. We should pick one and document it. Right
now, this file is chaos.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Ian Romanick [Wed, 18 Apr 2018 21:02:33 +0000 (14:02 -0700)]
intel/compiler: Implement untyped atomic float min, max, and compare-swap dataport messages
v2: Split changes to the message type field to another patch. Suggested
by Caio.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Ian Romanick [Wed, 18 Apr 2018 21:02:33 +0000 (14:02 -0700)]
intel/compiler: Expand untyped atomic message type field by a bit
This is necessary for a new Gen9 message type that will be added in the
next patch. There are also Gen8 message types that need the extra bit
(mostly for bindless).
v2: Split off from the next patch. Suggested by Caio.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Ian Romanick [Tue, 10 Apr 2018 17:37:42 +0000 (10:37 -0700)]
intel/compiler: Silence unused parameter warnings
src/intel/compiler/brw_disasm_info.c: In function ‘nir_print_instr’:
src/intel/compiler/brw_disasm_info.c:30:61: warning: unused parameter ‘instr’ [-Wunused-parameter]
__attribute__((weak)) void nir_print_instr(const nir_instr *instr, FILE *fp) {}
^~~~~
src/intel/compiler/brw_disasm_info.c:30:74: warning: unused parameter ‘fp’ [-Wunused-parameter]
__attribute__((weak)) void nir_print_instr(const nir_instr *instr, FILE *fp) {}
^~
src/intel/compiler/brw_disasm.c: In function ‘src_ia1’:
src/intel/compiler/brw_disasm.c:850:18: warning: unused parameter ‘_reg_file’ [-Wunused-parameter]
unsigned _reg_file,
^~~~~~~~~
src/intel/compiler/brw_fs_surface_builder.cpp: In function ‘void brw::surface_access::emit_byte_scattered_write(const brw::fs_builder&, const fs_reg&, const fs_reg&, const fs_reg&, unsigned int, unsigned int, unsigned int, brw_predicate)’:
src/intel/compiler/brw_fs_surface_builder.cpp:193:57: warning: unused parameter ‘size’ [-Wunused-parameter]
unsigned dims, unsigned size,
^~~~
v2: Update commit message. brw_fs_generator.cpp warnings were already
fixed by another patch. Noticed by Caio.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Ian Romanick [Wed, 18 Apr 2018 20:34:25 +0000 (13:34 -0700)]
nir: Add floating point atomic min, max, and compare-swap instrinsics
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Ian Romanick [Tue, 10 Apr 2018 01:36:22 +0000 (18:36 -0700)]
nir: Add floating point atomic add instrinsics
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Ian Romanick [Thu, 12 Apr 2018 00:53:22 +0000 (17:53 -0700)]
glsl: Add support for lowering shared-variable float atomics
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Ian Romanick [Mon, 9 Apr 2018 22:26:48 +0000 (15:26 -0700)]
glsl: Add support for lowering SSBO float atomics
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Ian Romanick [Wed, 18 Apr 2018 20:20:25 +0000 (13:20 -0700)]
glsl: Add built-in functions for INTEL_shader_atomic_float_minmax
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Ian Romanick [Wed, 18 Apr 2018 18:45:13 +0000 (11:45 -0700)]
mesa: Extension boilerplate for INTEL_shader_atomic_float_minmax
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Ian Romanick [Thu, 19 Apr 2018 22:35:42 +0000 (15:35 -0700)]
docs: Initial version of INTEL_shader_atomic_float_minmax spec
v2: Describe interactions with the capabilities added by
SPV_INTEL_shader_atomic_float_minmax
v3: Remove 64-bit float support.
v4: Explain NaN issues. Explain issues with atomicMin(-0, +0) and
atomicMax(-0, +0).
v5: Fix whitespace issues noticed by Caio.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Ian Romanick [Thu, 29 Mar 2018 22:16:38 +0000 (15:16 -0700)]
glsl: Add built-in functions for NV_shader_atomic_float
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Ian Romanick [Thu, 29 Mar 2018 21:59:47 +0000 (14:59 -0700)]
mesa: Extension boilerplate for NV_shader_atomic_float
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Gurchetan Singh [Wed, 22 Aug 2018 23:08:34 +0000 (16:08 -0700)]
meson: fix egl build for android
Haven't tested this, but we do include loader.h
in platform_android.c
Fixes:
c5ec1556859b7d33637c9fad13d3473c7b2f9eb3 ("meson: wire up egl/android")
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Gurchetan Singh [Wed, 22 Aug 2018 23:08:33 +0000 (16:08 -0700)]
meson: fix egl build for surfaceless
Without this, I get:
> platform_surfaceless.c:38:10: fatal error: 'loader.h' file not found
> #include "loader.h"
> ^~~~~~~~~~
> 1 error generated.
Fixes:
108d257a16859898f5ce02f4759c5c58f9b8c050 ("meson: build libEGL")
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
v2: Split up patches, modify commit message (Dylan)
Caio Marcelo de Oliveira Filho [Tue, 14 Aug 2018 22:06:38 +0000 (15:06 -0700)]
nir: Give end_block its own index
Since there's no particular reason for the index to be 0, choose an
index that is not used by other block. This is convenient when we
store "per-block" data in an array AND look for the successors
data (e.g. any kind of backwards data-flow analysis).
v2: Add a note about end_block's index. (Jason)
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Caio Marcelo de Oliveira Filho [Fri, 10 Aug 2018 23:04:04 +0000 (16:04 -0700)]
nir: Skip common instructions when comparing deref paths
Deref paths may share the same deref instructions in their chains,
e.g.
ssa_100 = deref_var A
ssa_101 = deref_struct "array_field" of ssa_100
ssa_102 = deref_array "[1]" of ssa_101
ssa_103 = deref_struct "field_a" of ssa_102
ssa_104 = deref_struct "field_a" of ssa_103
when comparing the two last deref instructions, their paths will share
a common sequence ssa_100, ssa_101, ssa_102. This patch skips to next
iteration if the deref instructions are the same. Path[0] (the var)
is still handled specially, so in the case above, only ssa_101 and
ssa_102 will be skipped.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Caio Marcelo de Oliveira Filho [Wed, 25 Jul 2018 18:01:24 +0000 (11:01 -0700)]
nir: Export deref comparison functions
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Caio Marcelo de Oliveira Filho [Mon, 13 Aug 2018 21:41:19 +0000 (14:41 -0700)]
util/dynarray: add a clone function
v2: Fix mem_ctx parameter type. (Thomas)
Reviewed-by: Thomas Helland <thomashelland90@gmail.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Mariusz Ceier [Wed, 22 Aug 2018 20:16:05 +0000 (22:16 +0200)]
amd/addrlib: Fix include path for c99_compat.h
Without this patch mesa doesn't compile:
In file included from ../mesa-9999/src/amd/addrlib/addrinterface.cpp:39:
../mesa-9999/src/util/macros.h:29:10: fatal error: c99_compat.h: No such file or directory
#include "c99_compat.h"
^~~~~~~~~~~~~~
compilation terminated.
Fixes:
15ca5ce99a80d9ebb5ef2b1aca6ea00784931de4
("amd/addrlib: mark returnCode as MAYBE_UNUSED in")
Signed-off-by: Mariusz Ceier <mceier+mesa-dev@gmail.com>
Acked-by: Kai Wasserbäch <kai@dev.carbon-project.org>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Grazvydas Ignotas [Mon, 20 Aug 2018 21:36:58 +0000 (00:36 +0300)]
vulkan/wsi: fix pointer-integer conversion warnings
For 32bit build. Trivial.
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Grazvydas Ignotas [Mon, 20 Aug 2018 21:21:57 +0000 (00:21 +0300)]
radv: use different builtin shader cache for 32bit
Currently if 64bit and 32bit programs are used interchangeably, radv
will keep overwriting the cache. Use separate cache files to avoid
that.
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Grazvydas Ignotas [Mon, 20 Aug 2018 21:08:54 +0000 (00:08 +0300)]
radv: place pointer length into cache uuid
Thanks to reproducible builds, binary file timestamps may be identical
for both 32bit and 64bit packages when built from the same source.
This means radv will use the same cache for both 32 and 64 bit
processes, which leads to crashes.
Conveniently there is a spare byte in cache_uuid, let's place the
pointer size there.
Fixes:
f4e499ec79 "radv: add initial non-conformant radv vulkan driver"
CC: 18.1 18.2 <mesa-stable@lists.freedesktop.org>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=107601
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105904
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Grazvydas Ignotas [Wed, 25 Jul 2018 17:27:48 +0000 (20:27 +0300)]
llvmpipe: add cc clobber to inline asm
The bsr instruction modifies flags, so that needs to be indicated to the
compiler. No effect on generated code, but still needed for correctness.
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Nanley Chery [Fri, 27 Jul 2018 22:34:05 +0000 (15:34 -0700)]
intel/isl: Avoid tiling some 16K-wide render targets
Fix rendering issues on BDW and SKL.
Fixes:
0288fe8d0417730bdd5b3477130dd1dc32bdbcd3
("i965/miptree: Use the correct BLT pitch")
Fixes the following regressions seen
exclusively on SKL:
* KHR-GL46.texture_barrier_ARB.disjoint-texels
* KHR-GL46.texture_barrier_ARB.overlapping-texels
* KHR-GL46.texture_barrier.disjoint-texels
* KHR-GL46.texture_barrier.overlapping-texels
and both on BDW and SKL:
* GTF-GL46.gtf21.GL2FixedTests.buffer_corners.buffer_corners
* GTF-GL46.gtf21.GL2FixedTests.stencil_plane_corners.stencil_plane_corners
v2: Note the fixed tests (Andres).
Don't cause failures with multisampled buffers (Andres).
Don't hamper SKL GT4 (Ken).
v3: Fix the Fixes tag (Dylan).
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=107359
Cc: <mesa-stable@lists.freedesktop.org>
Tested-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Nanley Chery [Mon, 23 Jul 2018 17:09:20 +0000 (10:09 -0700)]
i965/miptree: Fix can_blit_slice()
Check the destination's row pitch against the BLT engine's row pitch
limitation as well.
Fixes:
0288fe8d0417730bdd5b3477130dd1dc32bdbcd3
("i965/miptree: Use the correct BLT pitch")
v2: Fix the Fixes tag (Dylan).
Check the destination row pitch (Chris).
Reported-by: Dylan Baker <dylan@pnwbakers.com>
Cc: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Nanley Chery [Fri, 17 Aug 2018 22:03:14 +0000 (15:03 -0700)]
i965/miptree: Use miptree_map in map_blit functions
This struct contains all the data of interest. can_blit_slice() will use
it in the next patch to calculate the correct pitch.
Suggested-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Rafael Antognolli [Tue, 21 Aug 2018 16:49:03 +0000 (09:49 -0700)]
intel/tools/aubwrite: Always use physical addresses for traces.
It looks like we can't rely on the simulator to always translate virtual
addresses to physical ones correctly. So let's use physical everywhere.
Since our current GGTT maps virtual to physical addresses in a 1:1 way,
no further changes are required.
Additionally, we have other address spaces not in use right now. So
let's make it easier to switch which one we are using but putting the
default one into the aub_file struct.
Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Rafael Antognolli [Tue, 21 Aug 2018 16:47:17 +0000 (09:47 -0700)]
intel/tools/aubwrite: Rename "legacy" to "Trace Block".
Hopefully it's a little more descriptive, and more accurate.
Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Jason Ekstrand [Mon, 9 Jul 2018 04:55:40 +0000 (21:55 -0700)]
nir/vars_to_ssa: Don't build deref nodes for non-local variables
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Marek Olšák [Thu, 16 Aug 2018 01:43:32 +0000 (21:43 -0400)]
ac: fix WAITCNT flags for GFX9
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Kai Wasserbäch [Sat, 18 Aug 2018 10:01:14 +0000 (12:01 +0200)]
amd/addrlib: mark physicalSliceSize as MAYBE_UNUSED in Addr::V1::EgBasedLib::HwlGetSizeAdjustmentMicroTiled
Only used, when asserts are enabled.
Fixes an unused-but-set-variable warning with GCC 8:
../../../src/amd/addrlib/r800/egbaddrlib.cpp: In member function 'virtual long long unsigned int Addr::V1::EgBasedLib::HwlGetSizeAdjustmentMicroTiled(unsigned int, unsigned int, ADDR_SURFACE_FLAGS, unsigned int, unsigned int, unsigned int, unsigned int*, unsigned int*) const':
../../../src/amd/addrlib/r800/egbaddrlib.cpp:4111:13: warning: variable 'physicalSliceSize' set but not used [-Wunused-but-set-variable]
UINT_64 physicalSliceSize;
^~~~~~~~~~~~~~~~~
Signed-off-by: Kai Wasserbäch <kai@dev.carbon-project.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Kai Wasserbäch [Tue, 21 Aug 2018 15:25:46 +0000 (17:25 +0200)]
amd/addrlib: mark numPipes as MAYBE_UNUSED in Addr::V1::EgBasedLib::SanityCheckMacroTiled (v2)
Only used, when asserts are enabled.
Fixes an unused-variable warning with GCC 8:
../../../src/amd/addrlib/r800/egbaddrlib.cpp: In member function 'int Addr::V1::EgBasedLib::SanityCheckMacroTiled(ADDR_TILEINFO*) const':
../../../src/amd/addrlib/r800/egbaddrlib.cpp:982:13: warning: unused variable 'numPipes' [-Wunused-variable]
UINT_32 numPipes = HwlGetPipes(pTileInfo);
^~~~~~~~
v2: Don't realign other variable definitions, to keep in line with file
style (Marek)
Signed-off-by: Kai Wasserbäch <kai@dev.carbon-project.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Kai Wasserbäch [Tue, 21 Aug 2018 15:25:45 +0000 (17:25 +0200)]
amd/addrlib: mark *pEqToCheck as MAYBE_UNUSED in Addr::V2::Gfx9Lib::ComputeStereoInfo (v2)
Only used, when asserts are enabled.
Fixes an unused-variable warning with GCC 8:
../../../src/amd/addrlib/gfx9/gfx9addrlib.cpp: In member function 'ADDR_E_RETURNCODE Addr::V2::Gfx9Lib::ComputeStereoInfo(const ADDR2_COMPUTE_SURFACE_INFO_INPUT*, ADDR2_COMPUTE_SURFACE_INFO_OUTPUT*, unsigned int*) const':
../../../src/amd/addrlib/gfx9/gfx9addrlib.cpp:3879:34: warning: unused variable 'pEqToCheck' [-Wunused-variable]
const ADDR_EQUATION *pEqToCheck = &m_equationTable[eqIndex];
^~~~~~~~~~
v2: Don't realign other variable definitions, to keep in line with file
style (Marek)
Signed-off-by: Kai Wasserbäch <kai@dev.carbon-project.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Kai Wasserbäch [Sat, 18 Aug 2018 09:45:30 +0000 (11:45 +0200)]
amd/addrlib: mark microBlockDim as MAYBE_UNUSED in Addr::V2::Gfx9Lib::HwlComputeBlock256Equation
Only used, when asserts are enabled.
Fixes an unused-but-set-variable warning with GCC 8:
../../../src/amd/addrlib/gfx9/gfx9addrlib.cpp: In member function 'virtual ADDR_E_RETURNCODE Addr::V2::Gfx9Lib::HwlComputeBlock256Equation(AddrResourceType, AddrSwizzleMode, unsigned int, ADDR_EQUATION*) const':
../../../src/amd/addrlib/gfx9/gfx9addrlib.cpp:2473:15: warning: variable 'microBlockDim' set but not used [-Wunused-but-set-variable]
Dim2d microBlockDim = Block256_2d[elementBytesLog2];
^~~~~~~~~~~~~
Signed-off-by: Kai Wasserbäch <kai@dev.carbon-project.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Kai Wasserbäch [Fri, 17 Aug 2018 10:14:22 +0000 (12:14 +0200)]
amd/addrlib: mark returnCode as MAYBE_UNUSED in ElemGetExportNorm
Only used, when asserts are enabled.
Fixes an unused-but-set-variable warning with GCC 8:
../../../src/amd/addrlib/addrinterface.cpp: In function 'int ElemGetExportNorm(ADDR_HANDLE, const ELEM_GETEXPORTNORM_INPUT*)':
../../../src/amd/addrlib/addrinterface.cpp:835:23: warning: variable 'returnCode' set but not used [-Wunused-but-set-variable]
ADDR_E_RETURNCODE returnCode = ADDR_OK;
^~~~~~~~~~
Signed-off-by: Kai Wasserbäch <kai@dev.carbon-project.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Lionel Landwerlin [Fri, 3 Aug 2018 09:41:01 +0000 (10:41 +0100)]
intel: aubinator_viewer: add urb view
This is available through a "Show URB" button on the 3DPRIMITIVE
instructions.
v2: Fix urb allocation end value in tooltip (Rafael)
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
Lionel Landwerlin [Fri, 3 Aug 2018 09:40:43 +0000 (10:40 +0100)]
intel: aubinator_viewer: store urb state during decoding
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
Lionel Landwerlin [Sat, 28 Jul 2018 18:13:28 +0000 (19:13 +0100)]
intel: tools: add aubinator viewer
A graphical user interface version of aubinator.
Allows you to :
- simultaneously look at multiple points in the aub file (using all
the goodness of the existing decoding in aubinator)
- edit an aub file
v2: Switch from GLFW to GTK+3
v3: Fix warning when exiting
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Rafael Antognolli <rafael.antognolli@intel.com> (v1)
Lionel Landwerlin [Sun, 29 Jul 2018 18:32:45 +0000 (19:32 +0100)]
intel: tools: import ImGui
We want to add a new UI tool to decode aub files. This will use the
Dear ImGui library to render its interface. The build of this UI
toolkit is conditional to -Dwith_tools=intel-ui which superseeds
-Dwith_tools=intel.
The main way to use ImGui is to embed its source code at a particular
revision. Most embedding projects have to do a bit of integration
which is really specific to one's project. In our case the only
modification is to include libepoxy. We also choose to use Gtk+3 for
the window system integration. As oppose to the previous previous
version of this patch using GLFW, Gtk+ is able to handle X11/Wayland
session as well as property DPI scaling on retina monitors.
The import was done at this commit (https://github.com/ocornut/imgui) :
commit
6211f40f3d903dd9df961256e044029c49793aa3
Author: omar <omarcornut@gmail.com>
Date: Fri Jul 27 12:29:33 2018 +0200
Internals: Drag and Drop: default drop preview use a narrower clipping rectangle (no effect here, but other branches uses a narrow clipping rectangle that was too small so this is a fix for it) + Comments
v2: Switch from GLFW to GTK+ (Lionel)
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Rafael Antognolli <rafael.antognolli@intel.com>
Lionel Landwerlin [Mon, 6 Aug 2018 11:06:13 +0000 (12:06 +0100)]
intel: tools: aub_mem: reuse already mapped ppgtt buffers
When we map a PPGTT buffer into a continous address space of aubinator
to be able to inspect it, we currently add it to the list of BOs to
unmap once we're finished. An optimization we can apply it to look up
that list before trying to remap PPGTT buffers again (we already do
this for GGTT buffers).
We need to take some care before doing this because the list also
contains GGTT BOs. As GGTT & PPGTT are 2 different address spaces, we
can have matching addresses in both that point to different physical
locations.
This changes adds a flag on the elements of the list of mapped BOs to
differenciate between GGTT & PPGTT, which allows use to reuse that
list when looking up both address spaces.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
Lionel Landwerlin [Thu, 2 Aug 2018 09:15:41 +0000 (10:15 +0100)]
intel: tools: aubmem: map gtt data to aub file
This will allow the aubinator viewer tool to modify the aub data that
was loaded at a particular gtt address.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
Lionel Landwerlin [Sun, 5 Aug 2018 09:22:37 +0000 (10:22 +0100)]
intel: tools: create libaub
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Lionel Landwerlin [Sun, 29 Jul 2018 18:32:13 +0000 (19:32 +0100)]
intel: tools: aubwrite: wrap function declarations for c++
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
Lionel Landwerlin [Sun, 29 Jul 2018 00:01:36 +0000 (01:01 +0100)]
intel: tools: split memory management out of aubinator
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
Lionel Landwerlin [Sun, 29 Jul 2018 13:13:25 +0000 (14:13 +0100)]
util: rb_tree: add safe iterators
v2: Add helper to make iterators more readable (Rafael)
Fix rev iterator bug (Rafael)
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
Lionel Landwerlin [Sat, 28 Jul 2018 18:13:04 +0000 (19:13 +0100)]
intel: tools: split aub parsing from aubinator
v2: add parsing error callback (Lionel)
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com> (v1)
Mathieu Bridon [Fri, 17 Aug 2018 19:32:18 +0000 (21:32 +0200)]
meson: Run the test with Python 3
This is a patch from me and a patch from Mathieu Bridon squashed
together.
Signed-off-by: Dylan Baker <dylan@pnwbakers.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Reviewed-by: Mathieu Bridon <bochecha@daitauha.fr>
Mathieu Bridon [Fri, 17 Aug 2018 19:32:17 +0000 (21:32 +0200)]
python: Disable universal newlines
We are testing the behaviour of a tool, for different input files, each
one using a different newline sequence. ('\n' on UNIX, '\r\n' on
Windows, …)
Unfortunately, when opening a file in text mode, Python 3 will by
default enable the "universal newlines" mode, which means it replaces
all the known newline sequences by '\n'.
This (usually useful) behaviour breaks the tests, which are specifically
trying to handle files with newline sequences different from '\n'.
Disabling the universal newlines mode fixes the tests.
However, to keep the script compatible with both Python 2 and 3, we must
use the io.open() function instead of the open() builtin, as the latter
only knows about the `newline` argument on Python 3.
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Mathieu Bridon [Fri, 17 Aug 2018 19:32:16 +0000 (21:32 +0200)]
python: difflib prefers unicode strings
Python 3 does not automatically convert from bytes to unicode strings
like Python 2 used to do.
This commit makes sure we pass unicode strings to difflib.unified_diff,
so that the script works on both Python 2 and 3.
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Dylan Baker [Thu, 16 Aug 2018 21:17:24 +0000 (14:17 -0700)]
compiler/glsl/tests: Make tests python3 safe
v2: - explicitly decode the output of subprocesses
- handle bytes and string types consistently rather than relying on
python 2's coercion for bytes and ignoring them in python 3
v3: - explicitly set encode as well as decode
- python 2.7 and 3.x `bytes` instead of defining an alias
Reviewed-by: Mathieu Bridon <bochecha@daitauha.fr>
Juan A. Suarez Romero [Tue, 21 Aug 2018 10:34:48 +0000 (12:34 +0200)]
travis: SWR requires LLVM 6.0
v2: update clarification why ubuntu-toolchain-r-test is required (Emil)
Fixes:
0cef0cccf51 ("swr: bump minimum supported LLVM version to 6.0")
Cc: Dylan Baker <dylan@pnwbakers.com>
Cc: Eric Engestrom <eric.engestrom@intel.com>
Acked-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Samuel Pitoiset [Wed, 22 Aug 2018 12:42:51 +0000 (14:42 +0200)]
ac/nir: fix getting GLSL type of array of samplers for TG4
This fixes a crash in build_tex_intrinsic() when trying to
launch the Basemark GPU benchmark on GFX8. It looks like
there is still something wrong because some frames are black.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=106980
CC: 18.2 <mesa-stable@lists.freedesktop.org>
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Samuel Pitoiset [Wed, 22 Aug 2018 10:34:13 +0000 (12:34 +0200)]
radv: remove dead variables after splitting per member structs
Otherwise, nir_lower_clip_cull_distance_arrays might report
wrong number of output clips/culls because it relies on
shader output variables and some of them might be dead.
This fixes a rendering issue with Dolphin and Super Mario
Sunshine.
Fixes:
b0c643d8f5 ("spirv: Use NIR per-member splitting")
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=107610
CC: 18.2 <mesa-stable@lists.freedesktop.org>
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>