This allows us to pass in some more details, like handles to the
loader-library to avoid a hard dependency.
Reviewed-by: Hoe Hao Cheng <haochengho12907@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11550>
This is going to make it possible to avoid linking directly to the
vulkan-loader.
Reviewed-by: Hoe Hao Cheng <haochengho12907@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11550>
This matters in the next patch, where we call vk-functions through the
dispatch-table instead.
Reviewed-by: Hoe Hao Cheng <haochengho12907@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11550>
This is just prep-work to reduce the size of the final commit; this
allows us to store the extension-getter function in the screen object.
Reviewed-by: Hoe Hao Cheng <haochengho12907@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11550>
Besides transparent border color, the hardware has special support for
white and black borders when clamping to border is enabled, which should
be more efficient.
v2:
- Rename enumerations (Iago)
- Add comment to border_color_variant field (Iago)
- Refactor the sampler variant code selection (Iago)
v3:
- Add comment for case of not clamping to border (Iago).
- Generate only required sampler states (Iago).
v4:
- Use one array element for standard border color set (Iago).
v5:
- Fix variant setting for transparent black borders (Iago).
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16422>
Previously this check was skipped for pools with
VK_DESCRIPTOR_POOL_CREATE_FREE_DESCRIPTOR_SET_BIT unset, but after
96a240e1 we need to check this otherwise we risk overflowing
radv_descriptor_pool::entries into the host memory base
This fixes a crash to desktop when launching Dota 2, which overallocates
descriptor sets and expects an error to allocate another descriptor pool
Fixes: 96a240e176 ("radv: fix memory leak of descriptor set layout")
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16490>
In a later merge request to remove the old GLSL IR loop unrolling
code we will add two additional compiler options stucts here. This
commit helps to avoid duplication getting out of hand.
Reviewed-by: Neha Bhende <bhenden@vmware.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16489>
Instead of storing them in a linked list, put them in an array
in si_shader_selector. The keys are also stored separatly, to
avoid pointer chasing when searching a variant in si_shader_select_with_key.
This main point here is to simplify the code by storing everything
in the selector instead of splitting the list storage between the selector
and the shaders; this shouldn't affect performance in a meaningful way.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16273>
Instead of spawning all the threads on startup.
This speeds up short lived programs (eg: piglit runs duration is reduced by ±25%),
avoid wasting resources and still make use of multi-threaded capabilities.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16273>
The original code waiting for the queue to be full before adding more
threads. This makes the thread count grow slowly, especially if the
queue also uses UTIL_QUEUE_INIT_RESIZE_IF_FULL.
This commit changes this behavior: now a new thread is spawned if we're
adding a job to a non-empty queue because this means that the existing
threads fail to process jobs faster than they're queued.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16273>
If the mesh shaders exports the viewport index or the layer, the value
can't be NULL, and it should be implicitly zero.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16438>
From the Vulkan spec:
"VUID-VkGraphicsPipelineCreateInfo-PrimitiveId-06264
If the pipeline is being created with pre-rasterization shader
state, it includes a mesh shader and the fragment shader code
reads from an input variable that is decorated with PrimitiveId,
then the mesh shader code must write to a matching output variable,
decorated with PrimitiveId, in all execution paths"
So, if PS uses PrimitiveID, MS must export it (like GS).
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16438>
We currently default to 4h of execution time, but this is unnecessary
on all our platforms. Tighten it to 45 minutes per run, and overall of
1.5h to allow for restarts.
Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Acked-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16472>
I am not super happy about doing this, as this means we will
potentially be introducing regressions because CI would have papered
over a rare race condition... But the situation is that this is already
happening, and improving the reliability will likely come from tracking
the occurence rate of hangs in an external system.
Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Acked-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16472>
This allows to support at least one binary format in
GL_ARB_get_program_binary extension.
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16433>
This merges them into si_llvm_load_intrinsic and reuses load_tess_varyings.
RADV only implemented 1 callback.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16467>
because LLVM 15 doesn't insert s_waitcnt before s_barrier.
Now, control_barrier shouldn't add any waits because memory_barrier is
responsible for it, so remove the wait.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16304>
This fixes 7 loop piglit tests when loop unrolling is disabled.
The problem is that we were trying to be smart with breaks and
tried to save one predicate instruction for endif in some cases.
This worked for simple loops but brought problems for more complex
shaders, instead just switch to standard VE_PRED_SNEQ_PUSH
ME_PRED_SET_POP combo everywhere.
Shader-db results on RV530 show three hurt glmark tests, however
I believe the simplification should be worth it.
total instructions in shared programs: 123715 -> 123718 (<.01%)
instructions in affected programs: 54 -> 57 (5.56%)
total predicate in shared programs: 118 -> 121 (2.54%)
predicate in affected programs: 6 -> 9 (50.00%)
total temps in shared programs: 17304 -> 17307 (0.02%)
temps in affected programs: 12 -> 15 (25.00%)
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6468
Signed-off-by: Pavel Ondračka <pavel.ondracka@gmail.com>
Reviewed-by: Filip Gawin <filip.gawin@zoho.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16470>
One of the fragment shaders passes large values to sin() function that
causes some blinking artifacts on Intel. This workaround limits the input
range for sin() to [-2*pi : 2*pi] to have better precision.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6452
Signed-off-by: Vadym Shovkoplias <vadym.shovkoplias@globallogic.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16388>
With this option enabled range of input values for fsin and fcos is
limited to [-2*pi : 2*pi] by calculating the reminder after 2*pi modulo
division. This helps to improve calculation precision for large input
arguments on Intel.
-v2: Add limit_trig_input_range option to prog_key to update shader
cache (Lionel)
Signed-off-by: Vadym Shovkoplias <vadym.shovkoplias@globallogic.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16388>
Currently, the LAVA job submitter fetches the job results from the LAVA
XMLRPC call, but that is not necessary, as the job result is easily
found in the logs. E.g. the bare-metal and poe jobs uses that log to set
the final job status of their runs.
Another reason for the change is that the LAVA signals are not reliable
in some devices with one serial port, causing some troubles in a618
recently. So, if one signal fails to be sent/received, the job will
ultimately fail even when the hwci script has been successful.
Fixes: #6435
Signed-off-by: Guilherme Gallo <guilherme.gallo@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16425>
this existed in order to reset query pools in advance so they
would never overflow. now the pools are reset every time the query
is started, so this behavior is no longer necessary
fixes#6475
Fixes: 57dd05616f ("zink/query: rewrite the query handling code to pass validation.")
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16484>
Fix clip distance shadow copy writemask as the shadow copy
includes clip distances to all clip planes whether enabled or not.
Fixes spec@glsl-1.30@execution@interpolation@interpolation-*distance
Reviewed-by: Neha Bhende <bhenden@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16481>
Previously it would fail, and then we'd fall back to the transfer path
for things like readpix. But it would spam logcat w/ bo_mmap fail
messages. Since gralloc allocated buffers for GPU usage are allocate
without _USE_MAPPABLE, let's just assume we can't map imported bo's.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16477>
