vulkan: add support for VK_IMAGE_LAYOUT_ATTACHMENT_FEEDBACK_LOOP_OPTIMAL_EXT

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17883>

vulkan: Update the XML and headers to 1.3.224

Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Acked-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17883>

freedreno: Drop fixed upper bound on # of tiles

Placate things that think 16k x 16k FBO is a good idea.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6997
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17888>

freedreno/gmem: Fix col0 calc

Fix typo in calculation of position of start of a row of tiles.  This
could otherwise cause an out-of-bounds access in the next patch.

Fixes: 81d85be9a5c freedreno/gmem: Reverse order of alternative tile rows
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17888>

freedreno/drm: Fix potential bo cache vs export crash

Keep the list head valid (empty) after allocation from bo cache.  Avoids
a potential later crash in lookup_bo in the following sequence:

1. alloc, bo cache hit
2. export
3. re-import

Cc: mesa-stable
Fixes: f3cc0d27475 ("freedreno: import libdrm_freedreno + redesign submit")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6988
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17888>

virgl/ci: Update virglrenderer

Update virglrenderer to the latest version on time.

Signed-off-by: Corentin Noël <corentin.noel@collabora.com>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17853>

ci: separate wine setup into own script

It will be used by LAVA jobs too.

Reviewed-by: "Sergi Blanch Torne" <sergi.blanch.torne@collabora.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17887>

ci: make shellcheck happy on dxvk script

Reviewed-by: "Sergi Blanch Torne" <sergi.blanch.torne@collabora.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17887>

ci: move DXVK instalation outside of x86_test-vk

It will be used by LAVA jobs.

Reviewed-by: "Sergi Blanch Torne" <sergi.blanch.torne@collabora.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17887>

d3d12: Fixes compile error with mingw/gcc-x64 when static linkage to runtime library

Closes #6968

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Suggested-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Tested-by: Prodea Alexandru-Liviu <liviuprodea@yahoo.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17889>

ci: Turn off the entire Lima farm (devices report out-of-space)

Reference: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7009

Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17910>

anv: Don't require 32-bit addresses for scratch on Gen12.5+

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17908>

genxml: Add BVH data structures

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17908>

intel/rt: Handle multiple exits in lower_shader_returns

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17908>

intel/rt: Handle halts in any-hit shaders properly

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17908>

intel/fs_reg_allocate: Improve compressed instruction self-interference

The old version worked fine for SIMD16 instructions but SIMD8
instructions where the destination spans two registers have the same
problem.

Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17908>

intel/nir: specify synchronous value for tracing op

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17908>

intel/compiler: document units of brw_ubo_range fields

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17908>

intel/fs: fixup simd selection with shader calls

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17908>

intel/fs: store num of resume shaders in prog_data

That way we can look at the SBT entries for debug purposes.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17908>

r300: add khr r400 failures

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17895>

r300: add list of deqp gles2 r400 failures

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17895>

broadcom/compiler: use nir_opt_idiv_const

total instructions in shared programs: 12463625 -> 12463571 (<.01%)
instructions in affected programs: 1758 -> 1704 (-3.07%)
helped: 12
HURT: 0

total uniforms in shared programs: 3704589 -> 3704591 (<.01%)
uniforms in affected programs: 17 -> 19 (11.76%)
helped: 0
HURT: 1

total max-temps in shared programs: 2148088 -> 2148138 (<.01%)
max-temps in affected programs: 170 -> 220 (29.41%)
helped: 0
HURT: 10

Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17871>

broadcom/compiler: don't use imprecise_32bit_lowering for idiv lowering

This is known to produce bogus results for certain combinations of
operands, so don't use it. See this issue for details:

https://gitlab.freedesktop.org/mesa/mesa/-/issues/6555

With this change, the idiv lowering will produce mul_high instructions,
so we need to instruct the compiler to lower those with the ALU lowering
right after the idiv lowering by adding the lower_mul_high option (we
only need to add this to V3D, since V3DV already had it set). This will
cause injection of uadd_carry instructions, for which we have backend
implementations that produce better code for us than the NIR lowering.

total instructions in shared programs: 12457692 -> 12463625 (0.05%)
instructions in affected programs: 23115 -> 29048 (25.67%)
helped: 0
HURT: 111

total threads in shared programs: 416372 -> 416368 (<.01%)
threads in affected programs: 8 -> 4 (-50.00%)
helped: 0
HURT: 2

total uniforms in shared programs: 3704067 -> 3704589 (0.01%)
uniforms in affected programs: 5804 -> 6326 (8.99%)
helped: 2
HURT: 109

total max-temps in shared programs: 2147845 -> 2148088 (0.01%)
max-temps in affected programs: 2456 -> 2699 (9.89%)
helped: 6
HURT: 91

Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17871>

ac/llvm: handle external textures in ac_nir_lower_resinfo

Fixes: 4f622d62d0dd8 - ac/nir: add ac_nir_lower_resinfo
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6993

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17902>

radeonsi: fix a regression due to reordering PIPE_SHADER_*

Fixes: 27f46465c7a408d3 - gallium/tgsi: reorder pipe shader type defines.

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17902>

radeonsi/ci: update failing tests on navi21

These pass now.

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17902>

radeonsi: don't assume that TC_ACTION_ENA invalidates L1 cache on gfx9

Just got into a midnight discussion with a hw guy.
TC_ACTION_ENA apparently doesn't invalidate L1, so don't clear
the INV_VCACHE flag.

Fixes: 4056e953fe43bd667 - radeonsi: move emit_cache_flush functions into si_gfx_cs.c

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17902>

iris: Handle new untyped dataport cache flush PIPE_CONTROL field

Also while switching to GPGPU pipeline, make sure to flush the untyped
dataport cache. HDC pipeline flush bit must be set if we are flushing
untyped dataport L1 data cache.

v2: Add utrace support (Lionel)

Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16905>

anv: Handle bits to flush data-port's Untyped L1 data cache

v2: Drop ANV_PIPE_UNTYPED_DATAPORT_CACHE_FLUSH_BIT from invalidate bits (Lionel)
    Add utrace support
    Expand on comment about PIPE_CONTROL::UntypedDataPortCache

Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16905>

intel/ds: track untyped dataport flushes

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16905>

iris: Specify Untyped L1 cache policy for stateless accesses

Set write back L1 cache policy in STATE_BASE_ADDRESS instruction for A64
messages.

Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Suggested-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16905>

anv: Specify Untyped L1 cache policy for stateless accesses

Set write back L1 cache policy in STATE_BASE_ADDRESS instruction for A64
messages.

v2: Also set the value in genX_state.c (Lionel)

Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Suggested-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16905>

intel/isl: Setting L1 caching policy to Write-back mode

For a RW L1 cache, both reads and writes are cached in the L1, at high
priority (MRU position). For a RO L1 cache, reads are cached at higher
priority and writes bypass the cache.

v1: (Ken)
- Set caching policy for buffer surfaces too

Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16905>

anv: fixup PIPE_CONTROL restriction on gfx8

We're missing a condition that is currently papered over by having
ANV_PIPE_HDC_PIPELINE_FLUSH_BIT in the invalidate bits.

v2: rework with simplication (Caio)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: mesa-stable
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16905>

vc4: properly restore vc4 debug option

Otherwise VC4_DEBUG does not work.

Fixes: c3f5d276317 ("vc4/v3d: restore calling debug_get_option_vc4/v3d_debug")
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17882>

draw: don't touch info values that aren't valid.

These shouldn't be accessed, and shows up as an uninit access in
valgrind with piglit rasterpos

Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10641>

llvmpipe/ci: update ci results for clover.

Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10641>

lavapipe: scan shader for info before lowering images.

Otherwise the values will be wrong.

Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10641>

lavapipe: lower images to non-derefs in vulkan side

Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10641>

gallivm/nir/st: lower image derefs in advance.

This improves clover from crashing to just failing, but I mainly
want it this to cleanup the nir code first

It's also important the shaders coming from the state tracker
for feedback get images lowered when they are draw shaders now.

Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10641>

virgl/vtest: fix memory overwrite problem in virgl_vtest_send_get_caps()

Signed-off-by: Feng Jiang <jiangfeng@kylinos.cn>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17901>

venus: double the abort timeout to allow long shader compiles

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/17898>

venus: increase busy wait order in vn_relax

This is mainly to workaround a platform issue that has huge sleep
penalty, which could lead to a timeout if the small synchronous queries
are going to sleep.

This change adjusts the warn and abort order correspondingly so that to
match prior timing.

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/17898>

zink: remove an old RGBA4 nvidia workaround

this should be automatically handled

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: don't fixup sparse texops

this is broken, and these will never need to be fixed

Fixes: 3a475766876 ("zink: add a compiler pass to match up tex op dest types")

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: add all format modifiers when adding for dmabuf export

adding LINEAR before was a good starter step, but LINEAR
might not actually be supported (e.g., nvidia)

cc: mesa-stable

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: don't add modifiers if EXT_image_drm_format_modifier isn't present

cc: mesa-stable

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: use modifier_aspect to check for modifier plane in zink_resource_get_param

cc: mesa-stable

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: demote dmabuf tiling to linear if modifiers aren't supported

this is effectively the same as LINEAR, and it still allows dmabuf creation

cc: mesa-stable

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: add u_queue.h to kopper header

types are used, so include the header

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

nine: check return on resource_get_handle

this has a return code, and if it return false, this is probably an
exit condition

cc: mesa-stable

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: support PIPE_QUERY_TIMESTAMP_DISJOINT

the key point here is that disjoint is always false, but also return a handwavy
guess at frequency value based on timestampPeriod

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: export PIPE_CAP_MULTISAMPLE_Z_RESOLVE

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: rewrite all undefined shader reads as 0001 instead of undef

this is a little less hostile towards broken/buggy apps

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: handle invalid gl_TexCoord reads

if a texcoord input is read in a fragment shader but not written in
the previous stage, the texcoord variable must be preserved until
nir_lower_texcoord_replace has (possibly) run, and only then can it
be replaced with a 0000 read

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: explicitly set nir cursor in rewrite_and_discard_read

not sure if this is a bug but it looks weird

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: fix return for PIPE_CAP_DEPTH_CLIP_DISABLE

this uses the extension now

Fixes: 21ea19d504b ("zink: Always enable depth clamping, make depth clipping independent.")

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: handle !half_pixel_center

the shader is already getting a -0.5,-0.5 bias, but the viewport also
needs to be shifted by 0.5 to match

cc: mesa-stable

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: handle unscaled depth bias from nine

nine uses this to pass unscaled units for depth bias, which means
the units must be scaled based on the format of the depth buffer

cc: mesa-stable

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: drop mode_changed check from linewidth/depthbias draw updates

this doesn't need to be updated on primtype change since it's always
set

cc: mesa-stable

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: allow sw loading with D3D_ALWAYS_SOFTWARE for nine

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: use maintenance4

just by enabling this there are fewer validation warnings

truly incredible.

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: force a new framebuffer for clear_depth_stencil if the clear region is big

can't clear outside the framebuffer, so set a new one if necessary

Fixes: f1f08e3529d ("zink: massively simplify zink_clear_depth_stencil")

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

zink: force flush clears on fb change if fb geometry changes

Fixes: 66ceea7ed9a ("zink: lift clearing on fb state change up a level")

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17775>

tu: Implement VK_EXT_vertex_input_dynamic_state

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17554>

tu: Decouple vertex input state from shader

Emit VFD_DECODE and VFD_DEST separately, similarly to what Gallium does.
This means we emit a few more VFD_DECODE for binning shaders and when
there are unused attributes, but hopefully the overhead won't be too
much. In exchange we lose one draw state, and in the future we can
pre-compute the dynamic vertex state independently of the shader, so
there should be lower CPU overhead with dynamic vertex inputs.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17554>

freedreno: Disable UBWC for frontbuffer rendering

Note that glamor would still need to set this new
GBM_BO_USE_FRONT_RENDERING flag.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6865
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17799>

gbm: Add USE_FRONT_RENDERING flag

Provide a use flag to let the driver know that the allocated buffer will
be used for frontbuffer rendering.  For example, bandwidth compressed
formats should usually be avoided for frontbuffer rendering to avoid the
visual corruption that results from this display racing with the GPU
writing header data vs. pixel data.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17799>

ci/zink: Re-enable a630 zink-on-turnip traces job using gbm as the backend.

This is a workaround for #6753, without having to build and install a
custom xlib in our containers.  Maybe some day debian will have a
backport?

Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17549>

zink: Replace the "optimal_tiling" flag with a "linear" flag instead.

That's what it actually means now.

Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17549>

zink: Take the tiling path for resources with modifiers.

transfer_map (the only consumer of the flag now) was doing direct mapping
on resources with modifiers (UBWC compressed images on turnip) and
accessing them as linear.  We definitely want the staging blit path
instead.

Fixes glReadPixels() from gbm surfaces on turnip.

Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17549>

util/memstream: insert null terminator on windows

Libc's open_memstream is doing so on Unix systems already.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17334>

gallium: mark the input data as const in pipe_grid_info

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17334>

vtn: silence warning about linkage

For OpenCL kernels we simply link together SPIR-V files, so the only case
where we are left with linking shaders together is libclc and we handle
that just fine.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17334>

clc: undefine spirv defs to work around LLVMs headers

Clang unconditionally adds those definitions if using a spirv LLVM target.

That's not a problem on its own, but clang's internal OpenCL header enable
a bunch of OpenCL extensions if those are set.

Lucky for us, we can simply undefine them and spare us the trouble of
finding an upstream solution to this problem :)

This fixes the OpenCL CTS' compiler features_macro test.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17334>

microsoft/clc: drop inline sampler reordering

nir_dedup_inline_samplers is now taking care of it.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17334>

nir: Clean up and improve nir_dedup_inline_samplers

It now removes dead inline sampler variables and moves everything to the
end so we no longer need nir_move_inline_samplers_to_end().

Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17334>

nir: extract the clc inline sampler dedup pass from clc

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17334>

nir/lower_images: extract from clover

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17334>

nir: serialize printf metadata for CL kernels

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17334>

printf: add some unit tests

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17334>

printf: extract clovers printf impl

Also make the code cleaner and simplier.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Acked-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17334>

ci/bin: ignore Gitlab GQL cache

Fixes: 266e3a627fbf ("ci/bin: Cache GQL queries")

Reviewed-by: Guilherme Gallo <guilherme.gallo@collabora.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17885>

radv/nir_lower_abi: Use instructions_pass

nir_shader_instructions_pass is sufficient and it should be a bit faster
than nir_shader_lower_instructions.

Signed-off-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17727>

meson: Cleanup usage dri_drivers as it's a dead option

Initial value of with_dri have no relation with dri_drivers.length()
Do not show useless `DRI drivers:`

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17821>

meson: Place all pre_args about HAVE_DRI* into single place

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17821>

docs: update calendar for 22.2.0-rc1

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17890>

docs: update calendar and link releases notes for 22.1.5

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17890>

docs: update calendar and link releases notes for 22.1.4

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17890>

docs: Add sha256 sum for 22.1.4

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17890>

docs: add release notes for 22.1.4

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17890>

docs: Add sha256 sum for 22.1.5

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17890>

docs: add release notes for 22.1.5

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17890>

util/disk_cache: Add new mesa-db cache type

Introduce new cache type, the Mesa-DB. This is a single-file read/write
cache that is based on the read-only Fossilize DB cache. Mesa-DB supports
cache size capping. It's a much more efficient cache than the multi-file
cache because Mesa-DB doesn't have the inode overhead. The plan is to make
Mesa-DB the default cache implementation once it will be deemed as stable
and well tested. For now users have to set the new MESA_DISK_CACHE_DATABASE
environment variable in order to active the Mesa-DB cache.

Mesa-DB cache is resilient to corrupted cache files and doesn't require
maintenance from users and developers. The size capping is implemented
by evicting least recently used cache items and compacting the cache
database files with the evicted entries. In order to prevent frequent
compaction of the cache, at minimum a half of cache is evicted when cache
is full.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16888>

util/disk_cache: Make disk_cache_os.h usable by C++ code

The cache tests are written in C++ and the upcoming new Mese-DB cache
tests will need to include disk_cache_os.h that misses extern "C"
modifier required by C++ linkage. Add the modifier.

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16888>

util/disk_cache: Move struct cache_entry_file_data to the disk_cache_os.h

Relocate struct cache_entry_file_data to the header file, making it
accessible to the cache testing code. This is a preparatory step towards
addition of the new Mesa-DB cache type.

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16888>

util/disk_cache: Append to disk_cache_load_cache_index() with _foz

The disk_cache_load_cache_index() is solely about the foz-db cache and
we're going to add the new cache type. Append the function name with
_foz postfix to improve the naming, making clear what that function is
about.

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16888>

util/disk_cache: Enable testing of uncompressed caches

Mesa caches are compressed by default and we just added option to disable
compression that will be needed by the new Mesa-DB cache. Let's enable
testing of uncompressed caches for the multi-file and foz-db caches in
addition to the compressed to increase test coverage of the new code.

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Suggested-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16888>

util/disk_cache: Add option to disable compression

The shader cache files produced by multi-file cache have a minimum size
limitation imposed by filesystem, usually it's 4KB. In order to make
cache tests suitable for a single file caches, we need to bypass the data
compression, making size of written out data determined for the eviction
testing.

The disk_cache_create() now checks whether driver_id name is set to
"make_check_uncompressed" in order to disable cache data compression.

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16888>