The compiler only references `intel_device_info->subslice_masks` for
ray tracing workloads. Platforms which lack raytracing support can
share a cache even if they differ on this field.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28311>
The 64-bit type lowering for SEL in opt_algebraic had a pre-existing bug
where it only triggered when 64-bit float _and_ integer types were
unsupported. Meteorlake supports 64-bit float but not integer, so we
need to lower Q/UQ in that case still.
When I moved this to a later pass, opt_peephole_sel started generating
Q/UQ SEL instructions which were failing to be lowered.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/10867
Fixes: ea423aba1b ("intel/brw: Split out 64-bit lowering from algebraic optimizations")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28328>
This is similar to what RADV implements using the NIR_LOOP_PASS
helpers. I have not used those helpers for a couple of reasons:
1. They use the pointer to the optimization function, which doesn't
work if the same function is called multiple times in one invocation
of the loop (fixable)
2. After fixing them, due to Intel's use of sub-expressions, the amount
of code added to wrap the shared macro becomes more than simply
reimplementing them for the Intel compiler
On most workloads the results are a wash, but on compile heavy
workloads like Cyberpunk 2077 and Rise of the Tomb Raider, I saw
fossil-db runtimes fall by 1-2% on my ICL, with no changes to the
compiled shaders. Caio saw closer to 2.5% on TGL.
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27510>
The mlen tracking is in REG_SIZE units, but in Xe2 each GRF has
doubled the size. The optimization can only elide full GRFs, so
round down the amount of trailing zeros to ensure the optimization
will remove only full GRFs.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28279>
Folks, there's more than one accumulator. In general, when the
register file is ARF, the upper 4 bits of the register number specify
which ARF, and the lower 4 bits specify which one of that ARF. This
can be further partitioned by the subregister number.
This is already mostly handled correctly for flags register, but lots
of places wanted to check the register number for equality with
BRW_ARF_ACCUMULATOR. If acc1 is ever specified, that won't work.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28281>
Make the filename more descriptive and since the file is used by
multiple drivers, move it into appropriate util/ directory.
Cosmetics:
- use SPDX license tag
- add newline before main function
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27804>
There is no tiling in Xe KMD uAPI.
With this one more i915_drm.h include can be removed.
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28327>
This patch exposes support for the following three extensions:
* VK_GOOGLE_decorate_string
* VK_GOOGLE_hlsl_functionality1
* VK_GOOGLE_user_type
There's nothing for the driver to do; it's all handled in spirv_to_nir.
Signed-off-by: Valentine Burley <valentine.burley@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28155>
Multi-planar planes can be imported from VA-API or V4L2. In this case we
would currently report wrong values for `gbm_bo_get_offset()`.
This does not fix any know visible bug, as crocus hardware usually does
not support muliti-planar planes in the display engine - in which case
the wrong values would make Mutter glitch. Lets report correct values
regardless.
While on it, also use the helper function for counting planes.
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28225>
On macOS builds the code is tryng to use X11 and XCB code paths for WSI functions and structures,
i.e. `VkXcbSurfaceCreateInfoKHR` and `vkCreateXcbSurfaceKHR()`.
The MoltenVK implementation only supports `VkMacOSSurfaceCreateInfoMVK` and `vkCreateMacOSSurfaceMVK()`.
To get these working correctly Metal diusplay surfaces need to be setup, which xquarts does not expose.
Until this situation is resolved correctly the `VK_STRUCTURE_TYPE_XCB_SURFACE_CREATE_INFO_KHR` code paths has been
changed to return `VK_INCOMPLETE`. This allows them to run without asserting and terminating the runtime.
This allows for non-display applicaitons, like `glxinfo`, to run without terminating.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28259>
Vulkan does not support quads, removal of this type forces quads to be converted
in `primconvert_init_draw()`.
In cases where `screen->have_triangle_fans` excludes MESA_PRIM_TRIANGLE_FAN,
`u_index_generator()` is accessing a NULL entry in 'generate_quads'
for converting MESA_PRIM_TRIANGLE_FAN to MESA_PRIM_QUADS.
Which seems like an odd conversion to do (might be why it is missing).
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28259>
The MoltenVK (time of writing) only implements Vulkan 1.2 APIs and no `VK_KHR_format_feature_flags2` device extension.
This means VkFormatProperties3 is not supported, and is left blank when used in `populate_format_props()`.
If non-1.3 is detected then the same values are read from the `VkFormatProperties2` `props` location.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28259>
Vulkan behaviour has changed to require VK_INSTANCE_CREATE_ENUMERATE_PORTABILITY_BIT_KHR
and VK_KHR_portability_enumeration to be used on layered Vulkan implementations.
These are enbaled on macOS/Apple as MoltenVK is the only implmentation (time of writing)
and newer version require this.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28259>
From Vulkan 1.3.250 the MoltenVK include filenames changed.
Now include the newer 'MoltenVK/mvk_vulkan.h' instead of the older `MoltenVK/vk_mvk_moltenvk.h` headers.
From Vulkan 1.3.275 the MoltenVK include location changed.
The meson build options 'moltenvk-dir' now needs to point at the root of the Vulkan SDK install.
It will look for the presence of the older 'MoltenVK/include' loction and the newer 'macos/include' location.
Tested against Vulkan SDK versions: 1.3.250, 1.3.261, 1.3.268, 1.3.275.
Will not work compile with Vulkan 1.3.234 or earlier.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28259>
This will be useful to decode HW context in the next patch.
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27888>
xe_topic can't be inside of the for loop otherwise it will be set to
TOPIC_INVALID at every iteration.
TOPIC_INVALID was added after it was reviewed by Lionel because CI
complained that xe_topic may be not initialized, turns out leaving it
not initialized was causing the xe_topic value to keep the value set
in the previous interation makeing the parser to work by luck.
Fixes: 90e38bbb3b ("intel/tools/error_decode: Parse Xe KMD error dump file")
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27888>
In what is probably the most common case cross of compilation, x86_64
-> x86, it should be possible to build intel-clc for the host machine
and run it. Doing so simplifies the build by not needing to be able to
cross compile half of mesa, and should ease developer and distro strain
for building Intel drivers for x86.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28222>
in a sequence like:
* resize A
* clear
* resize B
* clear
* resize C
* clear
for a swapchain resource, the geometry for a given op after the resize
may desync for the op with which it was executed, but this is fine
since the underlying swapchain object will have to be re-created anyway
fixes#10827
cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28214>
Looks like it is the same as for ALU instructions: skpHp
This simplifies the special atomic handling a lot. Seen on blob
running a simple opencl shader on GC3000.
__kernel void kern(
__global int *a,
__global int *b,
__global int *result)
{
atomic_add(result, a[0]);
}
Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28272>
Currently 92% of our Vulkan CI tests hit "Not Supported" test
cases, which is ridiculously high. Add a bunch of skips, some
of which include very large categories of tests of features we
already know we don't support, so we stop wasting so much time
skipping tests.
With this, we can also increase the fraction of tests we execute
for vulkan significantly, while still keeping job run times
under control.
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28264>
A big chunk of the new flakes and timeouts are caused by enabling new
tests in f977e4d4f5 ("v3dv: Enable
EXT_swapchain_maintenance1").
I'm not quite sure what happened with
`dEQP-VK.wsi.wayland.swapchain.simulate_oom.*` but now at least half of
them are flaky on rpi4 (between Skip and Crash), so moving the whole
block to flakes. On rpi5 I haven't seen them flake yet so tentatively
removing them entirely, but there's a good chance the issue is common
and they'll have to be put into flakes on rpi5 at some point as well.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28316>
This will switch everyone to the isa specific functions.
Fixes the output of etnaviv's pre_instr_cb callback if
freedreno and etnaviv are build at the same time.
Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28176>
Create a static library that just contains isa_print(..). We
need to do this step to make lto happy.
Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28176>
Without this change the isa_decode_field(..) uses the wrong bitmask_t
type (freedreno: array with 2 elements, etnaviv: array with 4 elements)
and weird things happens.
Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28176>
Without this commit etnaviv_isa_disasm(..) will call into
find_field(..) and use the bitsets are the one from freedreno.
Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28176>
Any component that links against libetnaviv_decode should not need to
take care if the generated isa files exists.
Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28176>
Any component that links against libir3decode should not need to
take care if the generated isa files exists.
Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28176>
In the end we want to only the specific functions and remove
the generic ones from the public interface.
Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28176>
There are no users of these defines, structs and functions
outside of the generated isa.c file. I left the empty header
as it will be used in later commits.
Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28176>
and reorder the fields to pack them better
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Acked-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27953>
