mesa.git
20 months agodocs: Add release notes for 20.0.7
Dylan Baker [Thu, 14 May 2020 16:38:57 +0000 (09:38 -0700)]
docs: Add release notes for 20.0.7

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5041>

20 months agointel: Silence unused parameter warning in __intel_log_use_args
Ian Romanick [Mon, 11 May 2020 20:10:52 +0000 (13:10 -0700)]
intel: Silence unused parameter warning in __intel_log_use_args

...in every file that includes intel_log.h.

In file included from src/intel/vulkan/anv_private.h:93,
                 from src/intel/vulkan/genX_cmd_buffer.c:27:
src/intel/common/intel_log.h: In function ‘__intel_log_use_args’:
src/intel/common/intel_log.h:75:34: warning: unused parameter ‘format’ [-Wunused-parameter]

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4994>

20 months agoanv: Silence unused parameter warning in anv_image_get_clear_color_addr
Ian Romanick [Mon, 11 May 2020 20:09:17 +0000 (13:09 -0700)]
anv: Silence unused parameter warning in anv_image_get_clear_color_addr

...in every file that includes anv_private.h.

In file included from src/intel/vulkan/genX_cmd_buffer.c:27:
src/intel/vulkan/anv_private.h: In function ‘anv_image_get_clear_color_addr’:
src/intel/vulkan/anv_private.h:3690:57: warning: unused parameter ‘device’ [-Wunused-parameter]

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4994>

20 months agoanv/tests: Silence unused parameter warnings in main
Ian Romanick [Mon, 11 May 2020 20:06:56 +0000 (13:06 -0700)]
anv/tests: Silence unused parameter warnings in main

src/intel/vulkan/tests/block_pool_no_free.c: In function ‘main’:
src/intel/vulkan/tests/block_pool_no_free.c:147:14: warning: unused parameter ‘argc’ [-Wunused-parameter]
src/intel/vulkan/tests/block_pool_no_free.c:147:27: warning: unused parameter ‘argv’ [-Wunused-parameter]

src/intel/vulkan/tests/block_pool_grow_first.c: In function ‘main’:
src/intel/vulkan/tests/block_pool_grow_first.c:27:14: warning: unused parameter ‘argc’ [-Wunused-parameter]
src/intel/vulkan/tests/block_pool_grow_first.c:27:27: warning: unused parameter ‘argv’ [-Wunused-parameter]

src/intel/vulkan/tests/state_pool.c: In function ‘main’:
src/intel/vulkan/tests/state_pool.c:36:14: warning: unused parameter ‘argc’ [-Wunused-parameter]
src/intel/vulkan/tests/state_pool.c:36:27: warning: unused parameter ‘argv’ [-Wunused-parameter]

src/intel/vulkan/tests/state_pool_padding.c: In function ‘main’:
src/intel/vulkan/tests/state_pool_padding.c:27:14: warning: unused parameter ‘argc’ [-Wunused-parameter]
src/intel/vulkan/tests/state_pool_padding.c:27:27: warning: unused parameter ‘argv’ [-Wunused-parameter]

src/intel/vulkan/tests/state_pool_no_free.c: In function ‘main’:
src/intel/vulkan/tests/state_pool_no_free.c:115:14: warning: unused parameter ‘argc’ [-Wunused-parameter]
src/intel/vulkan/tests/state_pool_no_free.c:115:27: warning: unused parameter ‘argv’ [-Wunused-parameter]

src/intel/vulkan/tests/state_pool_free_list_only.c: In function ‘main’:
src/intel/vulkan/tests/state_pool_free_list_only.c:35:14: warning: unused parameter ‘argc’ [-Wunused-parameter]
src/intel/vulkan/tests/state_pool_free_list_only.c:35:27: warning: unused parameter ‘argv’ [-Wunused-parameter]

v2: Use 'int main(void)' instead.  Suggested by Jason.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4994>

20 months agoanv/tests: Don't rely on assert or changing NDEBUG in tests
Ian Romanick [Mon, 11 May 2020 19:33:24 +0000 (12:33 -0700)]
anv/tests: Don't rely on assert or changing NDEBUG in tests

This is the last part of the fix for #2903.

v2: Add test_common.h.

Fixes: f7c56475d25 ("anv/tests: compile to something sensible in release builds")
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4994>

20 months agoaco: fix WQM coalescing
Daniel Schürmann [Thu, 14 May 2020 11:40:55 +0000 (12:40 +0100)]
aco: fix WQM coalescing

get_reg_specified() doesn't handle special registers like SCC.
Fixes: a5fc96b533418dc2d68f17f3f19ac5f82d59b978 ('aco: coalesce parallelcopies during register allocation')
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5036>

20 months agoanv: Fix descriptor set clean-up on BO allocation failure
Jason Ekstrand [Thu, 14 May 2020 03:57:46 +0000 (22:57 -0500)]
anv: Fix descriptor set clean-up on BO allocation failure

This was a bit of rebase fail when writing 682c81bdfb.  We stopped
freeing descriptor sets back to the pool and started calling
vk_object_base_finish.  This commit reverts a that hunk should have
never made its way into the final patch.

Fixes: 682c81bdfb "vulkan,anv: Add a base object struct type"
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Tested-by: Mark Janes <mark.a.janes@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5032>

20 months agoanv: Call vk_object_base_finish for image views
Jason Ekstrand [Wed, 13 May 2020 21:26:22 +0000 (16:26 -0500)]
anv: Call vk_object_base_finish for image views

Fixes: 682c81bdfb7 "vulkan,anv: Add a base object struct type"
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5032>

20 months agozink: correct PIPE_SHADER_CAP_MAX_SHADER_IMAGES
Erik Faye-Lund [Wed, 13 May 2020 15:13:47 +0000 (17:13 +0200)]
zink: correct PIPE_SHADER_CAP_MAX_SHADER_IMAGES

We don't support shader-images yet, so this is premature to expose.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5024>

20 months agozink: do not expose real value for PIPE_CAP_MAX_VIEWPORTS
Erik Faye-Lund [Wed, 13 May 2020 12:40:47 +0000 (14:40 +0200)]
zink: do not expose real value for PIPE_CAP_MAX_VIEWPORTS

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5024>

20 months agometa: Remove support for multisample blits
Ian Romanick [Thu, 15 Jun 2017 21:47:54 +0000 (14:47 -0700)]
meta: Remove support for multisample blits

Since i965 no longer uses this function for blitting color buffers,
there is no driver left that will ever support multisample textures and
use _mesa_meta_BlitFramebuffer.

v2: Delete blit_state::msaa_shaders and enum blit_msaa_shader.

   text     data     bss      dec    hex  filename
12243286 1344936 1290748 14878970 e308fa  before/lib64/dri/i965_dri.so
12240398 1344936 1290748 14876082 e2fdb2  after/lib64/dri/i965_dri.so

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/856>

20 months agometa: Coalesce the GLSL and FF paths in meta_clear
Ian Romanick [Fri, 2 Jun 2017 23:29:20 +0000 (16:29 -0700)]
meta: Coalesce the GLSL and FF paths in meta_clear

   text     data     bss      dec    hex  filename
12243286 1344936 1290748 14878970 e308fa  before/lib64/dri/i965_dri.so
12243286 1344936 1290748 14878970 e308fa  after/lib64/dri/i965_dri.so

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/856>

20 months agometa: Use same vertex coordinates for GLSL and FF clears
Ian Romanick [Fri, 2 Jun 2017 23:21:25 +0000 (16:21 -0700)]
meta: Use same vertex coordinates for GLSL and FF clears

   text     data     bss      dec    hex  filename
12243446 1344936 1290748 14879130 e3099a  before/lib64/dri/i965_dri.so
12243286 1344936 1290748 14878970 e308fa  after/lib64/dri/i965_dri.so

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/856>

20 months agometa: Stop frobbing MatrixMode
Ian Romanick [Fri, 2 Jun 2017 23:15:52 +0000 (16:15 -0700)]
meta: Stop frobbing MatrixMode

   text     data     bss      dec    hex  filename
12243510 1344936 1290748 14879194 e309da  before/lib64/dri/i965_dri.so
12243446 1344936 1290748 14879130 e3099a  after/lib64/dri/i965_dri.so

v2: Use _mesa_load_matrix to set the projection matrix instead of
frobbing dirty bits directly.  Suggested by Jason.  Clean up some
comments in the neighborhood.  Since glOrtho isn't called, there's no
point in mentioning it in the comment.  Only _mesa_load_identity_matrix
on the projection matrix when it isn't set to an ortho matrix.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/856>

20 months agomesa: Add function to calculate an orthographic projection
Ian Romanick [Tue, 28 Apr 2020 17:18:47 +0000 (10:18 -0700)]
mesa: Add function to calculate an orthographic projection

Unlike the existing _math_matrix_ortho, the new _math_float_ortho
function just stores the calculated matrix in an array of floats.  It
does not multiply the new matrix with data already stored.

   text     data     bss      dec    hex  filename
12243486 1344936 1290748 14879170 e309c2  before/lib64/dri/i965_dri.so
12243510 1344936 1290748 14879194 e309da  after/lib64/dri/i965_dri.so

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/856>

20 months agomesa: Add matrix utility functions to load matrices
Ian Romanick [Fri, 2 Jun 2017 22:59:24 +0000 (15:59 -0700)]
mesa: Add matrix utility functions to load matrices

These are basically DSA versions of glLoadIdentity() and glLoadMatrix()
that are available for internal Mesa use.

   text     data     bss      dec    hex  filename
12243574 1344936 1290748 14879258 e30a1a  before/lib64/dri/i965_dri.so
12243486 1344936 1290748 14879170 e309c2  after/lib64/dri/i965_dri.so

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/856>

20 months agometa: Remove support for clearing integer buffers
Ian Romanick [Thu, 15 Jun 2017 00:35:28 +0000 (17:35 -0700)]
meta: Remove support for clearing integer buffers

Since i965 no longer uses this function for clearing color buffers,
there is no driver left that will ever support integer textures and use
_mesa_meta_glsl_Clear.

As a side note, the has_integer_textures check was rubbish anyway
because meta always smashes the API to API_OPENGL_COMPAT.

   text     data     bss      dec    hex  filename
12244406 1344936 1290748 14880090 e30d5a  before/lib64/dri/i965_dri.so
12243574 1344936 1290748 14879258 e30a1a  after/lib64/dri/i965_dri.so

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/856>

20 months agometa: Make _mesa_meta_setup_sampler static
Ian Romanick [Mon, 6 May 2019 22:45:47 +0000 (15:45 -0700)]
meta: Make _mesa_meta_setup_sampler static

   text     data     bss      dec    hex  filename
12244854 1344936 1290748 14880538 e30f1a  before/lib64/dri/i965_dri.so
12244406 1344936 1290748 14880090 e30d5a  after/lib64/dri/i965_dri.so

v2: Put static on the function definition too.  Suggested by Paulo.

v3: Reformat prototype.  Suggested by Jason.

Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com> [v2]
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/856>

20 months agometa: Make _mesa_meta_texture_object_from_renderbuffer static
Ian Romanick [Mon, 6 May 2019 22:05:26 +0000 (15:05 -0700)]
meta: Make _mesa_meta_texture_object_from_renderbuffer static

   text     data     bss      dec    hex  filename
12244974 1344936 1290748 14880658 e30f92  before/lib64/dri/i965_dri.so
12244854 1344936 1290748 14880538 e30f1a  after/lib64/dri/i965_dri.so

v2: Put static on the function definition too.  Suggested by Paulo.

v3: Reformat prototype.  Suggested by Jason.

Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com> [v2]
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/856>

20 months agoi965: Assert that blorp always handles color blits
Ian Romanick [Mon, 19 Jun 2017 18:10:38 +0000 (11:10 -0700)]
i965: Assert that blorp always handles color blits

Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/856>

20 months agonir/validate: validate the stride for deref_ptr_as_array
Karol Herbst [Thu, 5 Mar 2020 21:13:24 +0000 (22:13 +0100)]
nir/validate: validate the stride for deref_ptr_as_array

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4068>

20 months agonir/deref: copy ptr_stride when rematerializing
Karol Herbst [Thu, 5 Mar 2020 21:35:10 +0000 (22:35 +0100)]
nir/deref: copy ptr_stride when rematerializing

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4068>

20 months agotargets/opencl: fix build against LLVM>=10 with Polly support
Jan Palus [Fri, 10 Apr 2020 16:00:40 +0000 (18:00 +0200)]
targets/opencl: fix build against LLVM>=10 with Polly support

see https://bugs.llvm.org/show_bug.cgi?id=44870

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4511>

20 months agofreedreno: Avoid duplicate BO relocs in FD_RINGBUFFER_OBJECTs.
Eric Anholt [Fri, 8 May 2020 21:53:47 +0000 (14:53 -0700)]
freedreno: Avoid duplicate BO relocs in FD_RINGBUFFER_OBJECTs.

For the piglit drawoverhead case, 5/18 of the objects' relocs were
duplicated.  We can dedupe them at object create time (since objects are
long-lived) and avoid repeated relocation work at emit time.

nohw drawoverhead program statechange throughput 2.34082% +/- 0.645832%
(n=10).

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5020>

20 months agofreedreno: Fix resource layout dump loop.
Eric Anholt [Tue, 12 May 2020 21:41:53 +0000 (14:41 -0700)]
freedreno: Fix resource layout dump loop.

Apparently I've never dumped a fully populated slices array, so the 0-init
always terminated the loop.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5020>

20 months agozink: disable vkCmdResolveImage when respecting render-condition
Erik Faye-Lund [Tue, 12 May 2020 13:01:16 +0000 (15:01 +0200)]
zink: disable vkCmdResolveImage when respecting render-condition

vkCmdResolveImage doesn't respect render-condition, so let's fall back
to blitter in this case instead.

Fixes: 80d7cc6f129 ("zink: enable conditional rendering if available")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5008>

20 months agoanv: Fix deadlock in anv_timelines_wait
Danylo Piliaiev [Tue, 12 May 2020 09:27:58 +0000 (12:27 +0300)]
anv: Fix deadlock in anv_timelines_wait

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/2945
Fixes: 34f32a6d6648073e2fda3fb78377124fb32bb288
Signed-off-by: Danylo Piliaiev <danylo.piliaiev@globallogic.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5005>

20 months agogitlab-ci: Install g++-mingw-w64-x86-64-win32 instead of mingw-w64
Michel Dänzer [Mon, 27 Apr 2020 16:46:32 +0000 (18:46 +0200)]
gitlab-ci: Install g++-mingw-w64-x86-64-win32 instead of mingw-w64

mingw-w64 pulls in a lot more packages we don't need.

g++-mingw-w64-x86-64-win32 is only available in Debian testing, so get
all mingw packages from there.

Acked-by: Andres Gomez <agomez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4851>

20 months agogitlab-ci: Move lib{drm,pciaccess}-dev cross packages out of loop
Michel Dänzer [Thu, 30 Apr 2020 07:28:04 +0000 (09:28 +0200)]
gitlab-ci: Move lib{drm,pciaccess}-dev cross packages out of loop

Simpler like this, since they're only needed for one cross architecture
each.

Acked-by: Andres Gomez <agomez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4851>

20 months agogitlab-ci: Install WINE from Debian testing
Michel Dänzer [Mon, 27 Apr 2020 10:42:58 +0000 (12:42 +0200)]
gitlab-ci: Install WINE from Debian testing

Instead of a third-party repository which has proved unreliable at
times.

This pulls in glibc 2.30 from testing in the x86_build image, so we need
to update the x86_test-{gl,vk} images to match.

Acked-by: Andres Gomez <agomez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4851>

20 months agogitlab-ci: Add Debian testing repository for x86_build image
Michel Dänzer [Fri, 8 May 2020 09:30:09 +0000 (11:30 +0200)]
gitlab-ci: Add Debian testing repository for x86_build image

We don't want LLVM 8 packages to be pulled in from testing though (it
would make installing llvm-8-dev for cross architectures a lot more
complicated), so explicitly select buster-backports for them (they were
already implicitly installed from there before, since they're not
available in buster proper).

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4851>

20 months agogitlab-ci: Move down container_pre_build.sh invocation in x86_build.sh
Michel Dänzer [Thu, 30 Apr 2020 09:45:58 +0000 (11:45 +0200)]
gitlab-ci: Move down container_pre_build.sh invocation in x86_build.sh

It was in the middle of package installations.

Acked-by: Andres Gomez <agomez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4851>

20 months agogitlab-ci: Update to current templates
Michel Dänzer [Wed, 29 Apr 2020 16:47:08 +0000 (18:47 +0200)]
gitlab-ci: Update to current templates

Notable changes:

* No longer generate a separate *-built-by-job-* image tag, instead
  store the pipeline/job information as labels in the image.
* Clean up some package information files which were accidentally left
  before, possibly resulting in slightly smaller images.

Acked-by: Andres Gomez <agomez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4851>

20 months agozink. Changed sampler default name.
Duncan Hopkins [Wed, 25 Sep 2019 10:52:37 +0000 (11:52 +0100)]
zink. Changed sampler default name.

Changed the sampler variable name from 'sampler' to 'sampler_<num>' to stop symbol classes in the Metal MSL shaders, as 'sampler' is a keyword.
Improves human readability when debugging issues.

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4211>

20 months agoradv: enable shaderResourceMinLod
Samuel Pitoiset [Fri, 8 May 2020 17:26:19 +0000 (19:26 +0200)]
radv: enable shaderResourceMinLod

This feature was missing for unknown reasons.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4989>

20 months agoac/llvm: add support for texturing with clamped LOD
Samuel Pitoiset [Mon, 11 May 2020 14:51:33 +0000 (16:51 +0200)]
ac/llvm: add support for texturing with clamped LOD

This is a requirement for the shaderResourceMinLod feature which
allows to clamp LOD. This uses all image_sample_*_cl variants.

All dEQP-VK.glsl.texture_functions.texture*clamp.* pass.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4989>

20 months agoaco: add support for texturing with clamped LOD
Samuel Pitoiset [Mon, 11 May 2020 14:33:14 +0000 (16:33 +0200)]
aco: add support for texturing with clamped LOD

This is a requirement for the shaderResourceMinLod feature which
allows to clamp LOD. This uses all image_sample_*_cl variants.

All dEQP-VK.glsl.texture_functions.texture*clamp.* pass.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4989>

20 months agoaco: remove useless check for nir_tex_src_bias
Samuel Pitoiset [Mon, 11 May 2020 13:23:52 +0000 (15:23 +0200)]
aco: remove useless check for nir_tex_src_bias

I think only nir_texop_txb can have a bias operand anyways.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4989>

20 months agoCI: Windows: Build LLVM and llvmpipe
Daniel Stone [Tue, 5 May 2020 14:49:22 +0000 (15:49 +0100)]
CI: Windows: Build LLVM and llvmpipe

We will eventually need to build our own LLVM on Windows in order to
build libclc and other bits which are required for the d3d12 build, as
well as to be able to test SPIR-V/OpenCL on llvmpipe.

Start doing this now, building into the base container, and exercise
this by building llvmpipe under Windows.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4946>

20 months agollvmpipe: Expect increased exp precision on Windows
Daniel Stone [Thu, 16 Apr 2020 22:27:41 +0000 (23:27 +0100)]
llvmpipe: Expect increased exp precision on Windows

'Newer' versions of MSVCRT than 2013 appear to have fixed the bug around expf
precision which caused bb9e8c5090f0. It's not clear when this was
changed, but at least on Windows 10 machines with Visual Studio 2019,
expf behaves in line with other implementations.

As there is no clear way to test for the version of the VCRT in use,
simply mark this test as expected-pass rather than xfail.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4946>

20 months agofreedreno/ir3: use lower_wrmasks pass
Rob Clark [Wed, 6 May 2020 21:58:28 +0000 (14:58 -0700)]
freedreno/ir3: use lower_wrmasks pass

Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
20 months agonir: add pass to lower disjoint wrmask's
Rob Clark [Wed, 6 May 2020 21:20:09 +0000 (14:20 -0700)]
nir: add pass to lower disjoint wrmask's

Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
20 months agonir: add helper to copy const_index[]
Rob Clark [Wed, 6 May 2020 20:35:51 +0000 (13:35 -0700)]
nir: add helper to copy const_index[]

It seems less brittle to not assume they are in the same order for src
and dst instructions.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
20 months agonir: fix indices for ir3 ssbo_atomic intrinsics
Rob Clark [Wed, 6 May 2020 22:52:53 +0000 (15:52 -0700)]
nir: fix indices for ir3 ssbo_atomic intrinsics

Caught by the sanity checking in nir_intrinsic_copy_const_indices()
(which is introduced by the next patch).

Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
20 months agofreedreno/ir3: use const_index accessors
Rob Clark [Wed, 6 May 2020 20:14:39 +0000 (13:14 -0700)]
freedreno/ir3: use const_index accessors

Cleans up a couple spots that were still open-coding this.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
20 months agofreedreno/ir3: Drop wrmask for ir3 local and global store intrinsics
Kristian H. Kristensen [Wed, 13 May 2020 20:19:57 +0000 (13:19 -0700)]
freedreno/ir3: Drop wrmask for ir3 local and global store intrinsics

These intrinsics are supposed to map to the underlying hardware
instructions, which don't have wrmask. We use them when we lower
store_output in the geometry pipeline and since store_output gets
lowered to temps, we always see full wrmasks there.

20 months agonir: Add some docs to the metadata types
Jason Ekstrand [Wed, 13 May 2020 18:29:43 +0000 (13:29 -0500)]
nir: Add some docs to the metadata types

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5028>

20 months agofreedreno: Fix attempts to push UBO contents past the constlen on pre-a6xx.
Eric Anholt [Wed, 13 May 2020 17:34:25 +0000 (10:34 -0700)]
freedreno: Fix attempts to push UBO contents past the constlen on pre-a6xx.

The binning variant likely won't have any UBO load code in it, so we were
writing past constlen (and sometimes asserting about it) when loading more
than one ubo block.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5027>

20 months agodocs: update calendar for 20.1.0-rc3
Eric Engestrom [Wed, 13 May 2020 17:29:02 +0000 (19:29 +0200)]
docs: update calendar for 20.1.0-rc3

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5026>

20 months agopanfrost: Fix background showing when using discard
Icecream95 [Tue, 12 May 2020 02:08:09 +0000 (14:08 +1200)]
panfrost: Fix background showing when using discard

This fixes problems in a number of games, including SuperTuxKart,
OpenMW and RVGL.

v2: Use MALI_READS_ZS | 0x20 instead of MALI_WRITES_Z to match with
the blob. Keep using 0x400 | 0x20 when depth is disabled.

Closes: #2620
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5001>

20 months agoanv: Translate relative timeout to absolute when calling anv_timelines_wait
Danylo Piliaiev [Wed, 13 May 2020 16:28:59 +0000 (19:28 +0300)]
anv: Translate relative timeout to absolute when calling anv_timelines_wait

Fixes: 34f32a6d6648073e2fda3fb78377124fb32bb288
Signed-off-by: Danylo Piliaiev <danylo.piliaiev@globallogic.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5025>

20 months agoanv: Set MOCS in 3DSTATE_CONSTANT_* on Gen9+
Jason Ekstrand [Wed, 13 May 2020 03:55:30 +0000 (22:55 -0500)]
anv: Set MOCS in 3DSTATE_CONSTANT_* on Gen9+

While we're here, we add a nice detailed comment about why always
assuming internal is ok.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5022>

20 months agoanv: Set 3DSTATE_VF_INSTANCING on the SVGS element
Jason Ekstrand [Wed, 13 May 2020 03:31:54 +0000 (22:31 -0500)]
anv: Set 3DSTATE_VF_INSTANCING on the SVGS element

It probably doesn't matter because that buffer should have a stride of
zero.  However, it still seems like a good idea just to be safe.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5022>

20 months agofreedreno: Drop the noubo fails list for CI, since there aren't any now.
Eric Anholt [Tue, 12 May 2020 23:36:08 +0000 (16:36 -0700)]
freedreno: Drop the noubo fails list for CI, since there aren't any now.

The remaining two fails in the list are the same as for the normal CI run.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4858>

20 months agofreedreno/a6xx: Use LDC for UBO loads.
Eric Anholt [Mon, 20 Apr 2020 23:39:54 +0000 (16:39 -0700)]
freedreno/a6xx: Use LDC for UBO loads.

It saves addressing math, but may cause multiple loads to be done and
bcseled due to NIR not giving us good address alignment information
currently.  I don't have any workloads I know of using non-const-uploaded
UBOs, so I don't have perf numbers for it

This makes us match the GLES blob's behavior, and turnip (other than being
bindful).

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4858>

20 months agofreedreno: Trim num_ubos to just the ones we haven't lowered to constbuf.
Eric Anholt [Tue, 12 May 2020 23:07:50 +0000 (16:07 -0700)]
freedreno: Trim num_ubos to just the ones we haven't lowered to constbuf.

With the upcoming LDC usage in the GL driver, we don't want to be
uploading descriptors for every UBO when they aren't actually in use.
Trimming NIR's num_ubos will avoid that, and cleans up num_ubo handling
elsewhere right now.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4858>

20 months agofreedreno/ir3: Move i/o offset lowering after analyze_ubo_ranges.
Eric Anholt [Fri, 1 May 2020 23:00:17 +0000 (16:00 -0700)]
freedreno/ir3: Move i/o offset lowering after analyze_ubo_ranges.

I found that when moving more UBOs to load_ubo_ir3, analyze_ubo_ranges
would move things back in a broken way.  We can just run this pass later
and drop the _ir3 path.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4858>

20 months agofreedreno/ir3: Leave the cursor alone during ir3_nir_try_propagate_bit_shift.
Eric Anholt [Tue, 21 Apr 2020 00:14:53 +0000 (17:14 -0700)]
freedreno/ir3: Leave the cursor alone during ir3_nir_try_propagate_bit_shift.

Otherwise, we might end up inserting the nir_intrinsic_load_ubo_ir3()
after the non-offset src's definition, leading to nir_validate() failures.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4858>

20 months agofreedreno/ir3: Clean up a silly nir_src_for_ssa(src.ssa).
Eric Anholt [Mon, 20 Apr 2020 23:43:25 +0000 (16:43 -0700)]
freedreno/ir3: Clean up a silly nir_src_for_ssa(src.ssa).

Just copy the src through.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4858>

20 months agonir: Include num_ubos in the printed shader (if nonzero).
Eric Anholt [Tue, 12 May 2020 23:25:13 +0000 (16:25 -0700)]
nir: Include num_ubos in the printed shader (if nonzero).

I keep wanting this number for debugging shaders.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4858>

20 months agoutil/ra: Add [de]serialization support
Jason Ekstrand [Wed, 18 Sep 2019 19:57:11 +0000 (14:57 -0500)]
util/ra: Add [de]serialization support

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5019>

20 months agoutil/vma: Add a debug print helper
Jason Ekstrand [Thu, 12 Sep 2019 23:42:21 +0000 (18:42 -0500)]
util/vma: Add a debug print helper

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5019>

20 months agoutil/vma: Add an option to configure high/low preference
Jason Ekstrand [Thu, 29 Aug 2019 18:04:25 +0000 (13:04 -0500)]
util/vma: Add an option to configure high/low preference

The vma_heap allocator was originally designed to prefer high addresses
in order to find bugs in ANV's high address handling.  However, there
are cases where you might want the allocator to prefer lower addresses
for some reason.  This provides a configure bit for exactly this
purpose.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5019>

20 months agoutil/list: Add list_foreach_entry_from_safe
Caio Marcelo de Oliveira Filho [Tue, 17 Sep 2019 01:21:06 +0000 (18:21 -0700)]
util/list: Add list_foreach_entry_from_safe

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5019>

20 months agoutil/list: Add a list pair iterator
Jason Ekstrand [Mon, 5 Aug 2019 21:19:06 +0000 (16:19 -0500)]
util/list: Add a list pair iterator

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5019>

20 months agoanv: Implement VK_EXT_custom_border_color
Iván Briano [Thu, 23 Apr 2020 00:08:22 +0000 (17:08 -0700)]
anv: Implement VK_EXT_custom_border_color

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4898>

20 months agoanv: Add a way to reserve states from a pool
Iván Briano [Thu, 23 Apr 2020 00:17:38 +0000 (17:17 -0700)]
anv: Add a way to reserve states from a pool

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4898>

20 months agoanv: Disable B5G6R5_UNORM_PACK16
Iván Briano [Tue, 5 May 2020 20:20:48 +0000 (13:20 -0700)]
anv: Disable B5G6R5_UNORM_PACK16

It's not a required format and it causes issues with some features.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4898>

20 months agoanv: use the correct format on Android
Iván Briano [Thu, 7 May 2020 20:34:04 +0000 (13:34 -0700)]
anv: use the correct format on Android

Per https://android.googlesource.com/platform/frameworks/native/+/master/vulkan/libvulkan/swapchain.cpp#745
the format Android requires is R5G6B5, and we have it backwards here.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4898>

20 months agodrirc: Enable glthread for PCSX2
JibbityJobbity [Wed, 13 May 2020 12:38:07 +0000 (00:38 +1200)]
drirc: Enable glthread for PCSX2

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5023>

20 months agopost_version.py: stop adding release candidates to the index and relnotes
Eric Engestrom [Wed, 6 May 2020 22:20:46 +0000 (00:20 +0200)]
post_version.py: stop adding release candidates to the index and relnotes

Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/2870
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4929>

20 months agopost_version.py: invert `is_point` into `is_first_release` to make its purpose clearer
Eric Engestrom [Wed, 6 May 2020 22:19:51 +0000 (00:19 +0200)]
post_version.py: invert `is_point` into `is_first_release` to make its purpose clearer

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4929>

20 months agopost_version.py: fix branch name construction for release candidates
Eric Engestrom [Wed, 6 May 2020 22:16:24 +0000 (00:16 +0200)]
post_version.py: fix branch name construction for release candidates

Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/2870
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4929>

20 months agoglthread: stop using GLenum16 to get correct GL errors for out-of-bounds enums
Marek Olšák [Tue, 12 May 2020 20:22:29 +0000 (16:22 -0400)]
glthread: stop using GLenum16 to get correct GL errors for out-of-bounds enums

Reported by Ian Romanick.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5016>

20 months agoradeonsi: also enable tgsi_to_nir caching for compute shaders
Marek Olšák [Wed, 13 May 2020 17:33:00 +0000 (13:33 -0400)]
radeonsi: also enable tgsi_to_nir caching for compute shaders

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4993>

20 months agoradeonsi: Enable tgsi to nir disk cache
Axel Davy [Wed, 13 May 2020 16:48:48 +0000 (18:48 +0200)]
radeonsi: Enable tgsi to nir disk cache

Enable the tgsi to nir cache for radeonsi.

Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4993>

20 months agost/nine: Enable ttn cache
Axel Davy [Tue, 12 May 2020 20:26:46 +0000 (22:26 +0200)]
st/nine: Enable ttn cache

A trace of a Hat in Time, which builds thousands of shaders
takes 339 seconds to run the second time without this patch,
and 41 seconds with it (basically there is no more loading times).

Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4993>

20 months agottn: Implement disk cache
Axel Davy [Tue, 12 May 2020 20:24:32 +0000 (22:24 +0200)]
ttn: Implement disk cache

ttn is slow, let's disk cache it.

Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4993>

20 months agottn: Add new allow_disk_cache parameter
Axel Davy [Tue, 12 May 2020 19:52:40 +0000 (21:52 +0200)]
ttn: Add new allow_disk_cache parameter

For now this parameter doesn't do anything.
It means the implementation is allowed to use
a cache on disk.

Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4993>

20 months agofreedreno/a6xx: Fix UBWC mipmapping height alignment.
Eric Anholt [Wed, 6 May 2020 21:50:34 +0000 (14:50 -0700)]
freedreno/a6xx: Fix UBWC mipmapping height alignment.

After fixing the power of two sizing, pitches worked, but 1-pixel high and
unaligned height miplevels were off.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4931>

20 months agofreedreno/a6xx: Fix UBWC mipmap sizing.
Eric Anholt [Wed, 6 May 2020 19:21:01 +0000 (12:21 -0700)]
freedreno/a6xx: Fix UBWC mipmap sizing.

The HW requires a log2 width/height of the level 0 meta_* size in the
descriptors, making it pretty clear that UBWC mipmapping is all
power-of-two sized.  Fixes a bunch of failures in the upcoming unit UBWC
layout unit tests.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4931>

20 months agofreedreno/a6xx: Fix UBWC blockheight for RG8.
Eric Anholt [Wed, 6 May 2020 22:38:46 +0000 (15:38 -0700)]
freedreno/a6xx: Fix UBWC blockheight for RG8.

Using texturator on a P3A at 1024x1024, RG8 has log2w/h of 6x7 instead of
R16I/UI's 6x8.  The other blockw/h I verified other than cpp=1
(R8/R8I/R8UI didn't use UBWC) and 32 (would need a bigger type).

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4931>

20 months agofreedreno: Pull the tile_alignment lookup for a layout to a helper.
Eric Anholt [Tue, 12 May 2020 20:49:37 +0000 (13:49 -0700)]
freedreno: Pull the tile_alignment lookup for a layout to a helper.

The r8g8 case UBWC alignment will be changing in the next commit, so
fdl6_get_ubwc_blockwidth needs to start paying attention to r8g8 too.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4931>

20 months agofreedreno/a6xx: Add a testcase for UBWC buffer sharing.
Eric Anholt [Thu, 7 May 2020 01:01:36 +0000 (18:01 -0700)]
freedreno/a6xx: Add a testcase for UBWC buffer sharing.

These offsets are hand-computed referencing msm_media_info.h, and match
our driver's current behavior.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4931>

20 months agofreedreno/a6xx: Improve layout testcase logging for UBWC fails.
Eric Anholt [Wed, 6 May 2020 18:40:51 +0000 (11:40 -0700)]
freedreno/a6xx: Improve layout testcase logging for UBWC fails.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4931>

20 months agofreedreno/a4xx+: Increase max texture size to 16384.
Eric Anholt [Wed, 6 May 2020 21:43:02 +0000 (14:43 -0700)]
freedreno/a4xx+: Increase max texture size to 16384.

Noticed when poking around with texture layouts and found that my big
texture layout from the blob buffer overflowed.  Values come from
http://vulkan.gpuinfo.org for Adreno 418, 512, 630.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4931>

20 months agonir: reset ssa-defs as non-divergent during divergence analysis instead of upfront
Daniel Schürmann [Tue, 21 Apr 2020 16:07:56 +0000 (17:07 +0100)]
nir: reset ssa-defs as non-divergent during divergence analysis instead of upfront

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4062>

20 months agonir: simplify phi handling in divergence analysis
Daniel Schürmann [Wed, 5 Feb 2020 17:36:34 +0000 (18:36 +0100)]
nir: simplify phi handling in divergence analysis

This patch adds some control flow information to the
state to keep track whether a loop contains divergent
continue or break statements to not having to
recalculate this property for every phi.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4062>

20 months agonir: rework phi handling in divergence analysis
Daniel Schürmann [Wed, 5 Feb 2020 12:08:27 +0000 (13:08 +0100)]
nir: rework phi handling in divergence analysis

This patch splits the visit_phi() function into
three different ones according to the kind of phi
(merge-node, loop-header or loop-exit) and calls
them when visiting the cf_nodes.
This allows to revisit loops if the loop header's
phis have changed, only.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4062>

20 months agonir: refactor divergence analysis state
Daniel Schürmann [Wed, 5 Feb 2020 10:53:04 +0000 (11:53 +0100)]
nir: refactor divergence analysis state

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4062>

20 months agonir: add nir_intrinsic_elect to divergence analysis
Daniel Schürmann [Thu, 6 Feb 2020 16:35:25 +0000 (17:35 +0100)]
nir: add nir_intrinsic_elect to divergence analysis

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4062>

20 months agonir: Make "divergent" a property of an SSA value
Jason Ekstrand [Tue, 15 Oct 2019 19:48:10 +0000 (14:48 -0500)]
nir: Make "divergent" a property of an SSA value

v2: fix usage in ACO (by Daniel Schürmann)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4062>

20 months agogallium: remove more "state tracker" occurences
Marek Olšák [Tue, 12 May 2020 20:19:10 +0000 (16:19 -0400)]
gallium: remove more "state tracker" occurences

Trivial.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4902>

20 months agogallium: rename PIPE_RESOURCE_FLAG_ST_PRIV to FRONTEND_PRIV
Marek Olšák [Wed, 4 Dec 2019 02:22:43 +0000 (21:22 -0500)]
gallium: rename PIPE_RESOURCE_FLAG_ST_PRIV to FRONTEND_PRIV

Acked-by: Eric Anholt <eric@anholt.net>
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4902>

20 months agogallium: change comments to remove 'state tracker'
Marek Olšák [Wed, 4 Dec 2019 01:38:14 +0000 (20:38 -0500)]
gallium: change comments to remove 'state tracker'

Acked-by: Eric Anholt <eric@anholt.net>
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4902>

20 months agogallium: rename 'state tracker' to 'frontend'
Marek Olšák [Tue, 3 Dec 2019 23:01:31 +0000 (18:01 -0500)]
gallium: rename 'state tracker' to 'frontend'

Acked-by: Eric Anholt <eric@anholt.net>
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4902>

20 months agotu: Implement fallback linear staging blit for CopyImage
Connor Abbott [Tue, 12 May 2020 10:47:03 +0000 (12:47 +0200)]
tu: Implement fallback linear staging blit for CopyImage

Also, rewrite the format decision code so that we correctly decide when
the linear fallback is needed, even if UBWC is disabled. As part of
that, I also moved around some of the code to handle compressed formats
to make sure that copying compressed formats with a linear staging blit
works (this is now possible since we started allowing tiled compressed
textures).

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5007>

20 months agotu: Add noubwc debug flag to disable UBWC
Connor Abbott [Tue, 12 May 2020 15:45:26 +0000 (17:45 +0200)]
tu: Add noubwc debug flag to disable UBWC

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5007>

20 months agotu: Add a "scratch bo" allocation mechanism
Connor Abbott [Mon, 11 May 2020 16:46:04 +0000 (18:46 +0200)]
tu: Add a "scratch bo" allocation mechanism

This is simpler than a full-blown memory reuse mechanism, but is good
enough to make sure that repeatedly doing a copy that requires the
linear staging buffer workaround won't use excessive memory or be slowed
down due to repeated allocations.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5007>

20 months agoaco: improve phi affinities with p_split_vector
Rhys Perry [Fri, 8 May 2020 18:01:09 +0000 (19:01 +0100)]
aco: improve phi affinities with p_split_vector

Totals from 5860 (4.59% of 127638) affected shaders:
VGPRs: 460212 -> 460216 (+0.00%)
CodeSize: 65554356 -> 65464816 (-0.14%)
Instrs: 12655972 -> 12633578 (-0.18%)
Copies: 1309994 -> 1292163 (-1.36%)

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4990>