review.tizen.org Git - platform/upstream/mesa.git/log

nir: add nir_export_dual_src_blend_amd intrinsic

For GFX11 export dual source blend outputs when ACO.
ACO need a pseudo instruction to emit a block of
code which can't be done in nir currently.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22199>

ac/nir/ps: fix null export write mask miss set to 0xf

Fixes: c1821544562 ("ac/nir: add ac_nir_lower_ps")
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22199>

radv/rt: remove merged VkRayTracingShaderGroupCreateInfoKHR

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22686>

radv/rt: replace uses of pGroups with radv_ray_tracing_group

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22686>

radv: Hash pipeline libraries separately

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22686>

radv/rt: add shader stage indices to radv_ray_tracing_group

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22686>

radv/rt: rename radv_ray_tracing_module -> radv_ray_tracing_group

This name better reflects the purpose and content of this struct.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22686>

svga: set PIPE_CAP_VERTEX_ATTRIB_ELEMENT_ALIGNED_ONLY for VGPU10 device

Instead of forcing vertex buffer stride to be 4 byte aligned only,
DX10 actually allows the stride to be non 4-byte aligned but the
alignment of an element must be the nearest power of 2 greater or equal to the
width of the element's format, or 4, whichever is less. So the requirement is
better met with PIPE_CAP_VERTEX_ATTRIB_ELEMENT_ALIGNED_ONLY which if set to
TRUE, the sum of vertex element offset + vertex buffer offset + vertex buffer
stride must be aligned to the vertex attributes component size.
Note: PIPE_CAP_VERTEX_ATTRIB_ELEMENT_ALIGNED_ONLY cannot be set
with other alignment-requiring CAPs, so we have to return 0 for all the
other alignement CAPs.

This avoids some unnecessary software vertex translate fallback.

cc: mesa-stable

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22689>

intel/dev: report stepping for TGL systems

Workaround 14010672564 requires a check for the TGL B0 stepping.

Reviewed-by: Rohan Garg <rohan.garg@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22458>

intel/dev: update mesa_defs.json from defect database

These modifications represent:

* changes to defects made since Feb 16, 2023
* changes to automated processing of defect state

Reviewed-by: Rohan Garg <rohan.garg@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22458>

glx: add fail check for current context in another thread

The GLX spec for glXMakeCurrent (3.3):
"If ctx is current to some other thread, then glXMakeCurrent will generate
a BadAccess error"

The GLX spec for glXCopyContext (3.3):
"If the destination context is current for some thread then a BadAccess
error is generated"

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7961
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Illia Polishchuk <illia.a.polishchuk@globallogic.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22546>

radv/rt: properly destroy radv_ray_tracing_lib_pipeline on error

Also return the correct error code.

Fixes: 4dafb69d61820c4a9b71856e62797b51f13df91c ('radv/rt: defer library_pipeline allocation')
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22684>

ci/zink: Try to update TGL results for new MSAA behavior.

A few fixes, but mostly tons of new GPU hangs.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22691>

ci/crocus: Note a recent regression.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22691>

ci/lima: Skip ppgtt_memory_alignment that flaked a job with the oomkiller.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22691>

ci/panfrost: Drop tex3d-maxsize on g52.

Implicated in 3 job-level flakes where Xorg got killed yesterday.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22691>

ci: add a660 firmware into rootfs

Until we bump to Debian 12 (bookworm).

Acked-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22604>

ci/lava: add support for HDK 888 firmware

Acked-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22604>

ci/lava: implement fastboot support

Based on work from Tomeu Vizoso.

Acked-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22604>

ci: add Adreno 660 on sm8350 chipset (HDK 888)

Acked-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22604>

aux/draw: fix memory leak related to ureg_get_tokens()

Indeed, the function nir_to_tgsi() returns an ureg_get_tokens() allocated
object which is assigned locally. The ureg_get_tokens() allocated object
should be freed.

For instance, this issue is triggered with a llvm enabled lima,
"piglit/bin/gl-1.0-rendermode-feedback -auto -fbo":
Direct leak of 512 byte(s) in 1 object(s) allocated from:
    #0 0x7faeaa4500 in __interceptor_realloc (/usr/lib64/libasan.so.6+0xa4500)
    #1 0x7fa4a88f1c in tokens_expand ../src/gallium/auxiliary/tgsi/tgsi_ureg.c:239
    #2 0x7fa4a88f1c in get_tokens ../src/gallium/auxiliary/tgsi/tgsi_ureg.c:262
    #3 0x7fa4a900f4 in copy_instructions ../src/gallium/auxiliary/tgsi/tgsi_ureg.c:2079
    #4 0x7fa4a900f4 in ureg_finalize ../src/gallium/auxiliary/tgsi/tgsi_ureg.c:2129
    #5 0x7fa4a91dfc in ureg_get_tokens ../src/gallium/auxiliary/tgsi/tgsi_ureg.c:2206
    #6 0x7fa4b20a2c in nir_to_tgsi_options ../src/gallium/auxiliary/nir/nir_to_tgsi.c:4011
    #7 0x7fa4a0c914 in draw_create_vertex_shader ../src/gallium/auxiliary/draw/draw_vs.c:77

Fixes: b5e782f5f431 ("aux/draw: use nir_to_tgsi for draw shader in llvm path")
Signed-off-by: Patrick Lerda <patrick9876@free.fr>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21924>

radv: Don't leak the RT prolog binary

Fixes: 063d0c90 ("radv: Combine all the parts together with a main loop for an RT pipeline.")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22535>

radv/rt: Plug some memory leaks during shader creation

nir_inline_function actually clones instructions instead of moving them.
Free the shaders explicitly after inserting them instead.

Fixes: 207ce6d658 ("radv: Add helper to inline shaders into the main shader.")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22535>

zink: print the type of shader when dumping

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22692>

nouveau/codegen: Check nir_dest_num_components

instead of reaching into a union and pulling out garbage when
the dest is a reg

Cc: mesa-stable
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8863
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22674>

freedreno/a6xx: Change a618 tile_align_h back to 32

Commit 60bc7c0e221 ("freedreno: Specify GMEM tile alignment per GPU")
changed the tile_align_h from 32 to 16 (which _should_ be the correct
value). But this is causing failure in android 9 skqp dstreadshuffle.
(But not, seemingly, with the android 11 version of skia+skqp, which
picks the same tile size. So this is likely papering something over.)

For now, to unblock things, revert back to the previous tile_align_h.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22683>

freedreno: Fix resource tracking vs rebind/invalidate

We can now no longer rely on certain dirty bits to re-trigger draw time
resource tracking. We need to use the new fd_dirty*_resource() APIs.

Fixes `org.skia.skqp.SkQPRunner#gles_recordopts` on android 9.

Fixes: 0a62a874fc5 ("freedreno: Re-work dirty-resource tracking")
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22683>

radv/rmv: Fix import memory

For some import memory, it is valid to specify zero size.

Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22532>

radv/rmv: Fix creating RT pipelines

Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22532>

ci/dzn: Run almost the full CTS

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22639>

dzn: Re-design custom buffer descriptors

Previously, custom buffer descriptors were owned by a descriptor set. Now,
custom buffer descriptors are owned by the buffer. Additionally, we respect
the app-provided sizes when they're smaller than the buffer size, even if
robustness is not enabled, so that size queries work correctly.

This new design fixes several issues:
* Descriptor set copies were broken when they involved custom descriptors,
  because the original descriptor set owned the lifetime of the custom
  descriptor, the new one was just borrowing it. If those lifetimes didn't
  line up, problems would arise.
* A single buffer with the same sub-view placed in multiplel descriptor sets
  would allocate multiple slots, when it only really needed one.
* Custom buffer descriptors now lower the base offset to 0 to allow merging
  multiple overlapping (ending at the same upper bound) descriptors. Since
  the shader is already doing an offset add, making it nonzero is free.
* Dynamic buffer descriptors were incorrect before. The size passed into the
  descriptor set is supposed to be the size from the *dynamic* offset, not the
  size from the static offset. By allocating/populating the descriptor when
  placed into the set, it prevented larger offsets from working correctly. This
  buffer-owned design prevents cmdbufs from having to own lifetime of custom
  descriptors.

Fixes dEQP-VK.ssbo.unsized_array_length.float_offset_explicit_size

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22639>

dzn: Handle depth bias for point fill mode emulation

Fixes dEQP-VK.draw.renderpass.depth_bias.depth_bias_triangle_list_point

This is not complete, there's no slope scale or clamp handling, but it
does handle static or dynamic (though dynamic is untested).

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22639>

dzn: Handle opaque BC1

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22639>

dzn: Use unrestricted copy alignments when available

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22639>

ci/windows: Update Agility SDK to 1.610.2

Otherwise non-normalized sampling coords are unintentionally disabled

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22639>

zink: fix store subsitution in `lower_pv_mode_gs_store`

Previously it was assumed that between the and the variable there was
only one deref.

To handle all cases a new function is introduced that recreates a chain
of derefs.

Fixes: 5a4083349f3 ("zink: add provoking vertex mode lowering")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22678>

nir/zink: use sysvals in `nir_create_passthrough_gs`

Previously the passthrough gs shader loaded some values with uniform
loads using sevaral hardcoded values.

This was not flexible for other drivers and started becoming too
unflexible for zink itself.

Use system values instead and use a lowering pass in zink.

Acked-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22667>

pvr: Add PVR_DW_TO_BYTES()

We use dwords (32 bit) quite a bit around the code base. Previously
we used '* 4', '<< 2', or '* sizeof(uint32_t)' to go from dwords to
bytes. The conversion isn't always clear when other operations
happen in the same line, which can leave one wondering where the
multiplication came from.
PVR_DW_TO_BYTES() should make the code more obvious as well as
making the conversion more consistent.

Signed-off-by: Karmjit Mahil <Karmjit.Mahil@imgtec.com>
Reviewed-by: Frank Binns <frank.binns@imgtec.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22658>

anv: implement state cache invalidate for Wa_16013063087

Cc: mesa-stable
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22651>

anv: cleanup bitmask construction for PIPELINE_SELECT

Cc: mesa-stable
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22651>

iris: implement state cache invalidate for Wa_16013063087

Cc: mesa-stable
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22651>

broadcom/compiler: return early for SFU op latency calculation

Since we are returning a fixed latency for these check for them
earlier and return early if they match.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22675>

broadcom/compiler: fix incorrect ALU checks

We had a bunch of cases where we would check ALU parameters without
first checking if the ALU op was valid.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22675>

radeonsi: implement fw based mcbp

Some chips support firmware based mcbp. If supported this means
radeonsi needs to allocate 3 buffers and pass them to the firmware.

From there, the firmware will handle mcbp and register shadowing
on its own so we don't need to insert LOAD packet in the preamble.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21986>

amd: determine info->has_fw_based_shadowing

The shadow_size value will be 0 if unsupported.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21986>

amd: update amdgpu_drm.h

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21986>

venus: fix the RHEL8 build by using syscall for gettid

src/virtio/vulkan/vn_common.c: In function ‘vn_ring_monitor_acquire’:
src/virtio/vulkan/vn_common.c:129:16: error: implicit declaration of function ‘gettid’; did you mean ‘getgid’? [-Werror=implicit-function-declaration]
129 | pid_t tid = gettid();

Reviewed-by: Yiwei Zhang <zzyiwei@chromium.org>
Reviewed-by: Ryan Neph <ryanneph@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22489>

ci/valve: Add a workaround for finding libdrm on navi21s.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22495>

ci/zink: Drop anv/lvp validation exceptions that should be fixed in the CTS.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22495>

ci/deqp: Update to 1.3.5.1 and pull in additional bugfixes from main.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22495>

ci: Move some timeout xfails to skips.

We generally don't want to wait around for a minute for things to fail.
Note that some of these were already in their skips.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22495>

rusticl/device: allow enablement of fp64 via RUSTICL_FEATURES

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22649>

rusticl/platform: add RUSTICL_FEATURES boilerplate

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22649>

rusticl/platform: extract env variable parsing from Platform::init

In our platform initialization code we might want to access the parsed env
variables already. So do this in separate steps.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22649>

rusticl/platform: make the initialization more explicit

It's not a lazy loaded type so doing the Once::call_once in every
Platform::get gives us a pointless atomic, which might be slow on some
platforms.

Every application has to call clGetPlatformIDs so we only need to do it
there.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22649>

rusticl/device: improve advertisement of fp64 support

Enabling fp64 support makes rarely sense, but in case we do claim it, we
should also tell if it's a pure software implementation.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22649>

gallium: Add pipe_shader_state_from_nir

and use it in nouveau

Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22537>

nv50,nvc0: Use ttn for tgsi shaders by default

Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22537>

nv50,nvc0: Stop advertising TGSI by default

Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22537>

nv50,nvc0: Use nir in nv50_blitter_make_fp

Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22537>

nvc0: Use nir in nvc0_blitter_make_vp

Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22537>

nvc0: Use nir in nvc0_program_init_tcp_empty

Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22537>

lavapipe: asst. clean-ups in lvp_execute.c

Move loop var decls into loops.
Use designated initializers.
Replace some questionable memcpy/memset() calls with assignments.
Line-wrap, whitespace fixes.

Signed-off-by: Brian Paul <brianp@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22467>

lavapipe: clean-ups in lvp_physical_device_get_format_properties()

Use new pscreen local var to shorten the code.
Line-wrap and align code for easier reading.
Consolidate |= assignments.

Signed-off-by: Brian Paul <brianp@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22467>

lavapipe: clean-ups in lvp_GetQueryPoolResults()

Simplify the code which puts query results into the destination
buffer:
* Use a uint64 or uint32 pointer instead of uint8 to write the results
to the buffer for less casting and simplifient pointer incrementing.
* Use MIN2() macro to be more concise.
And fix some indentation.

Signed-off-by: Brian Paul <brianp@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22467>

ci/mold: bump to 1.11.0

Fixes LTO issue.

Release notes: https://github.com/rui314/mold/releases/tag/v1.11.0

Acked-by: Eric Engestrom <eric@igalia.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22642>

llvmpipe: do late init for llvm builder

this avoids doing any llvm work during lavapipe enumeration which
might otherwise explode if lavapipe is not the actual driver that
gets used

cc: mesa-stable

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22600>

gallivm: break out native vector width calc for reuse

breaks dependency on lp_build_init()

cc: mesa-stable

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22600>

v3d/ci: enable glsl 1.30 and 1.40 piglit tests

Now that we are exposing OpenGL 3.1, let's enable the proper supported
GLSL versions.

Acked-by: Eric Engestrom <eric@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22660>

freedreno: Add more tracepoint fields

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22646>

freedreno/a6xx: Pass ring to __ONE_REG()

This happened to work before because the arg to OUT_REG() was also
called 'ring' (or there was a suitable local var in scope!!)

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22646>

freedreno: Add dirty state logging

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22646>

freedreno/perfetto: Add shader_id for compute stages

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22646>

mesa/nir: Add some perfetto traces

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22646>

freedreno/ir3: More perfetto tracing

Some useful trace points that I had laying around.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22646>

util/log: Add missing "const"

Fixes warning:

warning: ISO C++11 does not allow conversion from string literal to 'char *' [-Wwritable-strings]

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22646>

freedreno/registers: C++ struct casting

Using C style `(struct foo){ ..args..}` at least confuses eclipse, so
for C++ code use `{ ..args.. }` instead.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22646>

freedreno: Extra casting to make C++ happy

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22646>

util/log: Fix log messages over 1024 characters.

The first attempt at the sprintf would have consumed part of va, so if
we're going to recurse on overflow to try again in a new allocation then
we have to do our work on a copy.

This was a common failure mode for MESA_GLSL=source, where it would just print:

Mesa: info: GLSL source for fragment shader 1:
Mesa: info: (null)

Fixes: 7a18a1712a0a ("util/log: improve logger_file newline handling")
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22618>

anv: support fast color clears on vkCmdClearAttachments

As long as some conditions are met.

Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20175>

anv: expose some helper functions

v2: (Rohan Garg)
- Make set_fast_clear_state take an image and format instead of an iview

Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20175>

anv: factor out code for ccs_op and mcs_op

Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20175>

anv: make anv_can_fast_clear_color_view more generally available

Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20175>

anv: Remove dead parameters from copy_fast_clear_dwords

There's only one caller of this function and always passes false.

v2: (Nanley Chery)
- Also remove the aspect parameter

v3: (Nanley Chery)
- Rename the function so it's more clear in which direction it works

Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20175>

anv: Fast clear depth/stencil surface in vkCmdClearAttachments

Instead of doing a slow depth clear, we can do depth fast clear in
vkClearAttachments.

Sascha Willems occlusionquery demo shows more than 2% perf boost with
this series.

On Felix's Tigerlake with the GPU at fixed frequency, this patch
improves performance of RoTR by +0.5%.

v2: (Nanley Chery)
- Clear stencil surface along with depth.
- Check for multilayer resources.
- Lookout for state.attachments.
- Fallback on slow clear for BDW and CHV if conditional rendering
enabled.
- Keep flush in same function.

v3: (Nanley Chery)
- Return immediately after fast clearing.
- Remove unnecessary comment.

v4: (Nanley Chery)
- Add assertion for BLORP_BATCH_NO_EMIT_DEPTH_STENCIL.
- Remove unnecessary local variable.
- Add 3DSTATE_WM_HZ_OP comment.

v5: (Nanley Chery)
- Fix comments.
- Don't take fast depth clear path if BLORP_BATCH_PREDICATE_ENABLE set.
- Refactor code in can_hiz_clear_att.

Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20175>

anv: Move and make anv_can_hiz_clear_ds_view non-static

v2:
- Pass const image view param. (Nanley)

Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20175>

anv: Factor out code from anv_image_hiz_clear

Refactoring code from anv_image_hiz_clear which helps in future patches
to support fast depth clear in vkCmdClearAttachments.

Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20175>

blob: Don't valgrind assert for defined memory if we aren't writing.

The VK pipeline cache passes a NULL bytes with a nonzero size to a
NULL-data blob to set up the size of the blob.  In this case, we don't
actually execute the memcpy, so the non-existent "bytes" doesn't need to
have defined contents.  Avoids a valgrind warning:

==972858== Unaddressable byte(s) found during client check request
==972858==    at 0x147F4166: blob_write_bytes (blob.c:165)
==972858==    by 0x147F4166: blob_write_bytes (blob.c:158)
==972858==    by 0x14695FFF: vk_pipeline_cache_object_serialize (vk_pipeline_cache.c:240)
[...]
==972858==  Address 0x0 is not stack'd, malloc'd or (recently) free'd

Fixes: 591da9877900 ("vulkan: Add a common VkPipelineCache implementation")
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22617>

ci/zink: Enable the validation layer on the TGL GL46 run.

We recently had an issue where an anv merge failed due to a bug in zink,
which validation would have caught. Get some coverage by default on the
main branch.

This increases runtime from 9:30 to 12:10. I don't feel good about this,
but I've got https://gitlab.khronos.org/Tracker/vk-gl-cts/-/issues/4398 in
flight to try to knock the time back down.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22459>

ci: Move zink's validation layer setup to deqp-runner.sh.

I want the path to be always set in case someone wants some
ZINK_DEBUG=validation, rather than having to do it per test job.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22459>

ci/zink: Re-enable traces now that !20319 has landed.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22459>

ci: Add the Vulkan validation layer to amd64 rootfs builds.

We're going to want this for zink testing on actual HW drivers. I haven't
sorted out the arm64 build yet, so no tu coverage.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22459>

ci: Make a variable for the repeated rootfs directory name.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22459>

frontends/va: return matching drm format for yuyv pipe format

return the matching drm format for YUYV pipe format in pipe_format_to_drm_format(), else
vlVaExportSurfaceHandle() fails to export a surface handle for YUYV surface.

Signed-off-by: Sathishkumar S <sathishkumar.sundararaju@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22497>

zink: fix array copying in pv lowering

Fixes: 5a4083349f3 ("zink: add provoking vertex mode lowering")

Reviewed-by: Antonino Maniscalco <antonino.maniscalco@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22669>

nir/gs: fix array type copying for passthrough gs

same mechanics as in zink passes

Fixes: d0342e28b32 ("nir: Add helper to create passthrough GS shader")

Reviewed-by: Antonino Maniscalco <antonino.maniscalco@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22669>

dzn: Align-up heap sizes when allocating memory

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8895
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22666>

radv: do not overallocate the CS array during submissions

Preambles/postambles are no longer added to the CS array.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22653>

radv: Search for guilty contexts at radv_check_status

When a GPU hung happens, all contexts are notified. They will receive
INNOCENT_CONTEXT if they are not the context that triggered the reset,
or GUILTY_CONTEXT otherwise.

At radv_check_status(), we return on the first context that was notified
as [GUILTY, INNOCENT]_CONTEXT, without further checks. This can make an
app think that it's innocent if the guilty context is not the first one
on the list of hw_ctx to be checked.

Check every context for a guilty one before returning CONTEXT_INNOCENT.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: André Almeida <andrealmeid@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22648>