radv: Use nir_gen_rect_vertices

Signed-off-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17535>

nir: Add a common gen_rect_vertices implementation

Signed-off-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17535>

radv: disable viewport depth clamping only when necessary

When the application uses depth values outside of the [0.0,1.0] range
with VK_EXT_depth_range_unrestricted, or when explicitly disabled.
Otherwise, the driver can clamp to [0.0,1.0] internally for optimal
performance.

From the Vulkan spec "in 28.10.1. Depth Clamping and Range Adjustment":

    "If depth clamping is not enabled and zf is not in the range [0, 1]
    and either VK_EXT_depth_range_unrestricted is not enabled, or the
    depth attachment has a fixed-point format, then zf is undefined
    following this step."

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16349>

v3dv: implement VK_KHR_buffer_device_address

This feature allows shaders to use pointers to buffers which may
not be bound via descriptor sets. Access to these buffers is done
via global intrinsics.

Because the buffers are not accessed through descriptor sets, any
live buffer flagged with VK_BUFFER_USAGE_SHADER_DEVICE_ADDRESS_BIT_KHR
can be accessed by any shader using global intrinsics, so the driver
needs to make sure all these buffers are mapped by the kernel when
it submits the job for execution.

We handle this by tracking if any draw call or compute dispatch in
a job uses a pipeline that has any such shaders. If so, the job is
flagged as using buffer device address and the kernel submission
for that job will add all live BOs bound to buffers flagged with the
buffer device address usage flag.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17275>

broadcom/compiler: track if a shader uses global intrinsics

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17275>

broadcom/compiler: implement 2x32 global intrinsics

Notice we ignore the high 32-bit component of the address because
we know it must be 0.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17275>

nir: add nir_address_format_2x32bit_global

This adds support for global 64-bit GPU addresses as a pair of
32-bit values. This is useful for platforms with 32-bit GPUs
that want to support VK_KHR_buffer_device_address, which makes
GPU addresses explicitly 64-bit.

With the new format we also add new global intrinsics with 2x32
suffix that consume the new address format.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17275>

v3dv: remove duplicate condition

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17536>

v3dv: explain why we clear certain state after a draw call

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17536>

v3dv: add a dirty state for pending push constants UBO updates

If we have 2 pipelines that consume the same push constant data
but where one of them only uses direct access and the other has
indirect access, a draw with the first pipeline would clear the
dirty flag without updating the UBO and by the time we bind and
draw with the second pipeline we won't upload the constants either
because the first draw cleared the dirty flag.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17536>

v3dv: allocate more push constant buffers if needed

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17536>

v3dv: stop tracking push constant buffer references

Since we allocate this ourselves we can immediately add it to the
job at the time we allocate it.

This also fixes a bug we introduced when we implemented inline
uniforms because since that commit, if we had an inline uniform
buffer at index 1 which happend to have indirect access we would
track it in slot 0 instead of slot 1, potentially overwriting
the push constant buffer reference.

Fixes: ea3223e7a46 ('v3dv: implement VK_EXT_inline_uniform_block')
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17536>

v3dv: don't allocate MAX_PUSH_CONSTANTS_SIZE bytes for the push constants UBO

We have code in there to allocate various segments of MAX_PUSH_CONSTANTS_SIZE
to handle the case of various draw calls in the same command buffer requiring
different push constants, so we are implicitly expecting it to be larger than
this. In fact, this only works now because when we allocate a BO we are always
at least allocating a full page, so the least we ever allocate is 4096 bytes,
so be explicit about it to avoid confusion.

Also, since we were always mapping MAX_PUSH_CONSTANTS_SIZE and the mapping
always starts at the beginning of the BO, it looks like after the first copy
when the resource offset is not zero, we would be writing outside the mapped
range. Always map the full size of the BO instead to ensure this doesn't
happen.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17536>

v3dv: limit upload of indirect push constant data

We have been always uploading MAX_PUSH_CONSTANTS_SIZE but now that
we track the actual size of the push constant buffer we can use
this instead.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17536>

v3dv: move push constant data to the command buffer state

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17536>

v3dv: only save/restore push constant data for meta operations if needed

If the command buffer didn't have any push constants or the meta
operation didn't write any new constants we don't need to restore
the state.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17536>

Revert "vulkan: Detect pNext chain loops in vk_foreach_struct()"

This reverts commit 4c56b535f5b415c24805dccc8d147a60af3f03e3.

anv: move restart index to gfx state

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/17601>

anv: fix primitive topology dynamic state emission on gfx7

This dynamic state uses the pipeline information so it needs to be
reemitted whenever the pipeline changes.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: mesa-stable
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17601>

anv: name non dynamic state fields correctly

Those fields have confusing names. They hold non dynamic state,
specified at pipeline creation that we copy into the dynamic state of
the command buffer whenever we bind a pipeline. This non dynamic state
might get picked up in the dynamically emitted instructions of the 3D
pipeline because our HW packets are not exactly splitted like the
Vulkan API.

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/17601>

anv: reorder & document fields of anv_graphics_pipeline

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/17601>

anv: move CreateRayTracingPipelines to common code

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/17601>

anv: move CreateComputePipelines to common code

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/17601>

anv: move CreateGraphicsPipelines to common code

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/17601>

anv: rename internal function for consistency

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/17601>

anv: Don't use the wrong ARRAY_SIZE

Even though this doesn't change anything, it's not good to use an
ARRAY_SIZE for one array to iterate over another.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17601>

anv: split graphics nir loading

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/17601>

anv: break up anv_pipeline_compile_graphics()

This function is pretty overwhelming.

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/17601>

anv: simplify dynamic buffer count in pipeline layout

anv_descriptor_set_layout already has the information we're gather
here.

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/17601>

anv: remove local computation of dynamic states

This bit mask is already computed in
anv_graphics_pipeline::dynamic_states in anv_graphics_pipeline_init().

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/17601>

vulkan: Detect pNext chain loops in vk_foreach_struct()

This implements the "tortoise and the hare" algorithm for detecting
cycles in graphs.  We use the caller's iterator as the hare and our own
internal copy as the tortoise.  Conveniently, VkBaseOutStructure (and
VkBaseInStructure which is identical except the pointer type on pNext)
have a pointer we can use for the tortoise and an sType which we can use
for a counter to ensure we only increment the tortose every other loop
iteration.

There are more efficient algorithms than tortoise and hare but they
require allocating memory for something like a hash set of seen nodes.
Since this for debug purposes only, it's ok for it to be a bit
inefficient in the case where it hits the assert.  In the usual case of
no loops, it's the same runtime efficiency as the unchecked version
except that it does a tiny bit of math and 50% more pointer chases.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17596>

intel/fs: Simplify brw_barycentric_mode() args.

Reduce a bit of mode lookup noise I was tracing through trying to resolve
the previous bug.

Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17381>

intel/fs: Set NonPerspectiveBarycentricEnable when the interpolator needs it.

[anholt: changed to make all drivers do the right thing by moving the
payload barycentric check into the compiler]

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17381>

ci: move apitrace instalation outside of x86_test-vk

It'll be re-used by piglit traces.

Reviewed-by: Guilherme Gallo <guilherme.gallo@collabora.com>
Reviewed-by: Andres Gomez <agomez@igalia.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17573>

freedreno/ir3: Enable core NIR's 16-bit ALU optimizations.

In addition to hopefully generating shorter code, this optimizes out a
comparison of a mediump-cast value in
dEQP-GLES2.functional.shaders.algorithm.rgb_to_hsl_fragment passed
through ANGLE, and allows the test to pass.  We believe it to be a
test bug, but emitting better code like apparently everyone else does
is also a fine result.

No change on GLES gfxbench shaders.

Fixes: #6585
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17546>

vulkan: Common vk_format_get_component_bits

RADV and PowerVR use the same implementation.
Turnip does use a slightly modified version but the helper only has one
use -> just inline it and get rid of turnip's vk_format.h.

Signed-off-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17515>

build(glx): Fix build by adding missing deps

dri3_glx.c includes xshmfence and glxcmds.c includes xf86vm, neither of
which are listed as dependencies of the glx lib in the meson.build file.
Consequently, those files would fail to compile on machines that did not
have xshmfence and xf86vm installed globally. This commit rectifies the
issue by adding the missing dependencies

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17585>

mesa/st: add implicit zeroing of clipdistance array

GL drivers have an implicit default of "in bounds" for unwritten clipdistance
values, but some (layered) drivers have to deal with api mismatch which
prevents that implicit value from being used after the shader
gets mangled by various compiler passes

to avoid issues here, write out all members of the clipdistance array
with zeroes at the very start of the shader such that any values the
shader actually writes will naturally overwrite the implicit zero and
any unwritten values will now be written as zero

fixes #6845

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17498>

zink: tu a630 baseline update

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17606>

zink: nv baseline update

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17606>

kopper: Implement {EGL,GLX}_EXT_buffer_age

Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17527>

zink: break out tc/trace context unwrapping

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17527>

venus: filter out VK_EXT_physical_device_drm on the driver side

Fixes: a1a22862c62 ("venus: implement VK_EXT_physical_device_drm")

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Reviewed-by: Ryan Neph <ryanneph@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17591>

kms/dri: add mutex lock around map/unmap

this can get called from multiple threads with the recent llvmpipe
overlapping rendering changes, so make sure to lock around the
map/unmapping so they can't race.

This should fixes some crashes seen with kwin.

Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Tested-by: Adam Williamson (Fedora)
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17531>

radv: implement PS epilogs

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17485>

aco: add support for compiling PS epilogs

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17485>

aco: refactor export_fs_mrt_color() for PS epilogs preparation

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17485>

radv,aco: rename radv_aco_build_prolog to radv_aco_build_shader_part

Will be re-used for PS epilogs.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17485>

aco: add aco_postprocess_shader() helper

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17485>

aco: prevent adding DONE/VM to the last export if the FS has an epilog

If the fragment shader exports MRTZ and the epilog some color exports,
DONE/VM should be added to the last export.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17485>

aco: do not abort if the FS doesn't export anything but has an epilog

The main fragment shader can only export MRTZ (if present) and the
epilog will export colors.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17485>

aco: emit p_jump_to_epilog if the main fragment shader has an epilog

MRTZ is still exported from the main shader.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17485>

aco: add new pseudo instruction p_jump_to_epilog

The first operand of this new pseudo-instruction is a 64-bit SGPR for
the continue PC, followed by a variable list of fixed VGPRS for the
color exports which are the PS epilog inputs.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17485>

radv: add a function that declares PS epilog shader arguments

The PS epilog would be a "normal" compiled shader using RA, etc. It
will declare up to 8x4 VGPRs for all color exports.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17485>

radv: declare a new user SGPR arg in FS for the epilog PC

The main FS would have to jump to the PC of the PS epilog. Given that
shaders are allocated in the 32-bit addr space, one user SGPR is fine.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17485>

radv,aco: introduce {radv,aco}_ps_epilog_key

To pass the necessary pipeline information for compiling PS epilogs.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17485>

radv,aco: track if a fragment shader needs an epilog

This is currently disabled but it will be used for testing first,
and then for graphics pipeline libraries.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17485>

frontends/va: WA for ffmpeg 10bit encoding crash

When doing 10bit encoding in ffmpeg it uses
VaDeriveImage, and that could result in missing
mapping the chroma buffer of the input frame.

This WA to disallow ffmpeg using VaDeriveImage
function, so that VaCreateImage and VaPutImage can
be used and WA the chroma buffer mapping issue.

Reviewed-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Ruijing Dong <ruijing.dong@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17472>

frontends/va: do texture_map when needed

When map buffer, and its target is texture,
texture_map/unmap need to be used.

Reviewed-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Ruijing Dong <ruijing.dong@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17472>

dzn: expose VK_KHR_driver_properties

We're not quite conformant with the extension, because we don't have
a valid conformance version.

That's not a quick-fix, so we should probably just accept some failures
for now.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16998>

dzn: fill misc props

This is just a bag of misc properties that we should fill in.

Not all of them are filled out super accurately, but this is the best we
can do for now.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16998>

dzn: fill in minmax props

This should be possible to support, but we don't support minmax blending
at all yet, so let's leave these as unsupported for now.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16998>

dzn: fill in depth/stencil resolve props

Before enabling Vulkan 1.2 support, we need to fix the TODO in here.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16998>

dzn: fill in bindless props

These might not be exactly right, but they are good enough for now.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16998>

dzn: fill in non-uniform-indexing props

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16998>

dzn: fill in float-control details

We can do better here in the future, but this is what's supported right
now.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16998>

dzn: fill in driver name and info

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16998>

radv: Only set rt stack size for dynamic stacks

When using a static callable stack, the required scratch has already
been allocated.
Dynamic stacks are located at the end of scratch memory
and are allocated on demand using radv_set_rt_stack_size.
Static stacks live at the start of scratch memory and are allocated in
create_rt_shader by setting scratch_size.

Signed-off-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-By: Tatsuyuki Ishi <ishitatsuyuki@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17579>

ac/nir/ngg: add a barrier before prim id export

When culling enabled, it will use LDS space, which overlap with
the prim id export.

Fixes: e97f0463a8f ("ac/nir: Implement NGG deferred attribute culling in NIR.")
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/17593>

ac/nir/ngg: fix nogs culling scratch size

Should be in bytes not dwords.

Fixes: e97f0463a8f ("ac/nir: Implement NGG deferred attribute culling in NIR.")
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/17593>

radv: Remove trailing whitespace introduced by DGC commits.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17580>

radv: Only initialize DGC state when DGC is enabled.

This function causes a crash with RADV_DEBUG=llvm and this commit
works around that crash.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17580>

zink: inject a 0,0,0,1 clear for RGBX formats

this ensures the alpha component is full if it must be read for fbfetch

fixes (RGBX swapchain config):
KHR-GL46.blend_equation_advanced*

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: simplify zink_framebuffer_clear_data union

no functional changes

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: delete srgb tracking for clears

no longer used

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: delete zink_fb_clear_util_unpack_clear_color

no longer used

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: remove out-of-renderpass clears

these are only ever going to hurt tiler perf, so remove the footgun

this also means there's no more srgb format conversion needed, so delete
all of that too

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: lift clearing on fb state change up a level

in the scenario where:
* at least 1 color buffer was bound and a depth buffer was bound
* no color clear was enabled
* a zs clear was enabled
* the zs clear was never flushed
* the zs clear needs a renderpass
* the fb state changes

the color buffer(s) would be unbound, following which the depth buffer unbind
would trigger a renderpass, which would utilize the just-unbound color buffers,
which have no batch tracking, thus creating a case where the surface was destroyed
while it was still in use

cc: mesa-stable

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: clamp color clear values based on format

formats like GL_RGB10_A2UI can be cleared with out of range values,
so to ensure consistent driver behavior, pre-clamp to the valid range

affects:
KHR-GL46.direct_state_access.renderbuffers_storage_multisample

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: add explicit (awful) handling for fb layer mismatch clears

this is terrible and (hopefully?) rare, so just force it out early to
avoid any kind of issue

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: track a bitmask of fb attachments with mismatched layer counts

these need special handling for clears

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: always use storeOp=STORE for depth renderpass

it's unknown whether there may be clears to the depth attachment at the start of
a renderpass, so always assume there will be

Fixes: c132a287455 ("zink: use store op NONE when necessary for depth usage")

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: remove u_blitter usage from zink_clear_render_target

this is more operations than needed

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: remove non-renderpass clear path from zink_clear_texture

this should always be faster

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: massively simplify zink_clear_depth_stencil

this now just uses renderpass clears

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: improve zink_clear_depth_stencil check for current attachment

this is technically more correct since it accounts for multi-context usage

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: stop using u_blitter for texture clears

this was really stupid: instead of just binding a new fb and firing off
a clear, the code was calling u_blitter to bind a new fb and do actual
draws

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: fix coverage check for texture clears

this wasn't actively harmful, but it was potentially differently-performant

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: remove format check from clear texture

this is going to be a render target no matter what

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: fix transient attachment rp assert

cc: mesa-stable

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

zink: remove bogus range tracking from texture clear

too much copy/paste

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17366>

lavapipe: enable variablePointers

This passes the CTS with no regressions.

Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17587>

lavapipe: drop unreachable pNext checks.

These are reachable, and
dEQP-VK.api.smoke.triangle_ext_structs,Crash is why.

Signed-off-by: Dave Airlie <airlied@redhat.com>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17586>

gallium: LLVM-15 contexts use non-opaque pointers

LLVM-15 enables opaque pointers by default. We temporarilly request
non-opaque pointers while we migrate our code to support non-opaque pointers.

This workaround needs to be removed before LLVM-16.

See #6615

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Kai Wasserbäch <kai@dev.carbon-project.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17514>

freedreno/gmem: Reverse order of alternative tile rows

Similar motivation as c426e21ff14 ("turnip: Reverse the order of walking
pipes or tiles on odd rows."), but instead we just swap the order of
alternate rows of fd_tile the the gmem stateobj.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17303>

Revert "CI: Disable Collabora lab"

This reverts commit 7a336c97ef692ed96cc93394596a7d0650983874.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17563>

radeonsi: Enable nir_lower_point_smooth lowering pass for point smoothing

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15117>

ac/llvm: Implement nir_intrinsic_load_point_coord_maybe_flipped opcodes

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15117>

ac/llvm : Adding Number of all interpolated inputs in ac_shader_abi

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15117>

nir: Add a lowering pass for point smoothing

When point smoothing is enabled then this lowering pass will
modifies the alpha component of every write to fragment output.
Anti-aliased points get rounded with respect to their radius instead
of square.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15117>