In the SVGA create shader functions, the original shader IR could have been
deleted after pipe_shader_state_to_tgsi_tokens() if it is to be converted
from NIR to TGSI. So to avoid accessing the deleted NIR IR,
set the shader state type to TGSI before passing the shader
state to the draw function.
Reviewed-by: Neha Bhende <bhenden@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16365>
use multiple contexts and submit in a round robin scheme to make
use of all the available jpeg engines simultaneously. During mjpeg
decode context need not be same across frames as they are discrete
jpeg images.
V2: number of ctx to be equal to number of engines and fix indent (Leo)
V3: decide ctx count in create_decoder, don't add a video param (Boyuan)
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/16355>
As a reminder primitive_restart should always be checked before any access to restart_index.
It seems that restart_index is only initialized when primitive_restart is set to a non-zero
value. This patch is equivalent to the previous code but written in a way that the compiler
will test primitive_restart first before trying to read restart_index.
With commit ad864a7c15:
Conditional jump or move depends on uninitialised value(s)
at 0xD33F1EC: panfrost_is_implicit_prim_restart (pan_cmdstream.c:2907)
by 0xD33F1EC: panfrost_emit_primitive (pan_cmdstream.c:3073)
by 0xD33F1EC: panfrost_draw_emit_tiler (pan_cmdstream.c:3440)
by 0xD33F1EC: panfrost_direct_draw (pan_cmdstream.c:3595)
by 0xD340467: panfrost_draw_vbo (pan_cmdstream.c:3889)
by 0xD219119: u_vbuf_draw_vbo (u_vbuf.c:1498)
by 0xD1C81F9: cso_multi_draw (cso_context.c:1644)
by 0xCFBA19B: _mesa_draw_arrays.part.11 (draw.c:1324)
by 0xCFBADA1: _mesa_draw_arrays (draw.c:1295)
by 0xCFBADA1: _mesa_DrawArrays (draw.c:1533)
by 0xD32EB: gl_vao_draw_data (in /usr/local/bin/mpv)
Uninitialised value was created by a stack allocation
at 0xCFBA14E: _mesa_draw_arrays.part.11 (draw.c:1289)
With mesa-22.1.0-rc4:
Conditional jump or move depends on uninitialised value(s)
at 0xD36369C: panfrost_is_implicit_prim_restart (pan_cmdstream.c:2895)
by 0xD36369C: panfrost_draw_emit_tiler (pan_cmdstream.c:3023)
by 0xD36369C: panfrost_direct_draw (pan_cmdstream.c:3215)
by 0xD3649BF: panfrost_draw_vbo (pan_cmdstream.c:3494)
by 0xD23DE7D: u_vbuf_draw_vbo (u_vbuf.c:1498)
by 0xD1ECBD1: cso_multi_draw (cso_context.c:1644)
by 0xCFD60FF: _mesa_draw_arrays.part.11 (draw.c:1324)
by 0xCFD6D11: _mesa_draw_arrays (draw.c:1295)
by 0xCFD6D11: _mesa_DrawArrays (draw.c:1533)
by 0xD32EB: gl_vao_draw_data (in /usr/local/bin/mpv)
Uninitialised value was created by a stack allocation
at 0xCFD60B2: _mesa_draw_arrays.part.11 (draw.c:1289)
Signed-off-by: Patrick Lerda <patrick9876@free.fr>
Reviewed-by: Alyssa Rosenzweig alyssa@collabora.com
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16389>
Due to an upstream bug, macOS can't link empty static libraries.
There is open Meson bug about this use case [1], though arguably the issue
is macOS's implementation of ar. Of course, the functionality is mostly
useless.
The removal of GLSL-to-TGSI trivialized a static library, causing
linking to fail. This commit garbage collects the useless library.
This fixes the build on macOS:
FAILED: src/mesa/state_tracker/tests/libmesa_st_test_common.a
rm -f src/mesa/state_tracker/tests/libmesa_st_test_common.a && ar csr src/mesa/state_tracker/tests/libmesa_st_test_common.a
ar: no archive members specified
usage: ar -d [-TLsv] archive file ...
ar -m [-TLsv] archive file ...
ar -m [-abiTLsv] position archive file ...
ar -p [-TLsv] archive [file ...]
ar -q [-cTLsv] archive file ...
ar -r [-cuTLsv] archive file ...
ar -r [-abciuTLsv] position archive file ...
ar -t [-TLsv] archive [file ...]
ar -x [-ouTLsv] archive [file ...]
[1] https://github.com/mesonbuild/meson/issues/3735
Fixes: 214c774ba6 ("mesa/st: Remove st_glsl_to_tgsi.")
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Emma Anholt <emma@anholt.net>
Tested-by: Vinson Lee <vlee@freedesktop.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16385>
Typically we free them when we upload the QPU code from the variant
to the assembly BO in the pipeline, however, if there is an error
during pipeline compilation that may not happen and we would leak
the QPU code from the variants.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16370>
We can output the final NIR form (which we store in the pipeline
stage) and the final QPU (which we can retrive from the assembly BO).
We should be careful not to fetch the shaders from the cache when
VK_PIPELINE_CREATE_CAPTURE_INTERNAL_REPRESENTATIONS_BIT_KHR is present,
since we don't store NIR shader in the pipeline shader data that is
cached, so a cache hit would leave us without the NIR shader. The spec
already contemplates this scenario:
"Enabling this flag must not affect the final compiled pipeline but
may disable pipeline caching or otherwise affect pipeline creation
time."
We also prevent disposing of the pipeline stages the variants when this
flag is requested to ensure this information is available later when
calling vkGetPipelineExecutableInternalRepresentationsKHR.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16370>
This is actually required by Vulkan 1.2 and to expose the extension,
so let's conform to this requirement, we don't really care since
image layouts are not relevant to our current implementation.
Fixes: 1442d77bc5 ('v3dv: trivially implement VK_KHR_separate_depth_stencil_layouts')
Fixes: dEQP-VK.info.device_mandatory_features
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16398>
When moving code into the main block or loop blocks, put the code into
its own :
if(true) { ... }
block so that we avoid break/continue/return issues.
v2: Also take care of the main block with return instructions
v3: Make deletion more obvious with dummy if blocks (Jason)
v4: Fixup assert for loops (Lionel)
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 8dfb240b1f ("nir: Add raytracing shader call lowering pass.")
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16036>
LLVM 15 ripped out the legacy coroutine passes. This means moving
to the new pass manager is the best option to move forward and is
long overdue.
I've tried to recreate the same set of passes in the new pass mgr
as the old, but I expect some tweaking may be needed to confirm this.
Acked-by: Kai Wasserbäch <kai@dev.carbon-project.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16289>
