llvmpipe: switch to NIR by default

Add LP_DEBUG=tgsi_ir (tgsi already taken) to fallback to TGSI paths.

Disable NIR_VALIDATE in CI (Michel/Eric acked)

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2303>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2303>

gallivm/nir: wrap idiv to avoid divide by 0 (v2)

This code is taken from the TGSI paths, and should fix the regression
seens with GLES2

v2: use the udiv path which has d3d10 defined return.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2303>

ac/surface: fix an assertion failure on gfx9 in CMASK computation

addrlib only allows the 2D resource type with CMASK.

Fixes: 69ea473eeb9 "amd/addrlib: update to the latest version"
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3187>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3187>

pan/midgard: Optimize comparisions with similar operations

Optimizes comparisions by removing the invert flag on operands
which we can prove to be equal without the invert.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3036>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3036>

lima: set shader caps to optimize control flow

With these new caps, nir is able to unroll loops and optimize
conditionals much more efficiently in both gpit and ppir.
panfrost and vc4 were used as reference for the values.

Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3176>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3176>

lima/ppir: remove assert on ppir_emit_tex unsupported feature

This assert causes testing tools such as shaderdb to abort on some test
cases. This is an unsupported feature and not a compiler bug. The
compilation error is already propagated correctly, so we can remove the
assert to allow testing tools to run to completion.

Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3176>

lima/ppir: fix lod bias src

ppir has some code that operates on all ppir_src variables, and for that
uses ppir_node_get_src.
lod bias support introduced a separate ppir_src that is inaccessible by
that function, causing it to be missed by the compiler in some routines.
Ultimately this caused, in some cases, a bug in const lowering:

  .../pp/lower.c:42: ppir_lower_const: Assertion `src != NULL' failed.

This fix moves the ppir_srcs in ppir_load_texture_node together so they
don't get missed.

Fixes: 721d82cf061 lima/ppir: add lod-bias support
Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3185>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3185>

lima: Fix dump file creation

Otherwise lima_dump_file_next() always opens a new file and creates the
dumps regardless of what the environment variables say.

Fixes d71cd245d74 ('lima: Rotate dump files after each finished pp frame')

Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Signed-off-by: Andreas Baierl <ichgeh@imkreisrum.de>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3179>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3179>

radeon/vcn2: enable rate control for hevc encoding

Based on b0626c1f306 ("radeon/vcn: enable rate control for hevc encoding").

Reviewed-by: Boyuan Zhang <boyuan.zhang@amd.com>
Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/2225
Fixes: 587b9c5dae6 ("radeon/vcn: implement vcn 2.0 encode")
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3134>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3134>

radv: rely on pipeline layout when creating push descriptors with template

descriptorSetLayout should be ignored for push descriptors. While
we are it, also ignore pipelineBindPoint.

Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/2210
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3180>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3180>

etnaviv: Replace bitwise OR with logical OR

The test here is testing whether either variable is non-zero.
While currently the test works fine, it's fragile. Replace it
with logical OR to avoid the fragility.

Signed-off-by: Marek Vasut <marex@denx.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>

etnaviv: update resource status after flushing

Currently piglit spec@arb_occlusion_query@occlusion_query_conform
spins for ever as the resource status is never reset. See
etna_hw_get_query_result(..) for more details.

Fixes: 1456aa61cc5 ("etnaviv: Rework resource status tracking")
CC: <mesa-stable@lists.freedesktop.org>
Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Tested-by: Marek Vasut <marex@denx.de>

intel: limit shader geometry on BDW GT1

Similar to the SKL GT1 fix introduced here:

https://gitlab.freedesktop.org/asimiklit/mesa/commit/b1ba7ffdbd54fdb5da18d086c7b7a830e06a1cff

we need to limit the .urb.max_entries[MESA_SHADER_GEOMETRY] on BDW GT1
to address failures in these two tests:

dEQP-GLES31.functional.geometry_shading.layered.render_with_default_layer_3d
dEQP-GLES31.functional.geometry_shading.layered.render_with_default_layer_2d_array

The value 690 was found via bisection.  691 is the actual max on the
hardware I'm using, but 690 seemed like a nice round number.

Signed-off-by: Ross Zwisler <zwisler@google.com>
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3173>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3173>

pan/midgard: Lower txd with lower_tex

This is a hack since we do have native gradient stuff, but for the
moment I'm more interested in conformance and the lowered code is good
enough. Fixes
dEQP-GLES3.functional.shaders.texture_functions.texturegrad.sampler2d_fixed_fragment

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3169>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3169>

pan/midgard: Fix crash with txs

This regressed since we implemented RECT textures natively, oops.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3169>

pan/midgard: Implement textureOffset for 2D textures

Fixes dEQP-GLES3.functional.shaders.texture_functions.textureoffset.sampler2d_fixed_fragment.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3169>

radv: ignore pColorBlendState if rasterization is disabled

Or if the subpass has no color attachments.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3167>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3167>

radv: tidy up radv_pipeline_init_blend_state()

This is needed for the next commit because pColorBlendState can
actually be NULL but some fields might have to be initialized
(eg. alpha to coverage with no color attachments).

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/3167>

radv: ignore pDepthStencilState if rasterization is disabled

Or if the subpass has no depth stencil attachment.

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/3167>

radv: ignore pTessellationState if the pipeline doesn't use tess

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/3167>

radv: ignore pMultisampleState if rasterization is disabled

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/3167>

radv: init a default multisample state for the resolve FS path

pMultisampleState must be a valid pointer.

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/3167>

spirv: Implement SPV_KHR_non_semantic_info

Do nothing for OpExtInst from extended instruction sets that name
start with "NonSemantic.".

Since they can be used within the "preamble" to annotate global
decorations, also don't stop iterating when one of them is found.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3154>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3154>

turnip: disable B8G8R8 vertex formats

Looks like swap doesn't work as expected on these, disable them.

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3170>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3170>

util/format: add missing vulkan formats

Add some missing vulkan formats to util/format, this solves all the missing
pipe format cases for the formats that turnip supports.

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3170>

turnip: minor warning fixes

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3177>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3177>

lima: Rotate dump files after each finished pp frame

This rotates the dump files like the mali-syscall-tracker does.

After each finished pp frame a new file is generated. They are
numbered like lima.dump.0000, lima.dump.0001 ...
The filename and path can be given with the new environment
variable LIMA_DUMP_FILE.

Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Signed-off-by: Andreas Baierl <ichgeh@imkreisrum.de>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3175>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3175>

lima: drop suballocator

Since we're using a separate per-draw BO for GP outputs we don't
need suballocator anymore.

Reviewed-by: Erico Nunes <nunes.erico@gmail.com>
Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3158>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3158>

lima: use single BO for GP outputs

Varyings, gl_Position and gl_PointSize are all GP outputs, so we
can use a single BO for them all. Also that allows us to get rid
of suballocator.

Reviewed-by: Erico Nunes <nunes.erico@gmail.com>
Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3158>

nir: fix assign_io_var_locations for vertex inputs

Also fixes fragment inputs using the wrong "base" value (which was working
only because FRAG_RESULT_DATA0 is less than VARYING_SLOT_VAR0)

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3108>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3108>

turnip: implement secondary command buffers

Uses a new "tu_cs_add_entries" function because tu_cs_emit_call doesn't
work inside draw_cs (which is already called by tu_cs_emit_call).

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3075>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3075>

turnip: compute gmem offsets at renderpass creation time

This makes it easier to implement secondary command buffers, since we no
longer need to know the render area to set the gmem offsets for input
attachments and CmdClearAttachments.

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3075>

turnip: emit_compute_driver_params fixes

Offset was wrong, it is in vec4 not dwords.

There's a hole between DP_NUM_WORK_GROUPS_Z and DP_LOCAL_GROUP_SIZE_X so
use the IR3 enums.

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3162>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3162>

turnip: emit base instance vs driver param

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3162>

freedreno/ir3: support load_base_instance

Not supported by hardware, uses same mechanism as base vertex.

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3162>

freedreno/registers: document vertex/instance id offset bits

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3162>

st/mesa: release tgsi tokens for shader states

Since we are using st_common_variant while creating variant for vertext
program, we can release tokens created in st_create_vp_variant which
are already stored in respective states.
This fix memory leak found with piglit tests

Fixes bc99b22a305b ('st/mesa: use a separate VS variant for the draw module')

Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>

Revert "nir/lower_double_ops: relax lower mod()"

This reverts commit 8172b1fa03fe74165728bfb182c98a3e62193d2b.

This commit was done taking in account Vulkan spec, but did not realize
it was affecting OpenGL too.

Closes: #2252

freedreno/a6xx: Set up multisample sysmem MRTs correctly

We had an extra factor of num_samples in the stride.

Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2848>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2848>

freedreno/a6xx: Rewrite compressed blits in a helper function

Similar to how we handle zs blits.

Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2848>

freedreno/a6xx: Move handle_rgba_blit() up

If we move this function up, we don't have to forward declare it.

Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2848>

freedreno/a6xx: Handle srgb blits on the blitter

Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2848>

freedreno/a6xx: Use A6XX_SP_2D_SRC_FORMAT_MASK macro

Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2848>

freedreno/a6xx: RB6_R8G8B8 is actually 32 bit RGBX

Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2848>

freedreno/a6xx: Use blitter for resolve blits

We have a SAMPLES_AVERAGE bit that does what we need for resolving
multisample buffers - let's use it.

Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2848>

freedreno/a6xx: Add fd_resource_swap() helper

Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2848>

freedreno/a6xx: Pick blitter swap based on resource tiling

The linear levels in a tiled resource are stored in the canonical
swap, WZYX.  We need to pick the swap based on whether or not the
resource is tiled, not whether the the level in question is tiled.

Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2848>

freedreno/a6xx: Program sampler swap based on resource tiling

It doesn't matter whether or not the level in question is linear.

Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2848>

freedreno: Add debug flag for forcing linear layouts

Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2848>

freedreno/a6xx: Make DEBUG_BLIT_FALLBACK only dump fallbacks

Use new macro, DEBUG_BLIT, for dumping all blits.

Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2848>

freedreno/ir3: fix vertex shader sysvals with pre_assign_inputs

The first pre_assign_inputs loop doesn't pre-assign sysvals, so skip the
second part for sysvals.

The sysvals don't need to be pre-assigned since the state for those isn't
shared between binning / nonbinning shaders.

Fixes assert failures in cases where the sysvals didn't end up in the same
registers for binning / nonbinning.

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3168>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3168>

st/va: Convert interlaced NV12 to progressive

In vlVaDeriveImage, convert interlaced NV12 buffers to progressive.

Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/1193
Signed-off-by: Thong Thai <thong.thai@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3157>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3157>

pan/midgard: Add uniform/work heuristic

Uniform/work registers are partitioned on a shader-by-shader basis as
determined by the compiler. We add a simple heuristic here running
before scheduling that prioritizes mitigating spilling at all costs.

A more sophisticated heuristic should run *after* scheduling, doing a
dry run of the register allocator itself to determine spilling. Fitting
this into our current scheduling model is difficult, so while this
heuristic does hurt some shaders, overall the results are acceptable:

total instructions in shared programs: 50065 -> 38747 (-22.61%)
instructions in affected programs: 37187 -> 25869 (-30.44%)
helped: 59
HURT: 77
helped stats (abs) min: 1 max: 757 x̄: 198.46 x̃: 151
helped stats (rel) min: 0.48% max: 62.89% x̄: 32.95% x̃: 36.27%
HURT stats (abs)   min: 1 max: 9 x̄: 5.08 x̃: 6
HURT stats (rel)   min: 0.92% max: 14.29% x̄: 6.71% x̃: 4.60%
95% mean confidence interval for instructions value: -111.15 -55.29
95% mean confidence interval for instructions %-change: -14.33% -6.67%
Instructions are helped.

total bundles in shared programs: 30606 -> 19157 (-37.41%)
bundles in affected programs: 23907 -> 12458 (-47.89%)
helped: 58
HURT: 74
helped stats (abs) min: 6 max: 757 x̄: 203.09 x̃: 152
helped stats (rel) min: 5.19% max: 77.00% x̄: 49.38% x̃: 53.79%
HURT stats (abs)   min: 1 max: 9 x̄: 4.46 x̃: 5
HURT stats (rel)   min: 1.85% max: 26.32% x̄: 11.70% x̃: 9.57%
95% mean confidence interval for bundles value: -115.46 -58.01
95% mean confidence interval for bundles %-change: -20.87% -9.41%
Bundles are helped.

total quadwords in shared programs: 31305 -> 32027 (2.31%)
quadwords in affected programs: 20471 -> 21193 (3.53%)
helped: 0
HURT: 133
HURT stats (abs)   min: 1 max: 9 x̄: 5.43 x̃: 5
HURT stats (rel)   min: 0.76% max: 15.15% x̄: 5.47% x̃: 4.65%
95% mean confidence interval for quadwords value: 5.00 5.86
95% mean confidence interval for quadwords %-change: 4.85% 6.08%
Quadwords are HURT.

total registers in shared programs: 2256 -> 2545 (12.81%)
registers in affected programs: 708 -> 997 (40.82%)
helped: 0
HURT: 95
HURT stats (abs)   min: 1 max: 8 x̄: 3.04 x̃: 3
HURT stats (rel)   min: 12.50% max: 100.00% x̄: 39.41% x̃: 37.50%
95% mean confidence interval for registers value: 2.64 3.45
95% mean confidence interval for registers %-change: 34.62% 44.19%
Registers are HURT.

total threads in shared programs: 1776 -> 1709 (-3.77%)
threads in affected programs: 134 -> 67 (-50.00%)
helped: 0
HURT: 67
HURT stats (abs)   min: 1 max: 1 x̄: 1.00 x̃: 1
HURT stats (rel)   min: 50.00% max: 50.00% x̄: 50.00% x̃: 50.00%
95% mean confidence interval for threads value: -1.00 -1.00
95% mean confidence interval for threads %-change: -50.00% -50.00%
Threads are HURT.

total spills in shared programs: 3868 -> 2 (-99.95%)
spills in affected programs: 3868 -> 2 (-99.95%)
helped: 60
HURT: 0

total fills in shared programs: 6456 -> 4 (-99.94%)
fills in affected programs: 6456 -> 4 (-99.94%)
helped: 60
HURT: 0

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3150>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3150>

ac: declare an enum for the OOB select field on GFX10

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3147>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3147>

radv/gfx10: fix the out-of-bounds check for vertex descriptors

When stride is 0, it should check against the offset not the index.

This fixes black character models with Beat Saber and missing snow
with Dragon Quest.

Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/2233
Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/1975
Cc: <mesa-stable@lists.freedesktop.org>
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/3147>

nir/lower_double_ops: relax lower mod()

Currently when lowering mod() we add an extra instruction so if
mod(a,b) == b then 0 is returned instead of b, as mathematically
mod(a,b) is in the interval [0, b).

But Vulkan spec has relaxed this restriction, and allows the result to
be in the interval [0, b].

This commit takes this in account to remove the extra instruction
required to return 0 instead.

Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2922>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/2922>

zink: implement nir_texop_txd

This lets us enable PIPE_CAP_FRAGMENT_SHADER_TEXTURE_LOD, which in turns
gives us ARB_shader_texture_lod.

Still fails one piglit test on ANV, namely
spec@arb_shader_texture_lod@execution@arb_shader_texture_lod-texgradcube,
but with 33 new passing tests, I think this is worth it.

Reviewed-by: Dave Airlie <airlied@redhat.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3140>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3140>

zink: enable PIPE_CAP_MIXED_COLORBUFFER_FORMATS

This just works in Vulkan, there's no work neeed to enable it.

Reviewed-by: Dave Airlie <airlied@redhat.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3148>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3148>

turnip: don't set SP_FS_CTRL_REG0_VARYING if only fragcoord is used

Fixes artifacts in the subpasses demo, which has a shader using fragcoord
without any varyings. It looks like setting this bit when there are no
varyings can cause weirdness in some cases (without this change, if the
previous shader had <= 8 varyings it would work, but with 9 varyings it
would have artifacts).

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3143>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3143>

turnip: add cache invalidate to fix input attachment cases

Fixes artifacts in the subpasses demo.

Workaround texture cache with input attachments from GMEM by adding a cache
invalidate between subpasses.

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3143>

loader: fix close on uninitialized file descriptor value

Using a drm syscall layer faking a kernel driver :

  ==581460== Conditional jump or move depends on uninitialised value(s)
  ==581460==    by 0x48A4C2B: close (drm-hooks.cpp:185)
  ==581460==    by 0x5A815F1: dri3_alloc_render_buffer (loader_dri3_helper.c:1469)
  ==581460==    by 0x5A82050: dri3_get_buffer (loader_dri3_helper.c:1827)
  ==581460==    by 0x5A82662: loader_dri3_get_buffers (loader_dri3_helper.c:2028)
  ==581460==    by 0x6C78109: intel_update_image_buffers (brw_context.c:1870)
  ==581460==    by 0x6C77805: intel_update_renderbuffers (brw_context.c:1499)
  ==581460==    by 0x6C7789D: intel_prepare_render (brw_context.c:1520)
  ==581460==    by 0x6C773D4: intelMakeCurrent (brw_context.c:1341)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 069fdd5f9fac ("egl/x11: Support DRI3 v1.1")
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3152>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3152>

freedreno: Fix CP_MEM_TO_REG flag definitions

These actually mean something completely different, at least on A5xx
and A6xx. The only other usage of the old flags on something older than
A6xx was a typo, so I don't know if it was always this way, but at the
same time it means that we don't have to worry too much about that.

Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3116>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3116>

freedreno: Use new macros for CP_WAIT_REG_MEM and CP_WAIT_MEM_GTE

Similar to the existing usage for CP_COND_WRITE5, this makes it clear
what each of the magic parameters are for.

Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3116>

a6xx: Add more CP packets

And add fields uncovered by looking at the firmware. I think this covers
all the memory, register, and scratch manipulation opcodes that exist on
A6xx, plus one additional nice find for Vulkan and describing a
previously unknown opcode and documenting CP_WAIT_REG_MEM.

Note that the bits for the CP_REG_TO_MEM count, as well as the formula
for computing the actual count for both CP_REG_TO_MEM and CP_MEM_TO_REG,
are changed because the A630 SQE firmware actually does something
different. I haven't investigated older microcodes to see whether this
extends back to A5xx and A4xx, but the only non-A6xx uses of this
field result in the same bit-pattern when using the A6xx bit range and
formula, so it should be safe to change the definition universally.

Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3116>

radv: Limit workgroup size to 1024.

Fixes a hang with geekbench.

The existence of RX 580 and NAVI10 results shows that the generations
before and after this do not have the issue. (They show up on the
website). So this is likely a GFX9 only issue.

This is not something weird like LDS size since none of the shaders
seem to use LDS.

CC: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3145>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3145>

docs: Add release notes, news, and update calendar for 19.2.8

docs/relnotes/19.2.8: Add SHA256 sum

docs: add relnotes for 19.2.8

docs: Add release notes, update calendar, and add news for 19.3.1

dcos: add releanse notes for 19.3.1

i965/iris/perf: factor out frequency register capture

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Mark Janes <mark.a.janes@intel.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3113>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3113>

freedreno/ir3: update prefetch input_offset when packing inlocs

If the input location changes then prefetch input_offset needs to change.

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3141>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3141>

ci: Fix caselist results archiving after parallel-deqp-runner rename.

Noticed while reviewing some lava parallel-deqp-runner changes.

Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3138>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3138>

freedreno/a6xx: Document the CP_SET_DRAW_STATE enable bits

There are bits for binning, gmem and sysmem.

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3131>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3131>

anv/gen12: Temporarily disable VK_KHR_buffer_device_address (and EXT)

For the sake of our testing infrastructure, disable this extension
for TGL until we can sort out a hang in Vulkan CTS.

Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>

intel/vec4: Fix lowering of multiplication by 16-bit constant

Existing code was ignoring whether the type of the immediate source
was signed or not.  If the source was signed, it would ignore small
negative values but it also would wrongly accept values between
INT16_MAX and UINT16_MAX, causing the atual value to later be
reinterpreted as a negative number (under 16-bits).

Fixes tests/shaders/glsl-mul-const.shader_test in Piglit for older
platforms that don't support MUL with 32x32 types and use vec4.

Cc: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

intel/fs: Fix lowering of dword multiplication by 16-bit constant

Existing code was ignoring whether the type of the immediate source
was signed or not.  If the source was signed, it would ignore small
negative values but it also would wrongly accept values between
INT16_MAX and UINT16_MAX, causing the atual value to later be
reinterpreted as a negative number (under 16-bits).

Fixes tests/shaders/glsl-mul-const.shader_test in Piglit for platforms
that don't support MUL with 32x32 types, including ICL and TGL.

Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/2186
Cc: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

pan/midgard: Set Z to shadow comparator for 2D

We still need to generalize for other types of (non-2D / array) shadow
samplers, but this is enough for sampler2DShadow to work with initial
dEQP tests passing.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3125>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3125>

pan/midgard: Set .shadow for shadow samplers

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3125>

pan/midgard: Hoist temporary coordinate for cubemaps

We'll reuse some of this code for shadow samplers, which are represented
by a distinct source in NIR.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3125>

pan/midgard: Use a reg temporary for mutiple writes

Bug in texelfetch implementation from inspection.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3125>

panfrost: Handle empty shaders

I didn't realize this was in spec, but it fixes a crash in shaderdb.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3125>

panfrost: Let precompile imply shaderdb

This cuts down the number of random environmental variables we need
flying around; now PAN_MESA_DEBUG=precompile is sufficient and
MIDGARD_MESA_DEBUG=shaderdb will be implied.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3125>

panfrost: Add PAN_MESA_DEBUG=precompile for shader-db

We would like to use run.c for shader-db runs (rather than capturing in
real-time, which is limiting).

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3125>

mesa: avoid triggering assert in implementation

When tearing down a GL context with an active performance query, the
implementation can be confused by a query marked active when it's
being deleted.

This shouldn't happen in the implementation because the context will
already be idle.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: <mesa-stable@lists.freedesktop.org>
Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/2235
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3115>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3115>

radv/gfx10: fix ngg_get_ordered_id

Ported from RadeonSI.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3133>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3133>

ci: Remove T820 from CI temporarily

Our lab will have continuous programmed power cuts until the 6th January 2020,
so it's safer to disable the T820 CI running on the BayLibre kernelCI lab
to avoid breaking CI.

Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3135>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3135>

i965: expose MESA_FORMAT_B8G8R8X8_SRGB visual

Patch adds BGRX sRGB visuals, required format translation information
to the __DRI_IMAGE_FOURCC_SXRGB8888 format and makes all BGRX visuals
sRGB capable just like is done with BGRA.

squashed patches from Yevhenii Kolesnikov:
  dri: Add __DRI_IMAGE_FOURCC_SXRGB8888 conversion
  i965: force visuals without alpha bits to use sRGB

Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/1501
Cc: <mesa-stable@lists.freedesktop.org>
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Signed-off-by: Yevhenii Kolesnikov <yevhenii.kolesnikov@globallogic.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3077>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3077>

dri: add __DRI_IMAGE_FORMAT_SXRGB8

Add format definition and required plumbing to create images.

Cc: <mesa-stable@lists.freedesktop.org>
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3077>

virgl: Increase the shader transfer buffer by doubling the size

With only linearly increasing the size of the shader transfer buffer
the transfer of very large shaders may fail, so with each attempt double
the size of the buffer.

CTS:
  dEQP-GLES31.functional.ssbo.layout.random.all_shared_buffer.48
  for VTK-GL-CTS b5dcfb9c5 and newer

virglrenderer bug:
  https://gitlab.freedesktop.org/virgl/virglrenderer/issues/150

Fixes: a8987b88ff1db4ac00720a9b56c4bc3aeb666537
    virgl: add driver for virtio-gpu 3D (v2)

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Gurchetan Singh <gurchetansingh@chromium.org>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3121>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3121>

turnip: Fix support for immutable samplers.

We were setting up the hardware sampler state when updating a combined
image sampler, but never looking at the immutable sampler for in the
separate case.

Fixes failures in
dEQP-VK.binding_model.shader_access.primary_cmd_buf.sampler_immutable.fragment.*

Reviewed-by: Jonathan Marek <jonathan@marek.ca>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3127>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3127>

turnip: don't set LRZ enable at end of renderpass

Fixes hanging with cases that use more than one renderpass.

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3122>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3122>

freedreno/ir3: lower pack/unpack ops

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3106>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3106>

nir: add option to lower half packing opcodes

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
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/3106>

turnip: Add support for descriptor arrays.

I had a bigger rework I was working on, but this is simple and gets tests
passing.

Fixes 36 failures in
dEQP-VK.binding_model.shader_access.primary_cmd_buf.sampler_mutable.fragment.*
(now all passing)

Reviewed-by: Jonathan Marek <jonathan@marek.ca>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3124>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3124>

turnip: Drop unused variable.

We really need -Werror in CI.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3124>

panfrost: Don't double-create scratchpad

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Fixes: 4f7fddbd716 ("panfrost: Pass size to panfrost_batch_get_scratchpad")
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3119>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3119>

panfrost: Simplify sampler upload condition

Makes it more obvious what's going on.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3119>

gallium/auxiliary: Handle count == 0 in u_vbuf_get_minmax_index_mapped

This makes u_vbuf_get_minmax_index_mapped return min = 0 / max = 0
when info->count == 0.

That should never happen anyway, but this commit makes it at least
return a sane value that callers expect, and also allows us - and
GCC - to assume count != 0 for optimization purposes.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3050>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3050>

gallium/auxiliary: Reduce conversions in u_vbuf_get_minmax_index_mapped

With this patch, GCC generates vectorized code that does the comparisons
without converting the indices to 32-bit first.

This optimization makes the aforementioned function almost twice as fast
for ARM NEON, and should speed up vectorised code on other platforms.

Without vectorisation, the function is still a percent or two faster,
but slightly larger.

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3050>