this was correct for 64bit loads and manually converted 32bit loads (e.g., bindless),
but it was broken for the case where 64bit was not supported, as the offset wasn't
being correctly adjusted
break out the offset division to hopefully make this a little clearer
Fixes: 150d6ee97e ("zink: move all 64-32bit shader load rewriting to nir pass")
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>
all of ntv requires scalarized io since the offsets are now array indices
instead of byte offsets, so enforce scalarization here to avoid breaking
the universe
Fixes: 150d6ee97e ("zink: move all 64-32bit shader load rewriting to nir pass")
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>
zink can't use lower_io, so this all has to be done manually and in
excruciating depth and detail
fixes (tu):
KHR-Single-GL46.arrays_of_arrays_gl.InteractionFunctionCalls2
KHR-GL46.gpu_shader_fp64.fp64.named_uniform_blocks
KHR-GL46.gpu_shader_fp64.fp64.varyings
KHR-GL46.vertex_attrib_binding.advanced-bindingUpdate
KHR-Single-GL46.enhanced_layouts.varying_array_components
KHR-Single-GL46.enhanced_layouts.varying_array_locations
KHR-Single-GL46.enhanced_layouts.varying_components
KHR-Single-GL46.enhanced_layouts.varying_locations
KHR-Single-GL46.enhanced_layouts.xfb_explicit_location
dEQP-GLES3.functional.transform_feedback.basic_types.interleaved.lines.highp_mat3x4
dEQP-GLES3.functional.transform_feedback.basic_types.interleaved.lines.mediump_vec3
dEQP-GLES3.functional.transform_feedback.basic_types.interleaved.points.mediump_vec4
dEQP-GLES3.functional.transform_feedback.basic_types.separate.points.highp_vec3
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>
I left this semi-unfinished back when I discovered that I could blast
out xfb values inline with variable declarations, but this is not viable
for all scenarios, so it has to work and it has to be able to pass cts
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>
fs_visitor::assign_curb_setup() maps UNIFORM registers to HW regs,
and contains the following assert:
assert(inst->src[i].stride == 0);
emit_a64_oword_block_header's striding tricks run afoul of this
restriction, by producing stride 1 values on a 64-bit UNIFORM source.
Work around this by copying the UNIFORM value to a VGRF first.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
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/16938>
Instead of attempting to signal based on the memory object, use the new
DMA_BUF_IOCTL_EXPORT_SYNC_FILE to get a sync_file for the dma-buf and
use that to signal the semaphore or fence. Because this happens before
we transfer ownership back to the driver, the resulting sync_file should
only contain dma_fences from the compositor and/or display and shouldn't
be mixed up with the driver in any way. This gives us a real semaphore
and fence (as opposed to the dummy objects we've used int the past)
without over-synchronization.
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>
Instead of treating the blit submit specially in the buffer_blit_queue
case, treat the dummy submit as special. This lets us keep all the
handling of special-queue blits together. It also means that the
wsi_memory_signal_submit_info gets chained into the final submit which
is what we want if we're to rely on it for implicit sync. If we chain
it into the dummy submit, we'll implicit sync on all work previous to
the blit but not the blit. This won't work if X11 or a Wayland
compositor is depending on that to synchronize the linear copy.
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>
The WSI code is about to start querying for available semaphore handle
types via GetPhysicalDeviceExternalSemaphoreProperties in wsi_init().
For drivers that use vk_sync, supported_sync_types needs to be
initialized before GetPhysicalDeviceExternalSemaphoreProperties is
called. Really, wsi_init() should be the very last step of physical
device setup.
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>
Also, stop setting wsi_device::signal_semaphore/fence_with_memory
because those cause the WSI code to call the function we just dropped.
Since the core WSI code is now setting dummy syncs by default, we don't
need any of this anymore.
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>
With the old value, anv didn't think that the hardware supported 48-bit
addresses, and hit this assert:
assert(device->supports_48bit_addresses == !device->use_relocations);
The new value of 1ull << 48 is the one reported on my Icelake machine.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16933>
Kristian H. Kristensen