ac/nir: use nir_intrinsic_load_hs_out_patch_data_offset_amd in tess lower

radeonsi load this from SGPR arg, can't use static value because TCS output
and TES input may not match (TCS output is not a key for TES) and
determined in runtime.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16705>

ac/nir: add nir_intrinsic_load_hs_out_patch_data_offset_amd

Also add radv and radeonsi implementation. Will be used in tess lowering.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16705>

ac/nir: remove unused parameter in tes input lower

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16705>

radeonsi: implement nir_intrinsic_load_tcs_num_patches_amd

Used by ac_nir_lower_tess_io_to_mem.c.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16705>

radeonsi: replace llvm based fixed tcs with nir

Create nir passthrough shader with explicit input/output and vertex
output count so that it can be handled by compiler same as user tcs.

The drawback is we create more si_shader_selector with different
input/output and vertex output count which was handled by compiler
backend before.

As fixed function tcs can be handled like user tcs, we don't need
the dedicated fixed_func_tcs_shader state either.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16705>

radeonsi: add si_create_passthrough_tcs

For replacing si_create_fixed_func_tcs.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16705>

radeonsi: support multi stage shader state creation in nir shaderlib

For creating tcs passthrough shader.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16705>

radeonsi: use si_shader as parameter in si_get_nir_shader

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16705>

radeonsi: deserialize nir binary in si_check_blend_dst_sampler_noop

We can do this parse with original nir instead of shader key pass
applied nir in si_get_nir_shader.

This can free si_get_nir_shader to just use si_shader as parameter.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16705>

radv: no need to do gs_alloc_req for newer chips in ngg vs/tes

Copy from radeonsi.

Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17130>

ac/llvm: conditionally check wave id in gs sendmsg

nir lowering already call this with wave id check, no need to
check inside ac_build_sendmsg_gs_alloc_req again.

Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17130>

radv: Deal with derefs from opaque types in function parameters.

Needs more copy propagation before nir_opt_derefs picks it up.

Note that the full general problem of opaque types stored in
intermediate variables is still open, but that seems like a whole
can of worms, and no sense to have gfxbench stay broken during the
time it takes to solve that.

Cc: mesa-stable
Gitlab: https://gitlab.freedesktop.org/mesa/mesa/-/issues/5945
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17012>

Android.mk: Intermediate output paths may already be absolute

That is the case when OUT_DIR_COMMON_BASE is set.
Only prefix paths with AOSP_ABSOLUTE_PATH if they're relative.

Reviewed-by: Roman Stratiienko <r.stratiienko@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16674>

Android.mk: Generate the dummy source in local-generated-sources-dir

A source file cannot be otherwise referenced by absolute path.
That happens when OUT_DIR_COMMON_BASE is set.

Reviewed-by: Roman Stratiienko <r.stratiienko@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16674>

Android.mk: Make mesa3d-lib work with absolute path meson outputs

LOCAL_PREBUILT_MODULE_FILE is the only variable that allows
specifying the absolute path to a prebuilt.
That happens when OUT_DIR_COMMON_BASE is set.

Since it does not have multilib variants, define two separate
libraries for multilib

Reviewed-by: Roman Stratiienko <r.stratiienko@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16674>

Android.mk: Cleanup mesa3d-lib

Properly cleanup variables before declaring a library.
Explicitly require library dependencies.

Reviewed-by: Roman Stratiienko <r.stratiienko@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16674>

iris: Update comment about 2GB dynamic state range

We tracked this down with the HW teams back in 2020 and there's now a
documented workaround.  Comments from the HW team say this applies all
the way through XeHP but we're not sure beyond that.

This is a bug that we hit but the Windows drivers didn't because Jason
decided to allocate our memory structures from the top end of the VMA
range explicitly to catch bugs like this, while Windows allocates from
zero and up, so they would need to allocate more than 2GB of dynamic
state before running into it.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4880
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17216>

venus: support VK_KHR_copy_commands2

Signed-off-by: Ryan Neph <ryanneph@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17226>

venus: enable VK_EXT_image_view_min_lod

Signed-off-by: Ryan Neph <ryanneph@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17227>

venus: update venus-protocol with VK_EXT_image_view_min_lod

Copy in auto-generated protocol bindings.

Signed-off-by: Ryan Neph <ryanneph@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17227>

vulkan/render_pass: Support VkAttachmentSampleCountInfoAMD

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16953>

vulkan/render_pass: Allow for mixed sample counts

RADV supports VK_AMD_mixed_attachment_samples which does exactly what it
sounds like.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16953>

vulkan/render_pass: Pass sample locations to barriers

This is required for depth/stencil images created with
VK_IMAGE_CREATE_SAMPLE_LOCATIONS_COMPATIBLE_DEPTH_BIT_EXT.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16953>

vulkan/render_pass: Use a special layout for self-dependencies

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16953>

anv: Use CmdBeginRendering for resumes in BeginCommandBuffer when possible

This lets us avoid the code duplication between BeginRendering and
BeginCommandBuffer and also lets us stop crawling core render pass
structs directly and instead focus on dynamic rendering concepts.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16953>

vulkan/render_pass: Add a better helper for render pass inheritance

Instead of making drivers dive into the render pass and framebuffer
themselves, provide a helper that constructs a VkRenderingInfo for a
render pass resume that they can use instead.  This should reduce code
duplication between driver implementations of BeginRendering and
BeginCommandBuffer.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16953>

zink: add a turnip driver workaround for EXT_depth_clip_enable

this is broken

ref #6732

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17237>

zink: fix-ish depth clipping without VK_EXT_depth_clip_enable

if this extension is unsupported, use the previous behavior and hope for the best

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17237>

nir/opt_memcpy: Add another case for function_temp

Reviewed-by: Kristian H. Kristensen <hoegsberg@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com> (1.5 years later)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13166>

nir: Add an options parameter to deref_instr_has_complex_use

Reviewed-by: Kristian H. Kristensen <hoegsberg@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com> (1.5 years later)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13166>

nir/opt_memcpy: lower copies to/from tightly packed types

v2: Add comment by Jason (Lionel)

Reviewed-by: Kristian H. Kristensen <hoegsberg@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com> (1.5 years later)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13166>

zink: remove another zink/tu fail

fixed in f1c1b9687e8d58b62df5bc563c784cffda325611

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17241>

ci/dzn: Copy testlog.{css,xsl} to the result dir

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17235>

microsoft/compiler: Fix emit_ubo_var()

get_dword_size() is misleading, its name implies it's returning
a size in dwords, but it's actually returning a size in bytes.
This led to a wrong size passed to emit_cbv(). Instead of fixing
get_dword_size(), let's inline the code in emit_ubo_var().

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17230>

dzn: Transition resource to RENDER_TARGET/DEPTH_WRITE before clears

When clear_attachment() is called, we must ensure the resource is
in the DEPTH_WRITE or RENDER_TARGET state.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17229>

dzn: Clamp depthBiasConstantFactor when doing the float -> int conversion

If we don't do that, we might end up with an integer overflow/underflow.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17236>

dzn: Fix CmdPushConstants()

The original offset value is overwritten in our first for(i: num_states)
iteration, messing up the compute push constant update if stageFlags
applies to both compute and graphics.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17233>

tu: Don't count 3d blits in QUERY_TYPE_PRIMITIVES_GENERATED

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17164>

tu: Use hw binning or sysmem with QUERY_TYPE_PRIMITIVES_GENERATED

Without hw binning in gmem primitives generated query result could be
multiplied by tile count, which is not expected by OpenGL users for
GL_PRIMITIVES_GENERATED.

See https://gitlab.khronos.org/vulkan/vulkan/-/issues/3131

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17164>

aco: cleanup force-waitcnt output

If we don't reset ctx.vm_cnt/gpr_map/etc, this will spam a lot of
s_waitcnt instructions.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17207>

CODEOWNERS: evelikov renamed to xexaxo

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17072>

CODEOWNERS: Update c11 code owners

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17072>

intel/blorp/gen6: Set BLEND_STATEChange only if emitting the blend state

This change is pretty straightforward: if set this field when we don't emit
the blend state, then the garbage at offset=0 will be set as a blend state,
and this will cause artifacts until the proper blend state will be given.

Cc: mesa-stable
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6544
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6232

Signed-off-by: Sviatoslav Peleshko <sviatoslav.peleshko@globallogic.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17132>

pvr: Rename loop iterator variable.

Signed-off-by: Karmjit Mahil <Karmjit.Mahil@imgtec.com>
Reviewed-by: Rajnesh Kanwal <rajnesh.kanwal@imgtec.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17206>

pvr: Fix off by 1 error in buffer_id for ubo pds program.

Signed-off-by: Karmjit Mahil <Karmjit.Mahil@imgtec.com>
Reviewed-by: Rajnesh Kanwal <rajnesh.kanwal@imgtec.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17206>

pvr: Handle vdm degen_cull_enable.

Signed-off-by: Karmjit Mahil <Karmjit.Mahil@imgtec.com>
Reviewed-by: Rajnesh Kanwal <rajnesh.kanwal@imgtec.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17206>

pvr: Fix physical device limits.

This commit changes to the physical device limits which were
missed during the 1.17 transition.

Signed-off-by: Karmjit Mahil <Karmjit.Mahil@imgtec.com>
Reviewed-by: Rajnesh Kanwal <rajnesh.kanwal@imgtec.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17206>

anv: make apply_pipeline_layout/compute_push_layout visible to NIR debug

Useful for debug.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17209>

radv: Don't check if we need to copy immutable samplers for non push templates.

This should allow the compiler to optimize this out because it knows that
cmd_buffer is NULL.

Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17166>

ci: add wrapper script for Valve's traces runner

Contributed by Charlie Turner.

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Martin Roukala <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17184>

ci: move b2c artifacts to its own directory

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Martin Roukala <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17184>

ci: upgrade to DXVK v1.10.1

Additionally, improve the building scripts.

Contributed by Andres Gomez.

Reviewed-by: Martin Roukala <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17184>

ci: upgrade gfxreconstruct build to v0.9.10

Additionally improve the style of the build script.

 - Instead of using platform build tools, use the CMake wrappers
 - Instead of build all targets, then manually stripping, use the
   CMake helpers.

Contributed by Andres Gomez.

Reviewed-by: Martin Roukala <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17184>

ci: add Linux build of apitrace

This is needed for some of Valve's GL traces to run.

Additionally, make the building commands for apitrace more standard.

Contributed by Andres Gomez.

Reviewed-by: Martin Roukala <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17184>

ci: fix shellcheck violations in the test-vk building script

Reviewed-by: Martin Roukala <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17184>

ci: install gfxinfo-mupuf in the test-vk image

This python module is needed for Valve's tracing jobs.

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Martin Roukala <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17184>

ci: install a more recent version of Wine in the test-vk image

Debian's Wine 5.0 has shown some problems when running Vulkan tests in
the past. Let's install the stable version provided by WineHQ (7.0 at
this time).

v2:
  - Remove OBS repository for Wine since it is unused (previously, it
    was providing libfaudio0) (Daniel).

v3:
  - Add WineHQ's repository GPG key (Michel).

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Martin Roukala <martin.roukala@mupuf.org> [v2]
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17184>

radv/ci: update vkd3d-proton results for AMD's Kabini

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Martin Roukala <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17184>

zink: use tracked barrier info for generated barriers

this should be simpler to read through and maintain while providing
the same results as well as some possible perf and compile time improvements

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17192>

zink: track gfx/compute descriptor barrier info

update_barriers has steadily grown more and more complex when the original
idea was for it to be a small function to handle deferred jit barriers and
simplify sync in patterns like bind_ubo -> draw -> buffer_subdata -> draw

instead, track the pending barrier info at bind time so that the stages and
access are already updated by the time draw/compute are reached

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17192>

zink: fix image bind counting

these must only be incremented if the image descriptor has changed

cc: mesa-stable

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17192>

zink: track overall samplerview bind counts

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17192>

zink: track ssbo bind counts

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17192>

zink: use the bigger of the variable type and interface type for bo sizing

this avoids the scenario where the full bo size isn't accounted for because
no variable for the block has been created

cc: mesa-stable

affects:
KHR-GL33.shaders.uniform_block.random.all_per_block_buffers.3

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17217>

util: use force_gl_map_buffer_synchronized workaround with RAGE

CC: 22.1 22.0 <mesa-stable>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/1326
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17199>

util: add dri config option to disable GL_MAP_UNSYNCHRONIZED_BIT

GL_MAP_UNSYNCHRONIZED_BIT depends on the app having its threading
handled correctly. This allows us to force disable the bit when
they get it wrong.

CC: 22.1 22.0 <mesa-stable>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17199>

lavapipe: skip post-copy pNext checking during pipeline creation for composites

these values should have all been set during pipeline compositing above,
so reapplying the values is at best, redundant, and at worst, broken

cc: mesa-stable

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17219>

lavapipe: add a pipeline library assert

ensure that pipeline libraries are created with the library bit

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17219>

lavapipe: fix renderpass info handling during pipeline creation

only the viewMask parameter of VkPipelineRenderingCreateInfoKHR can
be accessed in the fragment stage, so for pipeline libraries it should
be assumed that zs attachments exist for the purpose of copying dynamic
state values, and then these dynamic states will naturally be pruned
during final pipeline construction if the attachments turn out to not
be present

cc: mesa-stable

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17219>

lavapipe: copy more pNexts for pipeline creation

also add some unreachable() handlers for unknown types

cc: mesa-stable

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17219>

lavapipe: zero out blend info if blend isn't enabled

this makes reading traces easier

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17219>

panfrost: Use u_default_clear_buffer

[Alyssa: This is required for OpenCL.]

Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16044>

iris: Use u_default_clear_buffer

iris uses u_default_buffer_subdata for buffer uploads via a CPU map so
clearing shouldn't be substantially worse.  We can do it with BLORP in
the future if we decide it's useful.

[Alyssa: A BLORP implementation is available at
https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15931 however nobody
has taken to reviewing that solution.]

Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16044>

gallium: Add a u_default_clear_buffer helper

[Alyssa: Add a default CPU implementation of pipe->clear_buffer(). This hook is
mandatory for OpenCL support. Even though this implementation isn't optimal by
any means, having a conformant default available in core will lower the barrier
of entry to OpenCL support.]

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16044>

anv: disable perf queries on non RCS engines

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17015>

u_blitter: Remove util_blitter_copy_buffer

It is now unused. We cannot yet remove the streamout functionality in u_blitter
as r600g still uses it for clear_buffer on GPUs older than Evergreen.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17142>

r600g: Remove streamout-based buffer copy path

r600g is the only user of util_blitter_copy_buffer in tree, which implements
buffer copies with streamout. This path for r600g was added in 8ac9801669c
("r600g: accelerate buffer copying"), a commit from 2012. At that point there
was no DMA path for buffer copies. Since then, a DMA path has been added,
conditional only on the kernel version -- not the hardware. It appears the
required kernel support has been mainline for at least 4 years now. Mesa 22.2
doesn't need to provide optimal performance on an old kernel -- for performance,
a DMA-capable kernel should be used, and for compatability, the CPU fallback
(used for unaligned buffers as it is) is still available. Remove the streamout
path "in the middle" that appears ~unused today.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17142>

microsoft/spirv_to_dxil: Fix discard semantics

Unlike in nir, discard is not a super return in DXIL. Therefore, we
will lower discard and terminate to demote + return.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17179>

nir/range_analysis: Teach range analysis about fdot opcodes

This really, really helps on platforms where fabs() isn't free.  A great
many shaders use a * frsq(fabs(fdot(a, a))) to normalize a vector.
Since the result of the fdot must be non-negative, the fabs can be
eliminated by an existing algebraic rule.

shader-db results:

r300 (run on R420 - X800XL)
total instructions in shared programs: 1369807 -> 1368550 (-0.09%)
instructions in affected programs: 59986 -> 58729 (-2.10%)
helped: 609
HURT: 0

total vinst in shared programs: 512899 -> 512861 (<.01%)
vinst in affected programs: 1522 -> 1484 (-2.50%)
helped: 36
HURT: 0

total sinst in shared programs: 260690 -> 260570 (-0.05%)
sinst in affected programs: 1419 -> 1299 (-8.46%)
helped: 120
HURT: 0

total consts in shared programs: 957295 -> 957230 (<.01%)
consts in affected programs: 849 -> 784 (-7.66%)
helped: 65
HURT: 0

LOST:   0
GAINED: 3

The 3 gained shaders are all vertex shaders from XCom: Enemy Unknown.
I'm guessing that game is never going to run on my X800XL. :)

i915
total instructions in shared programs: 791121 -> 780843 (-1.30%)
instructions in affected programs: 220170 -> 209892 (-4.67%)
helped: 2085
HURT: 0

total temps in shared programs: 47765 -> 47766 (<.01%)
temps in affected programs: 9 -> 10 (11.11%)
helped: 0
HURT: 1

total const in shared programs: 93048 -> 92983 (-0.07%)
const in affected programs: 784 -> 719 (-8.29%)
helped: 65
HURT: 0

LOST:   0
GAINED: 36

Haswell, Ivy Bridge, and Sandy Bridge had similar results. (Haswell shown)
total instructions in shared programs: 16702250 -> 16697908 (-0.03%)
instructions in affected programs: 119277 -> 114935 (-3.64%)
helped: 1065
HURT: 0
helped stats (abs) min: 1 max: 20 x̄: 4.08 x̃: 4
helped stats (rel) min: 0.48% max: 10.17% x̄: 3.66% x̃: 3.94%
95% mean confidence interval for instructions value: -4.26 -3.89
95% mean confidence interval for instructions %-change: -3.76% -3.56%
Instructions are helped.

total cycles in shared programs: 880772068 -> 880734134 (<.01%)
cycles in affected programs: 2134456 -> 2096522 (-1.78%)
helped: 941
HURT: 324
helped stats (abs) min: 2 max: 2180 x̄: 123.06 x̃: 44
helped stats (rel) min: 0.04% max: 49.96% x̄: 7.08% x̃: 3.81%
HURT stats (abs)   min: 2 max: 2098 x̄: 240.33 x̃: 35
HURT stats (rel)   min: 0.04% max: 77.07% x̄: 12.34% x̃: 3.00%
95% mean confidence interval for cycles value: -47.93 -12.04
95% mean confidence interval for cycles %-change: -2.87% -1.34%
Cycles are helped.

No shader-db changes on any other Intel platform.

Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17181>

egl/wayland: Don't try to access modifiers u_vector as dynarray

The modifiers are u_vectors, but the code was trying to access them
as dynarrays. This resulted in a wrong number of modifiers, which then
later on would also lead to invalid reads used as modifiers.

In the case of the iris driver, a wrongly read number of modifiers > 0
would also trigger an error message.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6643
Fixes: b5848b2dac1 ("egl/wayland: use surface dma-buf feedback to allocate surface buffers")
Reviewed-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
Reviewed-by: Simon Ser <contact@emersion.fr>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17180>

vulkan: Add a vk_pipeline_shader_stage_to_nir helper

This is similar to vk_shader_module_to_nir only it takes a
VkPipelineShaderStageCreateInfo and handles
VK_KHR_graphics_pipeline_library semantics for when a
VkShaderModuleCreateInfo is provided instead of an actual module.

Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17196>

vulkan/nir: Make spirv_data const in vk_spirv_to_nir

Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17196>

pvr: csbgen: Make all generated enums unambiguous

This change involves two enums:
 * rogue_texstate.xml: All COMPRESSED_* members of FORMAT are moved
   to FORMAT_COMPRESSED (without the prefix). A second field is added
   to IMAGE_WORD0 (texformat_compressed) which overlaps with the
   original (texformat), and
 * rogue_pbestate.xml: REG_WORD0_LINESTRIDE was not a real enum; it's
   removed entirely. It only has value when feature
   pbe_stride_align_1pixel is present, so a FIXME comment was added to
   this effect.

Signed-off-by: Matt Coster <matt.coster@imgtec.com>
Reviewed-by: Karmjit Mahil <Karmjit.Mahil@imgtec.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17204>

r300: only run merge_movs pass on R500

This pass currently generates some swizzles that the R300 and R400
hardware can't handle, make it R500 for now.

Fixes: 6c2959c0

Signed-off-by: Pavel Ondračka <pavel.ondracka@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17150>

tu: Check dereferenced value of rop_reads_dst.

Fix defect reported by Coverity Scan.

Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking rop_reads_dst suggests that it may be
null, but it has already been dereferenced on all paths leading to the
check.

Fixes: 94be0dd0b86 ("tu: Implement extendedDynamicState2LogicOp")
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17099>

ir3: Fix vectorizer condition for SSBOs

SSBO access works very differently from UBO access. Straddling
loads/stores isn't an issue, loads/stores instead must be aligned to the
element size and can have up to 4 components.

We support 16-bit access with SSBOs on a650+, and sometimes the
vectorizer tries to create a misaligned 32-bit access when combining
32-bit and 16-bit accesses. The UBO-focused logic didn't reject this,
which is now fixed. This fixes a number of VK-CTS regressions on a650+.

Fixes: bf49d4a084b ("freedreno/ir3: Enable load/store vectorization for SSBO access, too.")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17040>

aco: don't skip VS->TCS barrier if TCS output vertices doesn't match input

TCS invocations correspond to output patch vertices, not input. If they
differ, TCS invocations can be in a different subgroup than VS invocations
of the input patch.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6564
Fixes: 152092b8ead ("aco: skip s_barrier if TCS patches are within subgroup")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17174>

meson: Enable wgl tests on mingw

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16084>

ci: Building all mesa functional with mingw on debian

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16084>

d3d12: Turn d3d12_format.h to include d3d12_common.h

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16084>

ci: Trigger the new mingw/linux dockers to be build

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16084>

ci: Prepare the container for building all mesa components with mingw under linux

`x86_build-base-wine.sh` are usd to install wine and xvfb
`x86_build-mingw-patch.sh` are used to pull packages from msys2 that can be directly used.
`x86_build-mingw-source-deps.sh` are used to building llvm, libclc, clang, spirv-tools and directx-headers from source

xvfb are used to enable wgl tests on debian

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16084>

ci/x86_build: Getting pushd popd be paired, avoid using cd

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16084>

dzn: Fixes incompatible pointer type error

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16084>

microsoft/clc: Disable clc_compiler_test on non-windows platform

The test can compile, but can not pass, so compile it but not running it

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16084>

microsoft/clc: Fixes narrowing error in clc_compiler_test.cpp with mingw/gcc

errors:
../../src/microsoft/clc/clc_compiler_test.cpp:563:19: error: narrowing conversion of '268435457' from 'int' to 'uint16_t' {aka 'short unsigned int'} [-Wnarrowing]
  563 |       0x00000000, 0x10000001, 0x20000020, 0x30000300,
      |                   ^~~~~~~~~~
../../src/microsoft/clc/clc_compiler_test.cpp:563:31: error: narrowing conversion of '536870944' from 'int' to 'uint16_t' {aka 'short unsigned int'} [-Wnarrowing]
  563 |       0x00000000, 0x10000001, 0x20000020, 0x30000300,
      |                               ^~~~~~~~~~
../../src/microsoft/clc/clc_compiler_test.cpp:563:43: error: narrowing conversion of '805307136' from 'int' to 'uint16_t' {aka 'short unsigned int'} [-Wnarrowing]
  563 |       0x00000000, 0x10000001, 0x20000020, 0x30000300,

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16084>

d3d12: Fixes compiling error in d3d12/wgl/d3d12_wgl_framebuffer.cpp with gcc

error message:
```
../../src/gallium/winsys/d3d12/wgl/d3d12_wgl_framebuffer.cpp:231:42: error: no matching function for call to 'operator new(sizetype, d3d12_wgl_framebuffer*&)'
  231 |    new (fb) struct d3d12_wgl_framebuffer();
      |                                          ^
<built-in>: note: candidate: 'void* operator new(long long unsigned int)'

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16084>

d3d12: Convert #include <Windows.h> to #include <windows.h> for mingw on linux

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16084>

d3d12: Use static_cast instead of dynamic_cast in d3d12_video_enc_h264.cpp

Because we may compile mesa with both rtti=enabled and rtti=disabled because of LLVM
Fixes errors:
../../src/gallium/drivers/d3d12/d3d12_video_enc_h264.cpp:777:7: error: 'dynamic_cast' not permitted with '-fno-rtti'
  777 |       dynamic_cast<d3d12_video_bitstream_builder_h264 *>(pD3D12Enc->m_upBitstreamBuilder.get());
      |       ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16084>

turnip: Use the new border color helpers

Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15359>