radv: Avoid deadlock on bo_list.

With the kernel timeline sysncobj changes, the kernel submits do
not necessarily happen in global vkQueueSubmit order. Which should
be fine, we added the appropriate waits for that. (See
DRM_SYNCOBJ_WAIT_FLAGS_WAIT_FOR_SUBMIT in the winsys)

However, all kernel submissions take a lock on the bo_list mutex,
and since we do the wait in the winsys, we wait while having the
bo_list mutex held. This means that as soon as a wait and a signal
submission are out of order we have a deadlock on the bo_list mutex
and the wait.

Solution is to use a shared reader lock during the kernel submission,
as we only need read access for the submission.

Fixes: 6bc5ce7a91d "radv: Add timeline syncobj for timeline semaphores."
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3446
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6478>

radv: Fix threading issue with submission refcounts.

If decrement == 0 then:

- it isn't safe to access the submission
- even if it is, checking that the result of the atomic_sub is 0
  doesn't given an unique owner anymore.

So skip it. The submission always starts out with refcount >= 1,
so first one to decrement to 0 still get dibs on executing it.

Fixes: 4aa75bb3bdd "radv: Add wait-before-submit support for timelines."
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6478>

intel/tools: Disassemble WAIT's argument as a destination

WAIT takes a notification register as a destination and a src0 argument.
Since the same notification register is specified in both fields, we
treat it as a special case and disassemble it only once.

If we disassemble it as if it is a source register, its scalar region
will be printed as <0,1,0>. This causes difficulties round-tripping
through the assembler <-> disassembler because that is not an acceptable
destination region. If we instead disassemble the destination, we
instead get a <1> region which is an acceptable and equivalent region
for source and destination.

The test .asm files are regenerated by round-tripping them through the
assembler/disassembler. Note that the <0> region in the tests was a
harmless mistake: the compiler translated it to a <0,1,0> source region
and a <1> destination region, since <0> isn't valid.

Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6543>

gallium/tgsi_exec: Fix up NumOutputs counting

We can get duplicate declarations for an index (for example dvec3 + float
packed into 2 vec4s, the second one won't pack into the first's array
decl), and we'd end up stepping by the wrong amount in GS vtx/prim emit.

Fixes vs-gs-fs-double, sso-vs-gs-fs-array-interleave piglit tests.

Fixes: 49155c3264d0 ("draw/tgsi: fix geometry shader input/output swizzling")
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6567>

gallium/tgsi_exec: Add missing DFLR opcode support.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6567>

nir/clone: Add a helper for cloning most instruction types

@anholt needed it for nir_to_tgsi, and the desire comes up frequently.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6567>

nir/opt_vectorize: Add a callback for filtering of vectorizing.

For NIR-to-TGSI, we don't want to revectorize 64-bit ops that we split to
scalar beyond vec2 width.  We even have some ops that we would rather
retain as scalar due to TGSI opcodes being scalar, or having more unusual
requirements.

This could be used to do the vectorize_vec2_16bit filtering, but that
shader compiler option is also used in algebraic so leave it in place for
now.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6567>

nir: Add simplistic lowering for bany_equal/ball_inequal.

It would be nice if we could do swizzling of an expression on the
replacement side so that we could have a single ieq/ine of the vector
after CSE.  However, if you do want vector operations, nir_opt_vectorize()
does just fine.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6567>

gallium/ureg: Set the next shader stage from the shader info.

Saves a loop over the linked shaders in glsl_to_tgsi which the GLSL linker
has already done for us.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6567>

gallium/tgsi: Add a helper for initializing ureg from a shader_info.

This moves a bunch of code from glsl_to_tgsi that will be reused by
tgsi-to-nir.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6567>

gallium/tgsi: Add some missing opcodes to tgsi_ureg.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6567>

gallium/tgsi: Add support for PRIMITIVEID as a system value.

NIR always represents this as a system value, so for NIR-to-TGSI we need
this support.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6567>

gallivm: Report the unsupported intrinsic instead of just assert(0);

Tripped over this in rendermode with softpipe NIR.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6567>

ci: Do not create manual test-docs job in post-merge pipelines

Only in pre-merge pipelines for MRs, or in pipelines for forked project
branches.

Having the manual job in post-merge pipelines prevented the pages job
from running automatically as well, which could prevent the public
website from getting updated.

Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6534>

intel/fs,vec4: remove unused assignments

Reported by Coverity.

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6126>

intel/compiler: mark debug constant as const

Should quiet Coverity's "'Constant' variable guards dead code".

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6126>

intel/compiler/test: use TEST_DEBUG env var consistently

Other tests use the same environment variable to decide whether they
should print debugging information.

Will quiet Coverity's "'Constant' variable guards dead code".

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6126>

intel/perf: don't generate logically dead code

When divisor is constant integer != 0 there's no point in checking
whether it's 0.

Complained about by Coverity.

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6126>

i965: verify format width in blorp_get_client_bo

Coverity complains about possible "Division or modulo by zero".

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6126>

iris: verify color component width in convert_fast_clear_color

Coverity complains about possible 2^32-1 bit shift.

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6126>

iris: handle os_dupfd_cloexec failure

Found by Coverity.

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6126>

freedreno/computerator: Use a render node

Fixes headless systems.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6562>

intel/compiler: Fix pointer arithmetic when reading shader assembly

start_offset is a byte offset.

Fixes: 04a995158084acbd1917b4c7e0f8d381e1c9222d
Signed-off-by: Danylo Piliaiev <danylo.piliaiev@globallogic.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6557>

mesa/st: Don't set alpha if ALPHA_TEST is lowered

When PIPE_CAP_ALPHA_TEST is zero, the driver does not support alpha
testing, so alpha shouldn't be set. In particular, alpha.enable should
be zero, since logically alpha testing is not used in the ZSA CSO when
it's lowered in the fragment shader key.

Fixes failing asserts in kicad, rvgl, etc with Panfrost since 6afd4ad.
(We could remove the assert in panfrost instead, but logically setting
alpha.enabled on top of lowering the shader seems wrong?)

As Erik pointed out, this should improve CSO cache behaviour.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reported-by: Icecream95 <ixn@keemail.me>
Tested-by: Urja Rannikko <urjaman@gmail.com>
Tested-by: Robin Murphy <robin.murphy@arm.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Fixes: 6afd4addefa ("panfrost: Simplify depth/stencil/alpha")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6523>

turnip: Implement VK_EXT_host_query_reset

Signed-off-by: Hyunjun Ko <zzoon@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6299>

turnip: Support pipeline statistics query

Signed-off-by: Hyunjun Ko <zzoon@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6299>

turnip: Refactor structs of tu_query

Since there are different number of results depending on query types,
this patch removes the result field out of the common struct and defines
query-specific results in each type of query struct.

Signed-off-by: Hyunjun Ko <zzoon@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6299>

intel/assembler: Add labels support

Use labels instead of numeric JIP/UIP offsets.
Works for gen6+.

v2:
 - Change asm tests to use labels on gen6+
 - Remove usage of relative offsets on gen6+
 - Consider brw_jump_scale when setting relative offset
 - Return error if there is a JIP/UIP label without matching target
 - Fix matching of label tokens

Signed-off-by: Danylo Piliaiev <danylo.piliaiev@globallogic.com>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4245>

intel/disasm: Label support in shader disassembly for UIP/JIP

Shader instructions which use UIP/JIP now get formatted with a label
in addition with immediate value, labels have "LABEL%d" format.

v2: - Consider brw_jump_scale when calculating label's offset

From: "Lonnberg, Toni" <toni.lonnberg@intel.com>
Signed-off-by: Danylo Piliaiev <danylo.piliaiev@globallogic.com>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4245>

intel/disasm: brw_label and support functions

Pre-work for shader disassembly label support.

Introduction of the structures and functions used by the shader disassembly
jump target labeling.

From: "Lonnberg, Toni" <toni.lonnberg@intel.com>
Signed-off-by: Danylo Piliaiev <danylo.piliaiev@globallogic.com>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4245>

intel/disasm: Change visibility of has_uip and has_jip

Pre-work for shader disassembly label support.

From: "Lonnberg, Toni" <toni.lonnberg@intel.com>
Signed-off-by: Danylo Piliaiev <danylo.piliaiev@globallogic.com>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4245>

anv: refresh cached current batch bo after emitting some commands

Fixes crashes in:
- Rise of the Tomb Rider (on benchmark start)
- Total War: Three Kingdoms (on game start)
- Total War: Warhammer II (on game start)

Fixes: 34a0ce58c7f ("anv: add a new execution mode for secondary command buffers")
Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6546>

radeonsi: use radeonsi_clamp_div_by_zero for SPECviewperf13, Road Redemption

Fixes SPECviewperf 13 creo rendering.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/2639
Cc: mesa-stable
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6259>

radeonsi,driconf: add clamp_div_by_zero option

Cc: mesa-stable
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6259>

ac/llvm: add option to clamp division by zero

Replace div(x) by min(div(x), FLT_MAX)) to avoid getting a NaN result
when x is 0.

A cheaper alternative would be to use legacy mult instructions but they're
not exposed by LLVM.

Cc: mesa-stable
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6259>

radeonsi: enable PIPE_CAP_NO_CLIP_ON_COPY_TEX

This fixes specviewperf13 catia test rendering.

Cc: mesa-stable
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6259>

mesa/st: introduce PIPE_CAP_NO_CLIP_ON_COPY_TEX

If supported this means that src_x/src_y/width/height parameters of
CopyTex functions will not be clipped using the read framebuffer's dimensions.

Cc: mesa-stable
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6259>

gallium/swr: Fix compilation TCS/TES compilation issues

Gallvm API changes in TCS/TES, but the compilation worked
for older LLVM APIs. With LLVM 12 update is needed.

Reviewed-by: Krzysztof Raszkowski <krzysztof.raszkowski@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6554>

gallium/swr: Fix compilation with LLVM 12

Update to reflect changes in VectorType and ElementCount APIs

Closes: #3457
Reviewed-by: Krzysztof Raszkowski <krzysztof.raszkowski@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6554>

spirv: fix retrieving dest type for OpFragmentMaskFetchAMD

Since the dest type is retrieved from the SPIR-V return type now,
we have to set it manually for OpFragmentMaskFetchAMD. The result
type must be a 32-bit unsigned integer type scalar.

Fix dEQP-VK.pipeline.multisample.shader_fragment_mask.* with RADV.

Fixes: a196f05fc22 ("nir/vtn: Use return type rather than image type for tex ops")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.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/6533>

vulkan: make VK_TIME_DOMAIN_CLOCK_MONOTONIC_RAW_EXT conditional

Only advertise VK_TIME_DOMAIN_CLOCK_MONOTONIC_RAW_EXT if CLOCK_MONOTONIC_RAW
is defined.  Fixes the build on OpenBSD which has CLOCK_MONOTONIC but not
CLOCK_MONOTONIC_RAW.

Fixes: 67a2c1493c0 ("vulkan: Add VK_EXT_calibrated_timestamps extension (radv and anv) [v5]")
Signed-off-by: Jonathan Gray <jsg@jsg.id.au>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6517>

radv: remove seccomp includes

57c95d2ce23 ("radv: a support for a secure compile fork at device creation")
added includes which were not removed in
7324977e420 ("radv: remove the secure compile support feature")

remove these no longer needed includes to fix the non-linux build

Fixes: 7324977e420 ("radv: remove the secure compile support feature")
Signed-off-by: Jonathan Gray <jsg@jsg.id.au>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6517>

util/os_misc: os_get_available_system_memory() for OpenBSD

Return the smallest value of available non-kernel physical memory and
the static per process data size limit as the amount of available
system memory on OpenBSD.

Fixes: b80930a6fea ("anv: add support for VK_EXT_memory_budget")
Signed-off-by: Jonathan Gray <jsg@jsg.id.au>
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6517>

anv: use os_get_available_system_memory()

Replace local get_available_system_memory() function with
os_get_available_system_memory().

Fixes: b80930a6fea ("anv: add support for VK_EXT_memory_budget")
Signed-off-by: Jonathan Gray <jsg@jsg.id.au>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6517>

util/os_misc: add os_get_available_system_memory()

Add os_get_available_system_memory() derived from
src/intel/vulkan/anv_device.c get_available_system_memory()

Fixes: b80930a6fea ("anv: add support for VK_EXT_memory_budget")
Signed-off-by: Jonathan Gray <jsg@jsg.id.au>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6517>

anv: use os_get_total_physical_memory()

Replace non-portable sysinfo() use with Mesa's
os_get_total_physical_memory()

Fixes: 060a6434eca ("anv: Advertise larger heap sizes")
Signed-off-by: Jonathan Gray <jsg@jsg.id.au>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6517>

freedreno/a6xx: Add ARB_depth_clamp and separate clamp support.

Passes piglit depth_clamp, depth-clamp-range,
amd_depth_clamp_separate_range.  This is part of enabling GL 3.2 (the
other is bumping PIPE_CAP_GLSL_FEATURE_LEVEL, which I'm hoping to do once
we have the KHR-GL* testing in place).

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6544>

ci: Enable KHR-GL30 CTS testing on freedreno a630.

Since we expose desktop GL, let's get at least a little bit of testing
coverage of it.

Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6526>

ci: Use the same VK-GL-CTS tree for GL/GLES as VK.

There's no need to have separate build scripts here, just choose what the
DEQP_TARGET is for the particular container being built.  This brings in a
tremendous number of GLES test fixes that haven't made it into a tagged
gles CTS release.

Closes: #2056
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6526>

ci: Bump vulkan CTS version to 1.2.3.2, and keep the GL CTS around.

The version bump gets us various testcase fixes, mostly to test
requirements).  While we're rebuilding the container, copy GL CTS stuff
from build-deqp-gl.sh -- we had already included the glcts binary in our
image, but we had unnecessary other binaries and were missing the mustpass
files (container size stays the same overall).  Also pull in all the GLES
mustpass lists, not just the main ones -- Rob wants them to increase our
coverage to match what Android CTS covers.

Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6526>

panfrost: Drop PIPE_CAP_MAX_COMBINED_HW_ATOMIC_COUNTER/BUFFERS

Just leave it at 0 and the frontend will lower for us.

Fixes dEQP-GLES31.functional.atomic_counter.get.1_counter_*

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6439>

panfrost: add support for atomics

Signed-off-by: Italo Nicola <italonicola@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6439>

panfrost: add atomic ops infrastructure

Signed-off-by: Italo Nicola <italonicola@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6439>

panfrost: add support for src[3] in LOAD_STORE ops

Signed-off-by: Italo Nicola <italonicola@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6439>

panfrost: introduce LDST_ATOMIC property

Signed-off-by: Italo Nicola <italonicola@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6439>

panfrost: add LDST_ADDRESS property to atomic ops

Atomic ops have to encode the address of the variable it's writing to.
This property is used to align the address to 64-bit boundaries.

Signed-off-by: Italo Nicola <italonicola@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6439>

panfrost: add atomic_cmpxchg opcode

Signed-off-by: Italo Nicola <italonicola@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6439>

panfrost: fix undefined value access on mir_set_intr_mask()

Signed-off-by: Italo Nicola <italonicola@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6439>

clover: Use 64-bit offsets for shader_in on 64-bit GPUs

This really shouldn't matter as inputs should have logical pointers.
However, nir_builder defaults to building derefs based on the pointer
size in the shader_info.  It's easier for now to just be consistent
everywhere.

Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6379>

nir/lower_explicit_io: Assert that compute address sizes match derefs

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6379>

spirv: Drop the constant_as_global as option

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6379>

clover/nir: Use nir_var_mem_constant for __constant memory

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6379>

llvmpipe: Add support for load_global_constant

Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6379>

nouveau/nir: Implement load_global_constant

Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6379>

intel/fs: Implement nir_intrinsic_load_global_constant

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6379>

spirv: Use nir_var_mem_constant for UniformConstant data in CL

For now, we leave the constant_as_global option intact and get rid of
the UBO path which no one upstream is using today.

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6379>

nir: Allow opt_large_constants to be run with constant_data_size > 0

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6379>

nir/lower_io: Add support for nir_var_mem_constant

This commit adds support for nir_var_mem_constant various places.  It
also adds a pass similar to nir_lower_vars_to_explicit_types except it
also scrapes out the constants and stuffs them into constant_data.

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6379>

nir/lower_io: Add a build_addr_for_var helper

The new version is more verbose but also more extensible.

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6379>

nir/lower_io: Use the variable mode for load_scratch_base_ptr checks

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6379>

nir: Add a load_global_constant intrinsic

This has the same semantics as load_global except the memory it reads is
known to be constant so load_global_constant intrinsics can be CSEd
rather than relying on more complex copy-propagation.

Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6379>

nir: Add a new nir_var_mem_constant variable mode

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6379>

nir/serialize: fix serialization of system values

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6545>

clover/spirv: fix vec3 alignment

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6433>

nvc0/cl: hande 64 bit pointers in nvc0_set_global_handle

clover gives us a uint32_t pointer into the kernel input bufffer, but also
for actual 64 bit pointers, so we can just use memcpy instead.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6433>

spirv: fix 64 bit atomic inc and dec

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jason Ekstrand <jason@jlekstran.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6433>

nvc0/ir: fix load propagation for sub 4 byte addressing

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6433>

clover/llvm: undefine __IMAGE_SUPPORT__ for devices without image support

libclang seems to define this on its own for SPIR targets, but the CTS
requires it to be not set if the device doesn't support images.

The SPIRV-LLVM-Translator also requires the spir triple to be set so we
can't really do anything else except to undefine.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jason Ekstrand <jason@jlekstran.net>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6433>

clover/nir: use offset for temp memory

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6433>

nv50/ir: fix cas lowering for 64 bit

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6433>

clover/nir: Lower function_temp to scratch.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6433>

egl/x11: simplify dri2_initialize_x11()

eglInitialize() already handles the "retry using the software path"
logic, there's no need to repeat it here.

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6353>

egl/wayland: simplify dri2_initialize_wayland()

eglInitialize() already handles the "retry using the software path"
logic, there's no need to repeat it here.

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6353>

egl/surfaceless: simplify dri2_initialize_surfaceless()

eglInitialize() already handles the "retry using the software path"
logic, there's no need to repeat it here.

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6353>

egl/android: simplify dri2_initialize_android()

eglInitialize() already handles the "retry using the software path"
logic, there's no need to repeat it here.

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6353>

intel/fs: Fix MOV_INDIRECT and BROADCAST of Q types on Gen11+

The immediate case is pretty uncommon to see but it can happen, in
theory.  BROADCAST is typically used to uniformize values and those are
usually 32-bit.  However, it does come up in some subgroup ops.

Fixes: 49c21802cbca "intel/compiler: Split has_64bit_types into float/int"
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6211>

nir: use enum operator helper for nir_variable_mode and nir_metadata

those are used quite a bit

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6520>

util: add helpers to define bitwise operators on enums for C++

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6520>

nir: use nir_var_all to get rid of casting

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6520>

nir: Improve the comment on num_inputs and friends

This doesn't fix the problem that no one knows what any of these mean
half the time but it at least makes them better documented to hopefully
make people's expectations more accurate.

Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6524>

nir: Rename num_shared to shared_size

This one is always a size in bytes.

Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6524>

spirv: Delete some dead workgroup variable handling code

This is dead since 5ed4e31c08dc0.

Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6524>

spirv: add some tests for volatile/available/visible

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6090>

spirv: Support big-endian strings

This should be all that's required for the SPIR-V parser to work in
big-endian systems.  SPIR-V requires that everything be in host
byte-order except for strings which are always little-endian.

Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6090>

spirv: implement SpvMemoryAccessVolatileMask

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6090>

spirv: implement Volatile image operand

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6090>

spirv: implement Volatile memory semantic

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6090>

spirv: implement MakePointerAvailable/MakePointerVisible for OpCopyMemory

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6090>

spirv: add vtn_emit_make_{visible,available}_barrier helpers

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6090>

spirv: make OpLoad/OpStore visibility/availablity barriers acquire/release

I think these are needed to order the visibility/availability operation
with the access.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6090>