mesa.git
18 months agoradv: clean up tessellation state emission
Samuel Pitoiset [Thu, 9 Jul 2020 08:42:57 +0000 (10:42 +0200)]
radv: clean up tessellation state emission

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agoradv: remove unnecessary radv_tessellation_state::lds_size
Samuel Pitoiset [Thu, 9 Jul 2020 08:36:28 +0000 (10:36 +0200)]
radv: remove unnecessary radv_tessellation_state::lds_size

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agoradv: set LDS TCS size at shaders creation for GFX9+
Samuel Pitoiset [Thu, 9 Jul 2020 08:33:16 +0000 (10:33 +0200)]
radv: set LDS TCS size at shaders creation for GFX9+

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agoradv: align the LDS size in calculate_tess_lds_size()
Samuel Pitoiset [Thu, 9 Jul 2020 08:06:43 +0000 (10:06 +0200)]
radv: align the LDS size in calculate_tess_lds_size()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agoradv: remove one unnecessary param to radv_generate_graphics_pipeline_key()
Samuel Pitoiset [Thu, 9 Jul 2020 06:40:13 +0000 (08:40 +0200)]
radv: remove one unnecessary param to radv_generate_graphics_pipeline_key()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agoradv: remove no-op si_multiwave_lds_size_workaround()
Samuel Pitoiset [Wed, 8 Jul 2020 19:45:09 +0000 (21:45 +0200)]
radv: remove no-op si_multiwave_lds_size_workaround()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agoradv: remove unnecessary radv_tessellation_state::num_patches
Samuel Pitoiset [Wed, 8 Jul 2020 19:35:23 +0000 (21:35 +0200)]
radv: remove unnecessary radv_tessellation_state::num_patches

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agoradv: clean up radv_compute_generate_pm4()
Samuel Pitoiset [Wed, 8 Jul 2020 15:48:49 +0000 (17:48 +0200)]
radv: clean up radv_compute_generate_pm4()

For consistency regarding how the graphics pipeline is built.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agoradv: reduce the number of allocated dwords for compute CS
Samuel Pitoiset [Wed, 8 Jul 2020 15:49:14 +0000 (17:49 +0200)]
radv: reduce the number of allocated dwords for compute CS

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agoradv: clean up PA_SC_CLIPRECT_RULE emission
Samuel Pitoiset [Wed, 8 Jul 2020 15:06:10 +0000 (17:06 +0200)]
radv: clean up PA_SC_CLIPRECT_RULE emission

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agoradv: clean up VGT_SHADER_STAGES_EN emission
Samuel Pitoiset [Wed, 8 Jul 2020 15:00:19 +0000 (17:00 +0200)]
radv: clean up VGT_SHADER_STAGES_EN emission

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agoradv: emit PA_SC_LINE_CNTL as part of the rasterization state
Samuel Pitoiset [Wed, 8 Jul 2020 14:51:54 +0000 (16:51 +0200)]
radv: emit PA_SC_LINE_CNTL as part of the rasterization state

While we are at it, remove one useless field in radv_multisample_state.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agoradv: emit more invariant registers as part of the initial gfx state
Samuel Pitoiset [Wed, 8 Jul 2020 13:59:42 +0000 (15:59 +0200)]
radv: emit more invariant registers as part of the initial gfx state

This reduces the number of emitted packets for pipelines.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agoradv: remove outdated TODO related to PA_SU_VTX_CNTL.PIX_CENTER
Samuel Pitoiset [Wed, 8 Jul 2020 13:56:47 +0000 (15:56 +0200)]
radv: remove outdated TODO related to PA_SU_VTX_CNTL.PIX_CENTER

It should be always 1, nothing more to check.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agoradv: remove set but unused radv_pipeline::vertex_elements
Samuel Pitoiset [Wed, 8 Jul 2020 13:02:00 +0000 (15:02 +0200)]
radv: remove set but unused radv_pipeline::vertex_elements

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agoradv: remove declared but unused radv_pipeline::is_dual_src
Samuel Pitoiset [Wed, 8 Jul 2020 12:58:58 +0000 (14:58 +0200)]
radv: remove declared but unused radv_pipeline::is_dual_src

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5837>

18 months agobin/khronos-update: add workaround for python bug 9625
Eric Engestrom [Fri, 24 Jul 2020 12:20:44 +0000 (14:20 +0200)]
bin/khronos-update: add workaround for python bug 9625

The bug causes `choices` to break `nargs='*'`.

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6057>

18 months agobin/khronos-update: add support for the SPIRV files
Eric Engestrom [Tue, 21 Jul 2020 00:27:33 +0000 (02:27 +0200)]
bin/khronos-update: add support for the SPIRV files

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6004>

18 months agobin/khronos-update: having a folder in include/ is not a requirement
Eric Engestrom [Tue, 21 Jul 2020 00:26:58 +0000 (02:26 +0200)]
bin/khronos-update: having a folder in include/ is not a requirement

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6004>

18 months agov3d: Retry with the fallback scheduler when RA fails
Neil Roberts [Fri, 17 Jul 2020 13:32:19 +0000 (15:32 +0200)]
v3d: Retry with the fallback scheduler when RA fails

v3d_compile is now split out into a helper function that gets called a
second time if compilation fails the first time with the result
reporting the register allocation failed. The second time it is run with
the fallback scheduler to try and increase the chances of successfully
allocating the registers.

v2: Add a performance debug message when using the fallback scheduler.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5953>

18 months agov3d: Changed v3d_compile:failed to an enum
Neil Roberts [Fri, 17 Jul 2020 10:49:59 +0000 (12:49 +0200)]
v3d: Changed v3d_compile:failed to an enum

Instead of just having a bool status for the failure, there is now an
enum so that the compilation can report a more detailed status.
Currently this is only used to report whether the failure was due to
failed register allocation. The “failed” bool doesn’t seem to actually
have been used anywhere so this doesn’t really change a lot.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5953>

18 months agonir/schedule: Add an option for a fallback scheduling algorithm
Neil Roberts [Fri, 17 Jul 2020 09:00:53 +0000 (11:00 +0200)]
nir/schedule: Add an option for a fallback scheduling algorithm

The current scheduling algorithm favors parallelism a bit too
aggressively and sometimes generates shaders that fail register
allocation. This happens even if the threshold is set to zero to force
it to always use the CSR instruction choosing algorithm.

This patch adds an option to use an even more aggressive fallback that
just always picks the instruction with the shortest maximum delay in the
hope that that will generate the least register pressure. The intention
is to use this as a last resort after register allocation fails in order
to at least have a working shader.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Acked-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5953>

18 months agov3d: Mark scheduling dependency for prim id and first output
Neil Roberts [Fri, 17 Jul 2020 07:49:33 +0000 (09:49 +0200)]
v3d: Mark scheduling dependency for prim id and first output

The input primitive ID is read from the VPM in the same memory segment
as the outputs. This means that writing the GS header to VPM location 0
needs to be done after reading the primitive ID. This patch adds a
dependency between the load_primitive_id intrinsic and the store_output
intrinsic for location 0 to stop the scheduler from reordering them.

v2: Use an enum for the dependency class number.
v3: Add "GS" to the class enum name.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5953>

18 months agonir/schedule: Add a callback for backend-specific dependencies
Neil Roberts [Fri, 17 Jul 2020 07:24:02 +0000 (09:24 +0200)]
nir/schedule: Add a callback for backend-specific dependencies

Adds a callback function to nir_schedule_options to give the backend a
chance to add custom dependencies between certain intrinsics. The
callback can assign a class number to the intrinsic and then set a read
or write dependency on that class.

v2: Use a linked-list of schedule nodes for the dependency classes
    instead of a fixed-sized array.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5953>

18 months agonir/schedule: Store a pointer to the options struct in scoreboard
Neil Roberts [Fri, 17 Jul 2020 07:17:29 +0000 (09:17 +0200)]
nir/schedule: Store a pointer to the options struct in scoreboard

Instead of copying the individual members of nir_schedule_options into
the scoreboard, it now just keeps a pointer to the options. This avoids
the duplicated comments and makes it easier to add more options later.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5953>

18 months agonir/scheduler: Move nir_scheduler to its own header
Neil Roberts [Fri, 17 Jul 2020 07:08:47 +0000 (09:08 +0200)]
nir/scheduler: Move nir_scheduler to its own header

nir_schedule already has a struct for options which makes it more than
just a function declaration. Later patches intend to add more structs to
complement these options. In order to make the code easier to manage,
this moves the nir_scheduler-related parts out of nir.h to their own
header.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5953>

18 months agospirv: Rework our handling of images and samplers
Jason Ekstrand [Fri, 29 May 2020 21:45:21 +0000 (16:45 -0500)]
spirv: Rework our handling of images and samplers

Previously, objects of type OpTypeImage or OpTypeSampler were treated as
vtn_pointers and objects of type OpTypeSampledImage were a special-use
vtn_sampled_image struct.  This commit changes that so that all of those
objects are stored in vtn_ssa_values.  Each of images, samplers, and
sampled images, are stored as a scalar or vector nir_ssa_def whose
components are NIR deref values.  We now use vtn_type_get_nir_type to
re-resolve those as-needed into GLSL sampler types for NIR.

This simplification has a number of benefits:

 1. We can git rid of the rest of our special-cases for handling images
    and samplers in function arguments.  Now that they're treated as
    structs at the glsl_type level, the generic paths can handle images
    and samplers.

 2. We can now construct composite values containing images and samplers
    internally.  It's unclear from the SPIR-V spec whether or not this
    is allowed and it's not a pattern that GLSLang currently generates
    thanks to GLSL rules.  However, if we do start seeing SPIR-V that
    contains such composites, we should now be able to handle it.

 3. SPIR-V OpNull and OpUndef instructions can now create samplers,
    images, and sampled images.  The NIR generated won't likely be fully
    valid but, given a NIR pass to do something sensible, it should be a
    thing we can compile.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agoanv,turnip,radv,clover,glspirv: Run nir_copy_prop before nir_opt_deref
Jason Ekstrand [Thu, 11 Jun 2020 18:29:02 +0000 (13:29 -0500)]
anv,turnip,radv,clover,glspirv: Run nir_copy_prop before nir_opt_deref

We're about to make the SPIR-V -> NIR path generate a bit more complex
SSA chains for certain derefs.  This will ensure we don't regress anyone
when we start making vec2's of derefs.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: More heavily use vtn_ssa_value in function parameter handling
Jason Ekstrand [Sat, 30 May 2020 01:48:23 +0000 (20:48 -0500)]
spirv: More heavily use vtn_ssa_value in function parameter handling

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Remove a dead case in function parameter handling
Jason Ekstrand [Sat, 30 May 2020 02:02:31 +0000 (21:02 -0500)]
spirv: Remove a dead case in function parameter handling

Ever since 31a7476335f911a, we've set something for vtn_type::type for
all pointer types.  For logical pointer types, it's uint32_t.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Add a helper for getting the NIR type of a vtn_type
Jason Ekstrand [Fri, 29 May 2020 22:57:34 +0000 (17:57 -0500)]
spirv: Add a helper for getting the NIR type of a vtn_type

There are a few cases, atomic counters being one example, where the type
used by vtn_ssa_value is not the same as the type we want NIR to use in
derefs and variables.  To solve this, we add a helper which converts
between the types for us.  In the next commit, we'll be adding another
major user of this: images and samplers.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Give atomic counters their own variable mode
Jason Ekstrand [Fri, 29 May 2020 23:39:30 +0000 (18:39 -0500)]
spirv: Give atomic counters their own variable mode

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Drop the sampled boolean from vtn_type
Jason Ekstrand [Fri, 29 May 2020 21:36:54 +0000 (16:36 -0500)]
spirv: Drop the sampled boolean from vtn_type

It was set but never used.  We always check the glsl_type instead.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Add better checks for SSA value types
Jason Ekstrand [Fri, 29 May 2020 21:10:28 +0000 (16:10 -0500)]
spirv: Add better checks for SSA value types

Primarily, we check for two things:

 1. That we only ever add SSA values via vtn_push_ssa_value and
    vtn_copy_value.

 2. That the type of the SSA value matches the SPIR-V destination type.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Hand-roll fewer vtn_ssa_value creations
Jason Ekstrand [Sat, 30 May 2020 17:27:54 +0000 (12:27 -0500)]
spirv: Hand-roll fewer vtn_ssa_value creations

Previously, we created our vtn_ssa_value in _vtn_variable_load_store
manually as we did the recursive load/store.  Instead, we now create the
SSA value before calling into the recursive function.  This is a tiny
bit less efficient but it removes a case of hand-rolling vtn_ssa_value
creation.  For symmetry, we make _vtn_block_load_store assume the value
is already created.  Finally, we remove a trivial hand-rolled case in
vtn_composite_extract.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Simplify vtn_ssa_value creation
Jason Ekstrand [Sat, 30 May 2020 17:23:50 +0000 (12:23 -0500)]
spirv: Simplify vtn_ssa_value creation

For three different functions which create vtn_ssa_values, we had three
completely different implementations.  This unifies them all to roughly
the same algorithm.  While we're at it, we take advantage of the
nir_build_imm helper to avoid some extra code in vtn_const_ssa_value.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv/subgroups: Refactor to use vtn_push_ssa
Jason Ekstrand [Fri, 29 May 2020 19:40:12 +0000 (14:40 -0500)]
spirv/subgroups: Refactor to use vtn_push_ssa

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv/subgroups: Stop incrementing w
Jason Ekstrand [Sat, 30 May 2020 16:07:46 +0000 (11:07 -0500)]
spirv/subgroups: Stop incrementing w

The w++ is to handle a differences between the KHR extension and Vulkan
1.1 feature where the Vulkan 1.1 instructions take an scope parameter.
While incrementing w technically works, it's really subtle and very easy
to miss when reading the code.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv/glsl450: Use vtn_push_ssa_value
Jason Ekstrand [Fri, 29 May 2020 21:04:02 +0000 (16:04 -0500)]
spirv/glsl450: Use vtn_push_ssa_value

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv/alu: Use vtn_push_ssa_value
Jason Ekstrand [Fri, 29 May 2020 20:57:42 +0000 (15:57 -0500)]
spirv/alu: Use vtn_push_ssa_value

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Refactor vtn_push_ssa
Jason Ekstrand [Fri, 29 May 2020 20:44:10 +0000 (15:44 -0500)]
spirv: Refactor vtn_push_ssa

We rename it to vtn_push_ssa_value, move it to spirv_to_nir, and remove
the unnecessary type parameter.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Use the new helpers in OpConvertUToPtr/PtrToU
Jason Ekstrand [Thu, 28 May 2020 23:39:45 +0000 (18:39 -0500)]
spirv: Use the new helpers in OpConvertUToPtr/PtrToU

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Add a vtn_get_nir_ssa helper
Jason Ekstrand [Wed, 27 May 2020 23:28:18 +0000 (18:28 -0500)]
spirv: Add a vtn_get_nir_ssa helper

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv/amd: Use vtn_push_nir_ssa
Jason Ekstrand [Wed, 27 May 2020 23:08:20 +0000 (18:08 -0500)]
spirv/amd: Use vtn_push_nir_ssa

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Add a vtn_push_nir_ssa helper
Jason Ekstrand [Wed, 27 May 2020 22:49:47 +0000 (17:49 -0500)]
spirv: Add a vtn_push_nir_ssa helper

This makes it easy to write a simple NIR SSA value

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Rename push_value_pointer to push_pointer
Jason Ekstrand [Wed, 27 May 2020 22:55:10 +0000 (17:55 -0500)]
spirv: Rename push_value_pointer to push_pointer

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Add a helpers for getting types of values
Jason Ekstrand [Wed, 27 May 2020 23:33:47 +0000 (18:33 -0500)]
spirv: Add a helpers for getting types of values

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Use nir_bany/ball for OpAny/All
Jason Ekstrand [Fri, 29 May 2020 04:00:21 +0000 (23:00 -0500)]
spirv: Use nir_bany/ball for OpAny/All

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Clean up OpSignBitSet
Jason Ekstrand [Thu, 28 May 2020 22:14:30 +0000 (17:14 -0500)]
spirv: Clean up OpSignBitSet

For some reason, we were doing a signed shift vectors and an unsigned
shift for scalars.  We then plug it into i2b so it should make no
difference whatsoever.  The fact that we're doing different things for
vectors vs. scalars is bonkers.  Let's simplify the code a bit.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Fix indentation in vtn_handle_ptr
Jason Ekstrand [Wed, 27 May 2020 22:41:23 +0000 (17:41 -0500)]
spirv: Fix indentation in vtn_handle_ptr

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Drop the void *ptr from vtn_value
Jason Ekstrand [Wed, 27 May 2020 21:15:39 +0000 (16:15 -0500)]
spirv: Drop the void *ptr from vtn_value

It isn't being used for anything.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agospirv: Simplify our handling of NonUniform
Jason Ekstrand [Thu, 28 May 2020 23:32:01 +0000 (18:32 -0500)]
spirv: Simplify our handling of NonUniform

The original implementation of SPV_EXT_descriptor_indexing was extremely
paranoid about the NonUniform qualifier, trying to fetch it from every
possible location and propagate it through access chains etc.  However,
the Vulkan spec is quite nice to us on this and has very strict rules
for where the NonUniform decoration has to be placed.  For image and
texture operations, we can search for the decoration on the spot when we
process the image or texture op.  For pointers, we continue putting it
on the pointer but we don't bother trying to do anything silly like
propagate it through casts.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5278>

18 months agonir/vtn: Add support for 8 and 16 vector ball/bany
Jesse Natalie [Tue, 23 Jun 2020 12:44:11 +0000 (05:44 -0700)]
nir/vtn: Add support for 8 and 16 vector ball/bany

Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6030>

18 months agonir: Support 8 and 16 component vectors for reduceable intrinsics
Jesse Natalie [Mon, 22 Jun 2020 23:48:43 +0000 (16:48 -0700)]
nir: Support 8 and 16 component vectors for reduceable intrinsics

Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6030>

18 months agonir: Support algebraic opts on vectors larger than 4
Jesse Natalie [Mon, 22 Jun 2020 18:49:41 +0000 (11:49 -0700)]
nir: Support algebraic opts on vectors larger than 4

Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6030>

18 months agonir: Support vec8/vec16 in nir_lower_bit_size
Jesse Natalie [Mon, 1 Jun 2020 22:46:03 +0000 (15:46 -0700)]
nir: Support vec8/vec16 in nir_lower_bit_size

Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6030>

18 months agofreedreno: sync registers from envytools
Rob Clark [Thu, 23 Jul 2020 20:02:08 +0000 (13:02 -0700)]
freedreno: sync registers from envytools

Pull in a bunch of fixes and updates.. mostly using varset correctly,
and fixes for implicit bools.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6052>

18 months agotu: Enable VK_EXT_depth_clip_enable
Connor Abbott [Thu, 23 Jul 2020 09:44:40 +0000 (11:44 +0200)]
tu: Enable VK_EXT_depth_clip_enable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6047>

18 months agoaco: ensure readfirstlane subdword operands are always dword aligned
Daniel Schürmann [Thu, 23 Jul 2020 07:58:11 +0000 (09:58 +0200)]
aco: ensure readfirstlane subdword operands are always dword aligned

Cc: 20.1 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6024>

18 months agoaco: prevent infinite recursion in RA for subdword variables
Daniel Schürmann [Thu, 23 Jul 2020 07:41:07 +0000 (09:41 +0200)]
aco: prevent infinite recursion in RA for subdword variables

Cc: 20.1 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6024>

18 months agoaco: don't split store data if it was already split into more elements
Daniel Schürmann [Wed, 22 Jul 2020 16:41:53 +0000 (18:41 +0200)]
aco: don't split store data if it was already split into more elements

Cc: 20.1 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6024>

18 months agoaco: ensure to not extract more components than have been fetched
Daniel Schürmann [Wed, 22 Jul 2020 14:41:41 +0000 (16:41 +0200)]
aco: ensure to not extract more components than have been fetched

Fixes: 7015d2c249e1f7814bf5681ccd049e49e4d6495c ('aco: fix scratch loads which cross element_size boundaries')
Cc: 20.1 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6024>

18 months agoradv: Add timeline syncobj for timeline semaphores.
Bas Nieuwenhuizen [Thu, 16 Jul 2020 00:44:22 +0000 (02:44 +0200)]
radv: Add timeline syncobj for timeline semaphores.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5600>

18 months agoradv: Add winsys functions for timeline syncobj.
Bas Nieuwenhuizen [Thu, 16 Jul 2020 00:43:32 +0000 (02:43 +0200)]
radv: Add winsys functions for timeline syncobj.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5600>

18 months agoradv: Add winsys support for submitting timeline syncobj.
Bas Nieuwenhuizen [Thu, 16 Jul 2020 00:42:56 +0000 (02:42 +0200)]
radv: Add winsys support for submitting timeline syncobj.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5600>

18 months agoradv: Add thread for timeline syncobj submission.
Bas Nieuwenhuizen [Mon, 22 Jun 2020 20:07:46 +0000 (22:07 +0200)]
radv: Add thread for timeline syncobj submission.

For cross-process timelines we have to have a thread to wait
till the requested points become available.

The functions actually dealing with timeline semaphores stubbed out, to
implement in the next patch. As such the thread code shouldn't trigger
yet.

The core idea is that we still use the refcount mechanism that we use with
emulated timelines, though the native timeline syncobj don't participate
in the refcounting. This way we keep the ordering of submission in a queue
as each submission is also blocked by its predecessor.

Where we change behavior is when the number of blockers reaches 0. In the
new code we check if we need to wait for the timeline semaphores to
be available and if so we won't execute the submission immediately but
pass it to the submission thread.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5600>

18 months agoradv/winsys: Add binary syncobj ABI changes for timeline semaphores.
Bas Nieuwenhuizen [Sun, 21 Jun 2020 23:11:03 +0000 (01:11 +0200)]
radv/winsys: Add binary syncobj ABI changes for timeline semaphores.

To facilitate cross-process timeline semaphores we have to deal with
the fact that the syncobj signal operation might be submitted a
small finite time after the wait operation.

For that we start using DRM_SYNCOBJ_WAIT_FLAGS_WAIT_FOR_SUBMIT during
the wait operation so we properly wait instead of returning an error.

Furthermore, to make this effective for syncobjs that get reused we
actually have to reset them after the wait. Otherwise the wait before
submit would get the previous fence instead of waiting for the new
thing to submit.

The obvious choice, resetting the syncobj after the CS submission
has 2 issues though:

1) If the same semaphore is used for wait and signal we can't reset it.
   This is solvable by only resetting the semaphores that are not in the
   signal list.
2) The submitted work might be complete before we get to resetting the
   syncobj. If there is a cycle of submissions that signals it again and
   finishes before we get to the reset we are screwed.

Solution:
Copy the fence into a new syncobj and reset the old syncobj before
submission. Yes I know it is more syscalls :( At least I reduced the
alloc/free overhead by keeping a cache of temporary syncobjs.

This also introduces a syncobj_reset_count as we don't want to reset
syncobjs that are part of an emulated timeline semaphore. (yes, if
the kernel supports timeline syncobjs we should use those instead,
but those still need to be implemented and if we depend on them in
this patch ordering dependencies get hard ...)

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5600>

18 months agoamd: Add detection of timeline semaphore support.
Bas Nieuwenhuizen [Sun, 21 Jun 2020 21:22:00 +0000 (23:22 +0200)]
amd: Add detection of timeline semaphore support.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5600>

18 months agonir/ lower_int_to_float: Handle umax and umin
Andreas Baierl [Thu, 23 Jul 2020 05:43:09 +0000 (07:43 +0200)]
nir/ lower_int_to_float: Handle umax and umin

8e1b75b3 introduced umax/umin in order to lower iand/ior for (n)eq zero.
That breaks the lower_int_to_float pass, because umax and umin weren't
handled there.

Tested with lima. The other users of nir_lower_int_to_float
(etnaviv, freedreno) should also have that issue.

Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Signed-off-by: Andreas Baierl <ichgeh@imkreisrum.de>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6043>

18 months agoci: Use half as many parallel softpipe / virgl test jobs
Michel Dänzer [Tue, 21 Jul 2020 14:36:28 +0000 (16:36 +0200)]
ci: Use half as many parallel softpipe / virgl test jobs

We're now using at least twice as many CPU cores per job (on shared
runners), so they only take about half as long, and should still be
under 10 minutes.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6011>

18 months agoci: Do not mark container / pages jobs as interruptible
Michel Dänzer [Tue, 21 Jul 2020 14:13:37 +0000 (16:13 +0200)]
ci: Do not mark container / pages jobs as interruptible

If another MR was merged while these were still running for the main
project, the result could be no updated images in the main project
registry (forcing a rebuild of the new images in all forked projects) or
an outdated Mesa website.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6011>

18 months agoci: Use FDO_CI_CONCURRENT in run-shader-db.sh as well
Michel Dänzer [Tue, 21 Jul 2020 14:03:57 +0000 (16:03 +0200)]
ci: Use FDO_CI_CONCURRENT in run-shader-db.sh as well

Noticed while checking job logs for it being used elsewhere.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6011>

18 months agoci: Namespace trace artifacts to the job number
Tomeu Vizoso [Wed, 22 Jul 2020 15:39:15 +0000 (17:39 +0200)]
ci: Namespace trace artifacts to the job number

Put artifacts in a per-job folder, because if a job is retried then it
will try to upload a file to the same key and fail with the following
error:

403 Client Error: Forbidden for url:
https://minio-packet.freedesktop.org/artifacts/daenzer/mesa/180609/gl-panfrost-t860/results.yml

Also, to prevent in the future similar clashes if several trace files
share the same name, upload the images with their checksums as their
names. This will also make it easier to fetch images for comparison with
the references.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Andres Gomez <agomez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6029>

18 months agonir_ allow nir_lower_clip_halfz to run in tess eval shader
Mike Blumenkrantz [Fri, 17 Jul 2020 13:46:47 +0000 (09:46 -0400)]
nir_ allow nir_lower_clip_halfz to run in tess eval shader

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6027>

18 months agonir: allow lower_psiz_mov to run in tessellation stages
Mike Blumenkrantz [Thu, 16 Jul 2020 13:38:46 +0000 (09:38 -0400)]
nir: allow lower_psiz_mov to run in tessellation stages

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6027>

18 months agoRevert "llvmpipe: Use the default behavior of ALLOW_MAPPED_BUFFERS."
Dave Airlie [Thu, 23 Jul 2020 05:50:19 +0000 (15:50 +1000)]
Revert "llvmpipe: Use the default behavior of ALLOW_MAPPED_BUFFERS."

This reverts commit 6ec49066498097c2647d030ebe1b99a00bc38734.

This broke:
GTF-GL45.gtf21.GL3Tests.texture_lod_bias.*

not sure why but revert for now.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6042>

18 months agollvmpipe/ms: fix sign extension bug in rasterizer.
Dave Airlie [Tue, 21 Jul 2020 06:32:08 +0000 (16:32 +1000)]
llvmpipe/ms: fix sign extension bug in rasterizer.

/glcts --deqp-surface-width=1024 --deqp-surface-height=64  --deqp-case=KHR-GL45.texture_view.view_sampling  --deqp-surface-type=fbo
was failing but only for width 1024.

The test was filling a 4x4 ms texture, but leaving the viewport set to 1024x64.

This was resulting in this code incorrectly sign extending a value, and passing
it into the mask generator and getting the wrong values. Explicit cast
avoids the sign extension and fixes the above test.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6006>

18 months agointel/ir/gen12+: Work around FS performance regressions due to SIMD32 discard divergence.
Francisco Jerez [Sun, 31 May 2020 21:56:40 +0000 (14:56 -0700)]
intel/ir/gen12+: Work around FS performance regressions due to SIMD32 discard divergence.

This avoids some performance regressions on Gen12 platforms caused by
SIMD32 fragment shaders reported in titles like Dota2, TF2, Xonotic,
and GFXBench5 Car Chase and Aztec Ruins.

The most obvious pattern in the regressing shaders I identified among
these workloads is that they all had non-uniform discard statements,
which are handled rather optimistically by the current IR analysis
pass: No penalty is currently applied to the SIMD32 variant of the
shader in the form of differing branching weights like we do for other
control flow instructions in order to account for the greater
likelihood of divergence of a SIMD32 shader.

Simply changing that by giving the same treatment to discard
statements as we give to other branching instructions seemed to hurt
more than it helped on platforms earlier than Gen12, since it reversed
most of the improvement obtained from SIMD32 fragment shaders in
Manhattan for no measurable benefit in other workloads (Manhattan has
a handful of shaders with statically non-uniform discard statements
which actually perform better in SIMD32 mode due to their approximate
dynamic uniformity).  For that reason this change is applied to Gen12+
platforms only.

I've been running a number of tests trying to understand the
difference in behavior between Gen12 and earlier platforms, and most
of the evidence I've gathered seems to point at EU fusion being the
culprit: Unlike previous generations, on Gen12 EUs are arranged in
pairs which execute instructions in lockstep, giving an effective warp
size of 64 threads in SIMD32 mode, which seems to increase the
likelihood for control flow divergence in some of the affected shaders
significantly.

Fixes: 188a3659aea6dec9acf1 "intel/ir: Import shader performance analysis pass."
Reported-by: Caleb Callaway <caleb.callaway@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5910>

18 months agoglx: Fix build and warnings with -Dglx=dri -Dglx-direct=false
Adam Jackson [Thu, 16 Jul 2020 21:29:26 +0000 (17:29 -0400)]
glx: Fix build and warnings with -Dglx=dri -Dglx-direct=false

Acked-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5958>

18 months agosoftpipe: Enable PIPE_CAP_TGSI_ANY_REG_AS_ADDRESS;
Eric Anholt [Wed, 15 Jan 2020 21:07:43 +0000 (13:07 -0800)]
softpipe: Enable PIPE_CAP_TGSI_ANY_REG_AS_ADDRESS;

tgsi_exec.c uses the generic src load path for indirects, so we don't
actually need addr regs.  Saves extra intructions.

shader-db results:
total instructions in shared programs: 3346685 -> 3249052 (-2.92%)
instructions in affected programs: 961832 -> 864199 (-10.15%)

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6018>

18 months agosoftpipe: Enable PIPE_CAP_TGSI_TEXCOORD.
Eric Anholt [Thu, 23 Jan 2020 19:43:04 +0000 (11:43 -0800)]
softpipe: Enable PIPE_CAP_TGSI_TEXCOORD.

The tgsi_exec path can handle it, and otherwise when we start using NIR
our MAX_VARYINGS value will cause us to have VARYING_SLOT_VARx above the
maximum.

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6018>

18 months agosoftpipe: Add support for reporting shader-db output.
Eric Anholt [Tue, 21 Jul 2020 22:55:40 +0000 (15:55 -0700)]
softpipe: Add support for reporting shader-db output.

In doing the softpipe NIR and NIR-to-TGSI transition, I want to make sure
I don't make shaders significantly worse, so I need shader-db output.

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6018>

18 months agosoftpipe: Convert to comma-separated SOFTPIPE_DEBUG for debug options.
Eric Anholt [Tue, 21 Jul 2020 22:11:56 +0000 (15:11 -0700)]
softpipe: Convert to comma-separated SOFTPIPE_DEBUG for debug options.

This makes us more like other drivers, and avoids having tons of different
names (particularly when you want to dump vs and fs in debugging).  In the
process, having a debug flag for vertex shaders just falls out.

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6018>

18 months agosoftpipe: Refactor pipe_shader_state setup.
Eric Anholt [Wed, 1 Jan 2020 00:31:54 +0000 (16:31 -0800)]
softpipe: Refactor pipe_shader_state setup.

We had repeated code that I want to repeatedly change for adding
nir-to-tgsi.

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6018>

18 months agollvmpipe: enable robust buffer access + GL 4.3, GLES 3.2 and robust buffer access...
Dave Airlie [Sun, 19 Jul 2020 23:22:34 +0000 (09:22 +1000)]
llvmpipe: enable robust buffer access + GL 4.3, GLES 3.2 and robust buffer access behaviour

Turning on robust buffer access enables GLES 3.2, also
finished GL 4.3 support.

The post depth coverage fail is expected, it's a test bug
This also introduce a fail in the invalid flag test that I can't reproduce out of CI.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5971>

18 months agollvmpipe: add device reset query context hook.
Dave Airlie [Sun, 19 Jul 2020 23:21:39 +0000 (09:21 +1000)]
llvmpipe: add device reset query context hook.

Add the device reset query hook needed for robustness

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5971>

18 months agoglx/drisw: add robustness support
Dave Airlie [Sun, 19 Jul 2020 23:55:30 +0000 (09:55 +1000)]
glx/drisw: add robustness support

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5971>

18 months agodrisw: add robustness extension support.
Dave Airlie [Sun, 19 Jul 2020 23:21:06 +0000 (09:21 +1000)]
drisw: add robustness extension support.

Port the code from dri2 so that drisw drivers can support the
robustness extension

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5971>

18 months agollvmpipe/draw: handle constant buffer limits and robustness (v1.1)
Dave Airlie [Sun, 29 Dec 2019 23:59:28 +0000 (09:59 +1000)]
llvmpipe/draw: handle constant buffer limits and robustness (v1.1)

TGSI expect vec4 of constants for it's current code paths, and when
doing indirect accesses it does the comparison on vec4 indexes,
however NIR does the indexing on packed float indexes.

This also align the compute path with the other shaders, and
should improve robustness (at least under Vulkan)

Fixes:
KHR-NoContext.gl43.robust_buffer_access_behavior.uniform_buffer

v1.1:
rename variable to something more meaningful (Roland)

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5971>

18 months agollvmpipe: enable EXT_texture_shadow_lod
Dave Airlie [Thu, 16 Jul 2020 00:50:14 +0000 (10:50 +1000)]
llvmpipe: enable EXT_texture_shadow_lod

The driver passes all the CTS tests for this.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5971>

18 months agonir/lower_io: Add support for global scratch addressing
Jason Ekstrand [Tue, 14 Jul 2020 18:32:19 +0000 (13:32 -0500)]
nir/lower_io: Add support for global scratch addressing

This provides an alternate lowering for scratch in which it uses global
reads/writes and bases scratch addresses on a base pointer.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5927>

18 months agonir/lower_io: Use b2b for shader and function temporaries
Jason Ekstrand [Wed, 15 Jul 2020 19:09:06 +0000 (14:09 -0500)]
nir/lower_io: Use b2b for shader and function temporaries

This way we can avoid some unnecessary conversions because there's no
need to sanitize to 0/1 for scratch.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5927>

18 months agonir/lower_io: Choose to set access based on intrinsic metadata
Jason Ekstrand [Wed, 15 Jul 2020 18:32:10 +0000 (13:32 -0500)]
nir/lower_io: Choose to set access based on intrinsic metadata

This should be far more reliable than trying to keep opcode lists
up-to-date.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5927>

18 months agonir: Allow for system values with variable numbers of destination components
Jason Ekstrand [Tue, 14 Jul 2020 18:27:53 +0000 (13:27 -0500)]
nir: Allow for system values with variable numbers of destination components

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5927>

18 months agodocs/releasing: improve wording
Eric Engestrom [Wed, 8 Jul 2020 23:26:31 +0000 (01:26 +0200)]
docs/releasing: improve wording

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5833>

18 months agobin/gen_release_notes: automatically commit release notes
Eric Engestrom [Wed, 8 Jul 2020 23:25:39 +0000 (01:25 +0200)]
bin/gen_release_notes: automatically commit release notes

Cc: mesa-stable
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5833>

18 months agopost_version.py: fix relnotes links
Eric Engestrom [Wed, 15 Jul 2020 21:53:59 +0000 (23:53 +0200)]
post_version.py: fix relnotes links

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5928>

18 months agopost_version.py: update the files in the current worktree, not the one with the scrip...
Eric Engestrom [Tue, 23 Jun 2020 00:24:00 +0000 (02:24 +0200)]
post_version.py: update the files in the current worktree, not the one with the script that we run

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5928>

18 months agopost_version.py: stop using non-existent functions and fix commit message
Eric Engestrom [Tue, 23 Jun 2020 00:22:58 +0000 (02:22 +0200)]
post_version.py: stop using non-existent functions and fix commit message

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5928>

18 months agopost_version.py: drop incorrect conf.py changes
Eric Engestrom [Tue, 23 Jun 2020 00:16:57 +0000 (02:16 +0200)]
post_version.py: drop incorrect conf.py changes

This needs to be done in the mesa3d.org repo; see
https://gitlab.freedesktop.org/mesa/mesa3d.org/-/merge_requests/19

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5928>