panfrost: decode: Rework the DUMP_{CL,ADDR}() macros

We don't need to pass the indentation level explicitly if we use
pandecode_log() instead of fprintf(). This helps keeping a consistent
indentation across descriptors.

While at it, let's rework those macros to allow creating titles from
a format string.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6797>

panfrost: decode: Make the indentation consistent with auto-generated print helpers

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6797>

panfrost: gen_pack: Allow enum/define values expressed in hexadecimal

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6797>

panfrost: gen_pack: Add a log2 modifier

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Suggested-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6797>

panfrost: gen_pack: Add an align() modifier

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6797>

panfrost: gen_pack: Allow empty structs

This is useful if we want to declare padding sections which can be
packed (filled with zeros) and unpacked (checked for non zero entries).

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6797>

panfrost: gen_pack: Add the aggregate concept

Panfrost descriptors are big and are usually built from a combination of
sub-descriptors. On top of that, layout of sub-descriptors might vary
depending on the architecture version. Since unions are not really an
option (too complex), here is a thin abstraction layer allowing us to
manipulate aggregates in their packed format. Each aggregate is formed
of one or more sections that are meant to be packed/unpacked/printed
separately. Section overlapping is allowed to facilitate handling of
descriptor variants.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6797>

panfrost: gen_pack: Move the group get_length() logic to its own method

So we can re-use it elsewhere.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6797>

panfrost: gen_pack: Add pan_{unpack,print}() helpers

And use them in pandecode.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6797>

panfrost: gen_pack: Fix __gen_unpack_uint()

The mask should be a 64-bit value and we should promote cl bytes to u64
before shifting them.

Fixes: 75cc5b8c2922 ("panfrost: Adopt gen_pack_header.py via v3d")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6797>

panfrost: gen_pack: Minor formatting improvement

Escape the \ in pan_pack() so the end result is actually a multi-line
macro.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6797>

Revert "panfrost: Drop implicit blend pooling"

This reverts commit dc7fbe114bac338e7075efdb325053ef41dad91c.

Fixes INSTR_INVALID_PC faults with the SuperTuxKart advanced rendering
pipeline, which occurred when blend shader BOs were allocated far apart.

Fixes: dc7fbe114ba ("panfrost: Drop implicit blend pooling")
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6785>

Revert "radv: emit {CB,DB}_RMI_L2_CACHE_CONTROL at framebuffer time"

This reverts commit d6bc0f26c918a67ebf85595d06628e8092e271d0.

These registers are now constant.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6726>

Revert "radv: set BIG_PAGE to improve performance on GFX10.3"

This reverts commit f4d861696dfb11dc2b6242a683a13238981f705f.

Turns out we cannot use BIG_PAGE with GTT and we can't tell
when a buffer is spilled to GTT.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6726>

radv,radeonsi: Disable compression on interop depth images

If we want to use HTILE correctly we need to communicate extra stuff
like clear colors. (Unlike DCC there is no HTILE FCE)

CC: mesa-stable
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6617>

ac/surface: Fix depth import on GFX6-GFX8.

Lets just do depth interop imports by convention between radv and
radeonsi for now. The only thing using this should be Vulkan interop
anyway.

CC: mesa-stable
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6617>

intel/fs/swsb: SCHEDULING_FENCE only emits SYNC_NOP

It's not really unordered in the sense that it can still stall on
ordered things and we don't need a SYNC_NOP for that because it is a
SYNC_NOP.  However, it also doesn't count when computing instruction
distances.

Fixes: 18e72ee2108 "intel/fs: Add FS_OPCODE_SCHEDULING_FENCE"
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6781>

nir: Add a pass for lowering CL-style image ops to texture ops

In CL 1.2, images are required to be either read-only or write-only.  We
can always translate the read-only image ops to texture ops.  In CL 2.0
(and an extension), the ability is added to have read-write images but
sampling (with a sampler) is only allowed on read-only images.  As long
as we only lower read-only images to texture ops, everything should stay
consistent.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6578>

spirv: Plumb access qualifiers through from image types

In SPIR-V, the access qualifiers for an image are provided on the image
type.  Assuming no one swaps the types around on us (I think that should
be illegal), this means we can reliably fetch the access qualifiers from
the type itself.  The ops which don't really have an easy-to-fetch type
are the atomics because they use OpImageTexelPointer.  However, those
are only allowed on read/write images and that's the default.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6578>

spirv: Access qualifiers are not a bitfield

They're an actual enum.  My bad.

Fixes: de36b5b805be "nir/vtn: Add support for kernel images to SPIRV-to-NIR"
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6578>

clover: Pull the stride from pipe_transfer for image maps

Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6578>

pan/mdg: Use nir_shader_instructions_pass for fdot2 lowering

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6718>

pan/mdg: Use nir_shader_instructions_pass for nir_lod_errata

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6718>

pan/mdg: Return a bool from midgard_nir_lod_errata

To match other NIR passes.

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6718>

panfrost: Set modifier_constant to true for exported resources

Not actually seen in the wild but could theoretically be a problem for
applications that explicitly import/export resources.

v2: Set it on exporting as well as importing (Daniel Stone)

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6717>

lima: fix vertex shader uniform buffer size

In some cases when switching shader programs, mesa does not switch the
currently set pipe_constant_buffer, which keeps pointing to the one
previously set.
If the two shader programs have a different number of uniforms, the size
of the constant buffer may be different and this needs to be considered
while generating the next draw command.
This patch fixes the uniform buffer creation in the lima vertex shader
command to avoid an out of bounds memcpy due to a previously set
pipe_constant_buffer.

Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6701>

lima: allocate new bo for stream draw

In stream draws, the resource bo might be in use in a previous draw.
Allocate a new one for the resource to avoid overwriting data in use.

Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Reviewed-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6503>

radeonsi: Move display dcc dirty tracking to framebuffer emission.

To improve performance.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6783>

radeonsi: Put retile map in separate buffers.

The retile maps are a software mechanism and hence very suceptible
to change. As such I'd like to avoid making it part of the cross
driver ABI.

Ideally we'd just use the cached tile info + a shader to avoid these
buffers altogether.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6783>

amd/common: Store non-displayable DCC pitch.

For DRM planes with modifiers.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6783>

radeonsi: fix user fence GPU address

User fence should have 4 QWORD memory space, I updated its CPU address
but forgot to update GPU address.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3513
Fixes: 3d5bed0e883 "radeonsi: fix user fence space when MCBP is enabled"
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6776>

glsl_type: Add packed to structure type comparison for hash map

Fixes: 659f333b3a4ff "glsl: add packed for struct types"
Reviewed-by: Marek Ol\9aák <marek.olsak@amd.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6767>

radv: fix vertex buffer null descriptors

Fixes: 0f1ead7b536 "radv: handle NULL vertex bindings"
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6773>

radv: fix null descriptor for dynamic buffers

Fixes: c1ef225d183 "radv: handle NULL descriptors"
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6772>

mesa: fix glUniform* when a struct contains a bindless sampler

Small example from #3271:

layout (bindless_sampler) uniform;
struct SamplerSparse {
  sampler2D tex;
  vec4 size;
  [...]
};
uniform SamplerSparse foo;

'foo' will be marked as bindless but we should only take the assign-as-GLuint64 path for 'tex'.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3271
Fixes: 990c8d15ac3 ("mesa: fix setting uniform variables for bindless samplers/images")
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6730>

etnaviv: add disk cache

Adds a shader disk-cache for shader variants.  Note that builds with
`-Dshader-cache=false` have no-op stubs with `disk_cache_create()` that
returns NULL.

This shader disk-cache gets used when using NIR only. Helps to save
about 1-2 minutes for a deqp run on gc2000.

Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6669>

etnaviv: shuffle some variant fields

Just to group together the parts that will get serialized when we have
shader disk-cache.

Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6669>

swr: fix _BitScanForward64 on unix

it must apply to 64 bits types, and use the ctzll intrinsic instead of ctz

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Krzysztof Raszkowski <krzysztof.raszkowski@intel.com>
Reviewed-by: Jan Zielinski <jan.zielinski@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6705>

swr: missing _BitScanForward64 on 32 bits win

the code does not compile on 32 bits systems
for mingw we can use gcc intrinsics like the unix side
for msvc a generic implementation is provided

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Krzysztof Raszkowski <krzysztof.raszkowski@intel.com>
Reviewed-by: Jan Zielinski <jan.zielinski@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6705>

glsl: take EXT_gpu_shader4 in to account when adding round

GL_EXT_gpu_shader4 adds truncate() and round() builtins.

Fixes: 12567de2be7 ("glsl: mark some builtins with correct glsl(es) version check")
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6731>

turnip: Release bo_mutex lock before potential error path.

Fix defect reported by Coverity Scan.

Missing unlock (LOCK)
missing_unlock: Returning without unlocking queue->device->bo_mutex.

Suggested-by: Jonathan Marek <jonathan@marek.ca>
Fixes: bea6290ca090 ("turnip: device global bo list")
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6768>

panfrost: Delete debug allocated syncobj.

Fix defect reported by Coverity Scan.

Logically dead code (DEADCODE)
dead_error_line: Execution cannot reach this statement: drmSyncobjDestroy(dev->fd, ...

Fixes: 64d6f56ad26f ("panfrost: Allocate syncobjs in panfrost_flush")
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6720>

radesonsi: Remove unsigned comparison to zero.

cbuf was changed to unsigned in commit 3fec2f67c3d9 ("radeonsi:
compact MRTs to save PS export memory space").

Fix defect reported by Coverity Scan.

Macro compares unsigned to 0 (NO_EFFECT)
unsigned_compare: This greater-than-or-equal-to-zero comparison of
an unsigned value is always true. cbuf >= 0U.

Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6739>

disk_cache: Fix filename leak on error path.

Remove filename ralloc comment. filename is allocated by asprintf.

Clean up disk_cache_get dead code left over from 367ac07efcc8
("disk_cache: move cache item loading code into
disk_cache_load_item() helper").

Fix defect reported by Coverity Scan.
Logically dead code (DEADCODE)
dead_error_line: Execution cannot reach this statement:
free(filename);

Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6738>

android: pan/bi: Use new packing

Fixes the following building error:

external/mesa/src/panfrost/bifrost/bi_pack.c:26:10: fatal error: 'bi_generated_pack.h' file not found
         ^~~~~~~~~~~~~~~~~~~~~
1 error generated.

Fixes: 2ff53879 ("pan/bi: Use new packing")
Signed-off-by: Mauro Rossi <issor.oruam@gmail.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6765>

android: pan/bi: Use new disassembler

Fixes the following building error:

external/mesa/src/panfrost/bifrost/bi_pack.c:409:24: error: implicit declaration of function 'pan_pack_fma_nop_i32' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                return pan_pack_fma_nop_i32(clause, NULL, regs);
                       ^
external/mesa/src/panfrost/bifrost/bi_pack.c:433:32: error: implicit declaration of function 'pan_pack_fma_fadd_f32' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                        return pan_pack_fma_fadd_f32(clause, bundle.fma, regs);
                               ^
external/mesa/src/panfrost/bifrost/bi_pack.c:433:32: note: did you mean 'pan_pack_fma_nop_i32'?
external/mesa/src/panfrost/bifrost/bi_pack.c:409:24: note: 'pan_pack_fma_nop_i32' declared here
                return pan_pack_fma_nop_i32(clause, NULL, regs);
                       ^
external/mesa/src/panfrost/bifrost/bi_pack.c:435:32: error: implicit declaration of function 'pan_pack_fma_fadd_v2f16' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                        return pan_pack_fma_fadd_v2f16(clause, bundle.fma, regs);
                               ^
external/mesa/src/panfrost/bifrost/bi_pack.c:435:32: note: did you mean 'pan_pack_fma_fadd_f32'?
external/mesa/src/panfrost/bifrost/bi_pack.c:433:32: note: 'pan_pack_fma_fadd_f32' declared here
                        return pan_pack_fma_fadd_f32(clause, bundle.fma, regs);
                               ^
external/mesa/src/panfrost/bifrost/bi_pack.c:442:32: error: implicit declaration of function 'pan_pack_fma_fcmp_f32' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                        return pan_pack_fma_fcmp_f32(clause, bundle.fma, regs);
                               ^
external/mesa/src/panfrost/bifrost/bi_pack.c:442:32: note: did you mean 'pan_pack_fma_fadd_f32'?
external/mesa/src/panfrost/bifrost/bi_pack.c:433:32: note: 'pan_pack_fma_fadd_f32' declared here
                        return pan_pack_fma_fadd_f32(clause, bundle.fma, regs);
                               ^
external/mesa/src/panfrost/bifrost/bi_pack.c:444:32: error: implicit declaration of function 'pan_pack_fma_fcmp_v2f16' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                        return pan_pack_fma_fcmp_v2f16(clause, bundle.fma, regs);
                               ^
external/mesa/src/panfrost/bifrost/bi_pack.c:444:32: note: did you mean 'pan_pack_fma_fadd_v2f16'?
external/mesa/src/panfrost/bifrost/bi_pack.c:435:32: note: 'pan_pack_fma_fadd_v2f16' declared here
                        return pan_pack_fma_fadd_v2f16(clause, bundle.fma, regs);
                               ^
external/mesa/src/panfrost/bifrost/bi_pack.c:449:41: error: implicit declaration of function 'pan_pack_fma_rshift_and_i32' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                                        pan_pack_fma_rshift_and_i32(clause, bundle.fma, regs) :
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:449:41: note: did you mean 'pan_pack_fma_fadd_f32'?
external/mesa/src/panfrost/bifrost/bi_pack.c:433:32: note: 'pan_pack_fma_fadd_f32' declared here
                        return pan_pack_fma_fadd_f32(clause, bundle.fma, regs);
                               ^
external/mesa/src/panfrost/bifrost/bi_pack.c:450:41: error: implicit declaration of function 'pan_pack_fma_lshift_and_i32' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                                        pan_pack_fma_lshift_and_i32(clause, bundle.fma, regs);
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:450:41: note: did you mean 'pan_pack_fma_rshift_and_i32'?
external/mesa/src/panfrost/bifrost/bi_pack.c:449:41: note: 'pan_pack_fma_rshift_and_i32' declared here
                                        pan_pack_fma_rshift_and_i32(clause, bundle.fma, regs) :
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:453:41: error: implicit declaration of function 'pan_pack_fma_rshift_and_v2i16' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                                        pan_pack_fma_rshift_and_v2i16(clause, bundle.fma, regs) :
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:453:41: note: did you mean 'pan_pack_fma_fadd_v2f16'?
external/mesa/src/panfrost/bifrost/bi_pack.c:435:32: note: 'pan_pack_fma_fadd_v2f16' declared here
                        return pan_pack_fma_fadd_v2f16(clause, bundle.fma, regs);
                               ^
external/mesa/src/panfrost/bifrost/bi_pack.c:454:41: error: implicit declaration of function 'pan_pack_fma_lshift_and_v2i16' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                                        pan_pack_fma_lshift_and_v2i16(clause, bundle.fma, regs);
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:454:41: note: did you mean 'pan_pack_fma_rshift_and_v2i16'?
external/mesa/src/panfrost/bifrost/bi_pack.c:453:41: note: 'pan_pack_fma_rshift_and_v2i16' declared here
                                        pan_pack_fma_rshift_and_v2i16(clause, bundle.fma, regs) :
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:458:41: error: implicit declaration of function 'pan_pack_fma_rshift_and_v4i8' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                                        pan_pack_fma_rshift_and_v4i8(clause, bundle.fma, regs) :
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:459:41: error: implicit declaration of function 'pan_pack_fma_lshift_and_v4i8' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                                        pan_pack_fma_lshift_and_v4i8(clause, bundle.fma, regs);
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:459:41: note: did you mean 'pan_pack_fma_rshift_and_v4i8'?
external/mesa/src/panfrost/bifrost/bi_pack.c:458:41: note: 'pan_pack_fma_rshift_and_v4i8' declared here
                                        pan_pack_fma_rshift_and_v4i8(clause, bundle.fma, regs) :
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:465:41: error: implicit declaration of function 'pan_pack_fma_rshift_or_i32' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                                        pan_pack_fma_rshift_or_i32(clause, bundle.fma, regs) :
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:465:41: note: did you mean 'pan_pack_fma_nop_i32'?
external/mesa/src/panfrost/bifrost/bi_pack.c:409:24: note: 'pan_pack_fma_nop_i32' declared here
                return pan_pack_fma_nop_i32(clause, NULL, regs);
                       ^
external/mesa/src/panfrost/bifrost/bi_pack.c:466:41: error: implicit declaration of function 'pan_pack_fma_lshift_or_i32' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                                        pan_pack_fma_lshift_or_i32(clause, bundle.fma, regs);
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:466:41: note: did you mean 'pan_pack_fma_rshift_or_i32'?
external/mesa/src/panfrost/bifrost/bi_pack.c:465:41: note: 'pan_pack_fma_rshift_or_i32' declared here
                                        pan_pack_fma_rshift_or_i32(clause, bundle.fma, regs) :
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:469:41: error: implicit declaration of function 'pan_pack_fma_rshift_or_v2i16' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                                        pan_pack_fma_rshift_or_v2i16(clause, bundle.fma, regs) :
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:470:41: error: implicit declaration of function 'pan_pack_fma_lshift_or_v2i16' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                                        pan_pack_fma_lshift_or_v2i16(clause, bundle.fma, regs);
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:470:41: note: did you mean 'pan_pack_fma_rshift_or_v2i16'?
external/mesa/src/panfrost/bifrost/bi_pack.c:469:41: note: 'pan_pack_fma_rshift_or_v2i16' declared here
                                        pan_pack_fma_rshift_or_v2i16(clause, bundle.fma, regs) :
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:474:41: error: implicit declaration of function 'pan_pack_fma_rshift_or_v4i8' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                                        pan_pack_fma_rshift_or_v4i8(clause, bundle.fma, regs) :
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:475:41: error: implicit declaration of function 'pan_pack_fma_lshift_or_v4i8' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                                        pan_pack_fma_lshift_or_v4i8(clause, bundle.fma, regs);
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:475:41: note: did you mean 'pan_pack_fma_rshift_or_v4i8'?
external/mesa/src/panfrost/bifrost/bi_pack.c:474:41: note: 'pan_pack_fma_rshift_or_v4i8' declared here
                                        pan_pack_fma_rshift_or_v4i8(clause, bundle.fma, regs) :
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:482:41: error: implicit declaration of function 'pan_pack_fma_rshift_xor_i32' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                                        pan_pack_fma_rshift_xor_i32(clause, bundle.fma, regs) :
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:483:41: error: implicit declaration of function 'pan_pack_fma_lshift_xor_i32' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
                                        pan_pack_fma_lshift_xor_i32(clause, bundle.fma, regs);
                                        ^
external/mesa/src/panfrost/bifrost/bi_pack.c:483:41: note: did you mean 'pan_pack_fma_rshift_xor_i32'?
external/mesa/src/panfrost/bifrost/bi_pack.c:482:41: note: 'pan_pack_fma_rshift_xor_i32' declared here
                                        pan_pack_fma_rshift_xor_i32(clause, bundle.fma, regs) :
                                        ^
fatal error: too many errors emitted, stopping now [-ferror-limit=]
20 errors generated.

Fixes: f8fc2105 ("pan/bi: Use new disassembler")
Signed-off-by: Mauro Rossi <issor.oruam@gmail.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6765>

glx: Allow depth-30 pbuffers to work without a depth-30 pixmap format

The backing pixmap needs to be big enough to hold depth-30 rendering,
but the server doesn't necessarily have a depth-30 pixmap format. Just
round up to the 32bpp format you'd be using anyway.

Fixes: mesa/mesa#3527
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6764>

anv: Fix the target_bo assertion in anv_reloc_list_add

Fixes: b75a92e2010e "anv: assert that the target bo is valid..."
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3528
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6759>

radv: initialize with expanded cmask if the destination layout needs it

If radv_layout_can_fast_clear() is false, 028C70_COMPRESSION is unset when
the image is rendered to and CMASK isn't updated. This appears to cause
FMASK to be ignored and the 0th sample to always be used.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3449
Fixes: 7b21ce401f7e81deca5ab6a4353b1369d059bcee
   ('radv: disable FMASK compression when drawing with GENERAL layout')

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6745>

nir: remove ubo_r600 instrinsic since ubo_vec4 is used now

As suggested by Eric.

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
eviewed-by: Eric Anholt <eric@anholt.net>

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6743>

r600/sfn: Use load_ubo_vec4 lowering pass

This replaces the lowering pass to align UBO loads at 16 byte
boundaries.

v2: use nir functions to query constants in ubo_vec4 (Eric)

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6743>

r600: enable lowering uniforms to UBO

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6743>

aco: fix incorrect assertion in emit_vop3a_instruction()

Fixes some float controls tests on Polaris10.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Fixes: 0b6448bbe791474a152d53bfaf750408807ac166
   ('aco/isel: refactor emit_vop3a_instruction() to handle 2 operand instructions')

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6744>

turnip: Add support for a615.

Verified RB_CCU_CNTL, 9805, and A0F8 values from blob traces.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6741>

util: Makefile.sources: Add disk_cache_os.{c,h}

The functions in disk_cache_os are referenced by disk_cache but these
files are not included in builds based on Makefile.sources such as
Android, resulting in linker errors.

Fixes: 4339ecde35c ("disk_cache: move cache dir generation into OS specific helper file")
Signed-off-by: Marijn Suijten <marijns95@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6702>

nir/lower_io: don't reduce range if parent length is zero

When handling arrays, range is increased based on the array size minus
one. But if such is zero, it has the effect of reducing the
range. Handle that case by returning the unknown range value.

v2:
  * Add missing braces.
  * Return unknown range in this case, instead of keeping the initial
    range.
v3: Simplify code, using existing "fail" label. (Jason)

Fixes the following using v3dv:
  dEQP-VK.graphicsfuzz.cov-simplify-clamp-max-itself

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6737>

pan/bi: Drop *FMIN reference

Even on G72, it's unsupported.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Remove unused packing data structures

Replaced by metaprogrammed variants.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Remove unused prints

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Use new packing

...and remove the old manual code.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Move packing helpers to dedicated file

We'll need to access them from the autogenerated section.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Use src1/dest_invert instead of src_invert[]

This maps more closely to the hardware, which makes for easier packing.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Use 8-bit shifts

Logically, it doesn't matter, but we want the IR to accurately reflect
the hardware behaviour.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Pass blend descriptor explicitly in IR

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Track compute_lod in IR

We'll need to differentiate tex and txl.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Add format field to IR

To make register_format packing explicit, and possibly in the future
support the auto mode.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Introduce segments into the IR

Needed to select between global, UBO, TLS, and WLS addressing modes,
required to implement loads/stores correctly.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Add dummy carry/borrow argument for iadd/isub

On FMA, a carry/borrow is required for iaddc/isubb (whereas the ADD
counterparts don't support carrying/borrowing). The trick is to model
this with an extra dummy (ZERO) argument which is free to encode on FMA,
and in the scheduler, "demote" to the non-carried versions if we want to
schedule to ADD.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Use canonical syntax for special constants

Adds some missing constants relevant to compute shaders, etc.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Use canonical syntax for registers/uniforms/imms

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Use new disassembler

We still use the clause/register decoding, but we now use the
metaprogrammed instruction decoding for the bulk of the operation.

We add a meson rule to call out to the Python generator script during
the build process.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Export dump_src

Needed in generated disassembler.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Add bi_disasm_dest_* helpers

Used to print the actual register/temporary for an instruction
destination given the port arrangement.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Add disassembly prototypes

Fix a header guard too.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Add disassembler generator

Given a parsed instruction set definition, this script generates
instruction disassembly routines responsible for decoding instruction
words and pretty-printing. Decoding is somewhat complex as with the
previous disassembler but can be automated.

Disssembly is complicated by indirect specifications of instruction
modifiers. These specifiers are given as logic expressions in the XML,
which optimizes for straightforwaard packing but makes disassembly
awkward. Instead of attempting to invert the logic directly, we generate
lookup tables of `modifiers -> encoding` maps which we may invert
directly to produce a lookup table for the `encoding -> modifiers` map
needed for disassembly.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Add packing generator

From the ISA definition, we can generate a function for each instruction
that looks at the bi_instruction in the intermediate representation and
emits a 20- or 23-bit word (for ADD/FMA respectively) containing that
instruction with all of its modifiers.

These will approximate the old packing routines, although the mapping of
bi_instruction to machine instructions will be hardcoded (at least for
now).

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Add ISA parser

This Python script parses the ISA.xml file to produce a normalized
in-memory representation suitable for the disassembly and packing
scripts to consume. In particular, it papers over details about
duplicate encodings and default fields.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Add XML describing the instruction set

Throughout this series, this XML file will serve as architectural ground
truth. It contains every instruction in the instruction set with all
programmable modifiers, as well as logic for computing derived values
(indirectly specified modifiers) and swapping operands as needed by
numerous encodings. It also allows for multiple encodings per
instruction differentiated by exact bits (a generalization of opcodes),
with different derived fields in each encoding, and logic tests to
select between the encodings at pack time.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

pan/bi: Lower flrp16

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Fixes: 3c8934a644b ("nir/algebraic: add flrp patterns for 16 and 64 bits")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>

anv: assert that the target bo is valid when adding a reloc list

this catches some undefined behavior like e.g., using a stale descriptorset
that references deleted bos, which I would absolutely never do

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6747>

anv: improve error message when failing to open device path

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6747>

docs: update calendar and link releases notes for 20.1.8

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6748>

docs: add release notes for 20.1.8

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6748>

gallium+mesa/st: lower uniforms based on compiler flag instead of packed uniforms cap

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6316>

llvmpipe: set lower_uniform_to_ubo compiler flag

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6316>

intel/compiler: Set lower_uniform_to_ubo compiler flag

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6316>

freedreno/ir3: set lower_uniforms_to_ubo compiler flag

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6316>

radeonsi: set compiler flag lower_uniforms_to_ubo

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6316>

nir: Add option lower_uniforms_to_ubo

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6316>

gallium/aux: remove unused u_blit.[ch]

There's no users of this code, so we might as well get rid of it to
avoid confusion with u_blitter.[ch].

Acked-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6651>

vc4: remove unused header

Acked-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6651>

v3d: remove unused header

Acked-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6651>

r600/sfn: Fix comparison with different signedness

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6706>

r600/sfn: more fixing of vec4 fetching

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6706>

r600/sfn: Fix source swizzle for gradient queries

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6706>

r600/sfn: Fix bitfield ops and 2x16 split_y

Don't reuse the dest register to make instruction dependency tracking
easier.

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6706>

r600/sfn: Fix split_alu_modifiers

Don't reuse the dest register so we can better track instruction
dependencies later.

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6706>

r600/sfn: Lower *sign opcodes in nir

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6706>

r600/sfn: avoid some copies

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6706>

r600/sfn: Fix ordering of tex param moves

Both moves should happen in the same instuction group, otherwise the
lod/bias value will be overwritten by the shadow compare value.

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6706>