* optimize not(compare(a,b)), nir_opt_algebraic does this only if the
comparison result is used only once, but on a vector arch we still get
an advantage when doing this, because it reduces dependencies.
* optimize b2f32(compare(a,b)), this is r600 specific
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37205>
v2: Add more targets and a little test lib to add stubs for missing symbols
v3: Condition building tests on a recent enough meson (@leftmostcat)
v4: Set suite parameter for new tests (@leftmostcat)
Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35188>
Just to stay consistent. We might want to use them instead of declaring
them inside meson, but I'd rather prefer we have a consistent list across
all crates and that's easier done in meson.
Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35188>
No manual changes here, this is simply running
$ ninja -C build/ clang-format
Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com>
Reviewed-by: Pavel Ondračka <pavel.ondracka@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37224>
in some places this assert is used to hang check, but here we
can assume that the batch_id is always valid (i.e., not a hang) and
check it accordingly to avoid spurious crashes
cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37239>
Replace the basic debug message for unsupported blits with a comprehensive
dump of the blit parameters and make it a hard failure. This helps identify
exactly what blit operations are not supported and ensures they don't fail
silently in debug builds.
Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37034>
Wayland EGL applications often set SwapInterval 0 and then use surface
frame callbacks independently to pace their updates. These applications
don't actually want the display to tear, and we do not currently use
wayland tearing_control protocol on EGL to enable tearing behaviour.
Using a tearing presentation mode in Vulkan will lead to unexpected
behaviour for these applications.
Use MAILBOX on wayland instead.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34842>
In theory this is of no consequence, but letting the scheduler run
it's normal algorithm is better.
Fixes: a6b6ce84f7 ("r600/sfn: Prepare scheduler to handle WaitAck instructions")
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37232>
It seems emmitting more than four MEM_RAT instructions in
a row may lead to a GPU hang, so start scheduling free
instructions earlier.
The problem is triggered by
KHR-GL45.compute_shader.shared-max
Fixes: fc40002de7 ("r600/sfn: Simplify scheduling")
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37232>
We don't (yet) have the ability to say that more than one value is
acceptable, and the yaml module we use complains about duplicate keys,
so let's just comment them out as it doesn't care about missing keys and
doesn't remove comments.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37215>
The definitive linux kernel counterpart update is now merged.
The linux kernel commit dc5c742f41c0a1c2b14e4357c752851e015f3bcd
(v6.17-rc1) set the new r600 drm version requirement. This change
updates the drm version accordingly to 2.51.0.
Fixes: 9e1180b335 ("r600: implement ARB_indirect_parameters")
Signed-off-by: Patrick Lerda <patrick9876@free.fr>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37118>
The function resource_copy_region is expected to have a memcpy()
like behavior. Some formats, like r11g11b10_float, do not preserve
the original raw values. This was the problem. This change
updates r600_resource_copy_region() to use the generic copy path
for those formats.
This change takes into account how 8235d3aa19 "radeonsi: preserve
NaNs in draw-based resource_copy_region" proceeds.
This change fixes the remaining khr-gl and deqp-gles31 copy_image tests. This
change was tested on palm, barts and cayman. Here are the tests fixed:
khr-gl4[2-6]/texture_view/view_classes: fail pass
khr-gl4[3-6]/copy_image/functional_src_target_texture_2d_array_src_format_rgb9_e5_dst_target_texture_2d_array_dst_format_r11f_g11f_b10f: fail pass
khr-gl4[3-6]/copy_image/functional_src_target_texture_2d_array_src_format_rgb9_e5_dst_target_texture_3d_dst_format_r11f_g11f_b10f: fail pass
khr-gl4[3-6]/copy_image/functional_src_target_texture_2d_array_src_format_rgb9_e5_dst_target_texture_rectangle_dst_format_r11f_g11f_b10f: fail pass
khr-gl4[3-6]/copy_image/functional_src_target_texture_3d_src_format_rgb9_e5_dst_target_texture_2d_array_dst_format_r11f_g11f_b10f: fail pass
khr-gl4[3-6]/copy_image/functional_src_target_texture_3d_src_format_rgb9_e5_dst_target_texture_3d_dst_format_r11f_g11f_b10f: fail pass
khr-gl4[3-6]/copy_image/functional_src_target_texture_3d_src_format_rgb9_e5_dst_target_texture_rectangle_dst_format_r11f_g11f_b10f: fail pass
khr-gl4[3-6]/copy_image/functional_src_target_texture_rectangle_src_format_rgb9_e5_dst_target_texture_2d_array_dst_format_r11f_g11f_b10f: fail pass
khr-gl4[3-6]/copy_image/functional_src_target_texture_rectangle_src_format_rgb9_e5_dst_target_texture_3d_dst_format_r11f_g11f_b10f: fail pass
khr-gl4[3-6]/copy_image/functional_src_target_texture_rectangle_src_format_rgb9_e5_dst_target_texture_rectangle_dst_format_r11f_g11f_b10f: fail pass
deqp-gles31/functional/copy_image/non_compressed/viewclass_16_bits/.*: fail pass
deqp-gles31/functional/copy_image/non_compressed/viewclass_32_bits/.*: fail pass
Cc: mesa-stable
Signed-off-by: Patrick Lerda <patrick9876@free.fr>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37117>
It alwways comes in through the create flags now.
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36957>
it's otherwise possible for this to race and hit the draw before
precompile finishes without ever waiting on the fence
I guess this just worked coincidentally before?
cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37197>
