As described in the comment, enabling the round_to_nearest_even results
in the upper 2^-9 of the texel i being sampled at i+1. This appears to
be allowed by the spec, but triggers a CTS bug[1]. Changing this behavior
is not necessary (we could fix the CTS), but is desirable regardless
because of the precision improvement.
[1]: https://gitlab.khronos.org/Tracker/vk-gl-cts/-/issues/5547
Signed-off-by: Benjamin Lee <benjamin.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32985>
The documentation says that if we don't use force_delta, the LOD will be
-infinity for non-active lanes before bias and clamp are applied. This
is not what we want, so let's instead assume all threads are active, and
let helper-invocations do their job to compute correct values.
While this is only needed for the second iteration, let's just leave it
on for both for simplicity.
Fixes: e317136536 ("pan/va: Add support for nir_texop_lod")
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33069>
Update expectation files for the test
runs with kernel 6.13-rc4.
Signed-off-by: Vignesh Raman <vignesh.raman@collabora.com>
Reviewed-by: David Heidelberg <None>
Reviewed-by: Sergi Blanch Torné <None>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32788>
Move to 6.13-rc4 for all mesa-ci jobs except anv-jsl.
Signed-off-by: Vignesh Raman <vignesh.raman@collabora.com>
Reviewed-by: David Heidelberg <None>
Reviewed-by: Sergi Blanch Torné <None>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32788>
Applications tend to forget to describe subpass dependencies, especially
when it comes to write -> read dependencies on attachments. The
proprietary driver forces "others" invalidation as a workaround, and this
invalidation even became implicit (done as part of the RUN_FRAGMENT) on
v13+.
We will consider adding a dri-conf hook for this option in the future,
but for now, let's just keep it as an opt-in debug flag.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33056>
Input attachment reads are lowered to image reads and thus require
a flush of the read-only L1 caches.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33056>
The expansion of DUMP_CL is missing parenthesis, making the dumping of
descriptors incorrect.
Fixes: 3b69edf825 ("pan/genxml: Enforce explicit packed types on pan_[un]pack")
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33040>
Right now, we have a problem when we flush draws inside a render pass
and we don't have enough information to re-emit the framebuffer/tiler
descriptors.
Turns out the only situations where this happens is when an occlusion
query end happens, but we shouldn't really flush the draws in that case.
What we should do instead is record the OQ in our command buffer, so we
can signal OQ availability when the fragment job is done.
In order to solve that, we add an OQ chain to the command buffer to
track OQs ending inside the render pass. We then walk this chain at
fragment job emission time to signal the syncobjs attached to each
query.
This also simplifies the whole occlusion query synchronization model:
instead of waiting for each syncobj individually, we now wait on
the iterators to make sure all OQs have landed. Thanks to this new
synchronization, we can batch OQ reset/copy operations and make the
command stream a lot shorter when big query ranges are copied/reset.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Reviewed-by: Benjamin Lee <benjamin.lee@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32973>
We have wrappers distinguishing staging registers from sratch registers,
so let's use cs_sr_reg64() here.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Reviewed-by: Benjamin Lee <benjamin.lee@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32973>
The SYSTEM scope triggers CPU interrupts we don't really need, so let's
use the CSG scope to avoid those. Note that the scope doesn't encode
the visibility aspect, meaning changes to the sync object with a CSG
scope will still be instantly visible to the CPU, it's just that the
CPU needs to poll the value to detect a change, which is basically what
we're doing for syncobjs attached to events/queries, so we're good.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Reviewed-by: Benjamin Lee <benjamin.lee@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32973>
Let's prevent clang-format from adding the semi-colon on a new line when
we use cs_{continue,break}();
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Reviewed-by: Benjamin Lee <benjamin.lee@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32973>
We already do that in the other cs_emit(b, BRANCH, I), so let's fix this
path too.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Reviewed-by: Benjamin Lee <benjamin.lee@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32973>
The increment was wrong, which ended up generating a lot more stores
than we need.
Fixes: bf05842a8d ("pan/cs: Add an event-based tracing mechanism")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Reviewed-by: Benjamin Lee <benjamin.lee@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32973>
The spec says
vkCmdCopyQueryPoolResults is considered to be a transfer operation,
and its writes to buffer memory must be synchronized using
VK_PIPELINE_STAGE_TRANSFER_BIT and VK_ACCESS_TRANSFER_WRITE_BIT before
using the results.
While STORE_MULTIPLE is not exactly VK_PIPELINE_STAGE_TRANSFER_BIT /
VK_ACCESS_TRANSFER_WRITE_BIT, we can still rely on user barriers to do
the right thing (e.g., flush caches for host access).
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32697>
When VK_QUERY_RESULT_WAIT_BIT is set, we rely on sync wait. When
VK_QUERY_RESULT_WAIT_BIT is not set, no wait is needed.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32697>
We can guarantee ordering with this sequence of async cmds
RUN_FRAGMENT ->
(signal and wait SB_ITER) ->
FLUSH_CACHE2 ->
(signal and wait DEFERRED_FLUSH) ->
SYNC_SET32
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32697>
The spec says
VUID-vkCmdBeginQueryIndexedEXT-None-00807
All queries used by the command must be unavailable
and panvk_cmd_reset_occlusion_queries is synchronous.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32697>
The spec says
Resetting a query via vkCmdResetQueryPool or vkResetQueryPool sets the
status to unavailable and makes the numerical results undefined.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32697>
The spec says
VUID-VkQueryPoolCreateInfo-queryCount-02763
queryCount must be greater than 0
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32697>
The spec says
After query pool creation, each query is in an uninitialized state and
must be reset before it is used.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32697>
This switch all __gen_unpack functions to macros to keep address space
information when working with OpenCL C.
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32962>
Because of OpenCL C, we need a way to retain address space information
contains with the pointers.
As a result this switch all [un]pack functions to macros, resulting in
pointers retaining their respective address space information.
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32962>
Provide a pan_cast_and_[un]pack() to help with the transition.
Those helpers should only be used when the caller is sure the
destination is big enough to emit the descriptor.
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32962>
We are now strongly typing everything, pan_[un]pack wil enforce this at
the API level next.
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32962>
Will be useful to easily define packed type variables from the
pan_[un]pack() functions, which we'll need during the pan_pack
revamp.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32962>
We are going to use packed structs in [un]pack next so we need those to
be emitted before them.
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32962>
The generic bits in autogen pack helpers files were extracted in a
common header, so let's include it from the autogenerated file rather
than copying its content there.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32962>
Masking with an ~7ull promotes the value to 64-bit, leading
to a size mismatch when we cast it to a pointer.
Make sure we're using an uintptr_t type for the mask.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32938>
On x86-32, long long are aligned on 4-bytes only, which breaks
the assumption we had about our sysvals struct layouts.
Define an aligned_u64 embedding the alignment attribute to
keep the alignment sane.
While at it, enforce this alignment with an alignment attribute
on the struct itself.
This fixes the build on x86-32, and should do what we expect,
though it's not been tested in practice.
Fixes: ae76a6a045 ("panvk: Pack push constants")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/12429
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32938>
This feature actually worked, but accidentally triggered some
CTS failures due do an unrelated sample_mask bug. That bug
is fixed in the previous commits in this series, so re-enable
fragmentStoresAndAtomics.
There are two spurious CI failures created by this change; one is in
a ycbcr test (which we're still working on) and the other is a CTS
bug (it's using a feature we don't advertise).
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32879>
The ATEST instruction needs sample_mask as an input, but if the
shader writes to color before sample_mask we could emit them
in the wrong order. Fix this in pan_nir_lower_zs_store by
deferring the color write until after the sample_mask write.
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Benjamin Lee <benjamin.lee@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32879>
We have an assert requiring that the sample_mask source for
ATEST be a register. However, the sample mask code can
sometimes produce a FAU index, and this actually works
correctly (tested with the vk and gl CTSes).
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32879>
We have various clang-format issues around on some common code macros.
This should fix this in panvk at least
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32939>
Same as 7ca01506c9 ("panvk: hack to improve depth clipping with
small viewport depth range") but applied to the JM backend.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Benjamin Lee <benjamin.lee@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32905>
vk_rasterization_state_depth_clip_enable() checks the clip and clamp
modes, not the cull mode. RS_DEPTH_CLIP_ENABLE got confused with
RS_CULL_MODE in 7ca01506c9 ("panvk: hack to improve depth clipping
with small viewport depth range").
Fixes: 7ca01506c9 ("panvk: hack to improve depth clipping with small viewport depth range")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Benjamin Lee <benjamin.lee@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32905>
It's empty now, so we don't need to include it from the packer headers.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32899>
This is where those macros are used, and those are the last two
definitions preventing us from dropping panfrost-job.h.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32899>
Move MALI_EXTRACT_INDEX to pan_format.h where all format-related macros
live and kill the unused MALI_EXTRACT_TYPE and MALI_FORMAT_COMPRESSED
macros.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32899>
We use uintXX_t most of the time, so let's be consistent and use the
stdint types everywhere instead of their shorter uXX variants.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32899>
mali_ptr is no shorter than uint64_t, and we already have a few places
where we use uint64_t to store GPU virtual addresses in src/panfrost, so
let's just kill this typedef and use uint64_t types everywhere.
That's one step towards panfrost-job.h removal.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32899>
We're about to add more sysvals, and the more we add, the bigger the
sysvals region gets, which increases the amount of memory we have to
allocate when push_uniforms are dirty.
Instead of allocating FAUs for all sysvals/push_constants, track FAU
usage per-shader, and pack those. This implies emitting an FAU buffer
per stage instead of trying to share it, but that's an acceptable
trade-off.
While at it, automate the sysval dirty tracking a bit.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32415>
Now that panvk_cmd_buffer.h is accessible from
src/panfrost/vulkan/panvk_vX_xxx.c files, there's no reason to pass
a gazillon arguments to blend_emit_descs(). We can just pass a cmdbuf
and let the helper extract the other parameters from there. It also
allows for extra automation, like dirtying the push_uniform buffer
when the new blend config reads the blend constant.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32415>
We're about to make the sysval logic a bit more complication when
introducing push constant packing. Let's first factor-out the sysvals
handling so the JM/CSF backend don't have to duplicate the thing.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32415>
We use blend shaders, so load_blend_const_color_rgba should
not be present in the fragment shader. We might want to re-introduce
this code if we decide to specialize fragment shaders when the
blend configuration is static, but let's drop it for now.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32415>
load_vertex_id_zero_base() is supposed to return the zero-based
vertex ID, which is then offset by load_first_vertex() to get
an absolute vertex ID. At the same time, when we're in a Vulkan
environment, load_first_vertex() also encodes the vertexOffset
passed to the indexed draw.
Midgard/Bifrost have a sligtly different semantics, where
load_first_vertex() returns vertexOffset + minVertexIdInIndexRange,
and load_vertex_id_zero_base() returns an ID that needs to be offset
by this vertexOffset + minVertexIdInIndexRange to get the absolute
vertex ID. Everything works fine as long as all the load_first_vertex()
and load_vertex_id_zero_base() calls are coming from the
load_vertex_id() lowering. But as mentioned above, that's no longer
the case in Vulkan, where gl_BaseVertexARB will be turned into
load_first_vertex() and expect a value of vertexOffset in an
indexed draw context.
We thus need to fix the mismatch by introducing two new
panfrost-specific intrinsic so we can stop abusing load_first_vertex()
and load_vertex_id_zero_base().
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32415>
