Rather than assuming a6xx+ means mergedregs. We can actually (mostly?)
do splitregs on a6xx as well. And GS/DS/HS currently require it, which
might be papering over a bug, or might be something to do with how
chaining shaders works. At any rate, we should at least be consistent,
and not have the compiler thinking we are doing mergedregs when we are
actually doing splitregs.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5458>
Reduce linear alignment, and rework the layout code a bit.
This rework has a side effect of also increasing the alignment on linear
levels of tiled (non-ubwc) cpp=1 and cpp=2 layouts. Since we should be
UBWC for those cases anyway, its not a big loss.
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5013>
"FETCHSIZE" is actually a "minimum pitch" or "pitchalign" value that's
relevant for mipmaps. The 0 value means 64-bytes. Understanding this allows
some simplifications and will make it possible to use less alignment on
linear formats.
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5013>
All squashed into a single commit because it shouldn't have any
behaviour change, except that it might work now on platforms where it
was broken because F_DUPFD_CLOEXEC is not supported but FD_CLOEXEC is.
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5369>
Implement GMEM input attachments by using non-bindless texture state which
is emitted at the start of every subpass.
This achieves two things:
* More vulkan-like CmdBindDescriptorSets
* Fixing secondary command buffer input attachments with GMEM
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5446>
This reworks dynamic states to use draw states, and reworks draw states.
This moves towards doing as little as possible in bind_draw_states.
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5446>
This moves some logic out of bind_draw_states, moving towards the eventual
goal of doing very little in bind_draw_states.
Split this out as a separate patch to make the DIRTY_INPUT_ATTACHMENTS more
visible: it can be safely removed because pipelines are subpass specific,
so there will always be a pipeline change to go with the CmdBeginRenderPass
and CmdNextSubpass (the CmdBindPipeline may not be in the subpass, but the
draw that flushes the pipeline update will be).
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5446>
The comment about fragment shader state overwriting compute shader state
is wrong, if either path is overwriting the other's state then it is a
mistake.
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5446>
The only two fields were always true, and I don't think we'd ever have
use for them. If we want to disable optimizations then we'd need a
different approach, and I don't even know what include_binning_pass was
for.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5500>
This will be necessary once we start compiling multiple variants due to
different const size limits, and it will also be necessary for properly
implementing the pipeline cache.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5500>
ir3_shader_from_nir() calls ir3_optimize_nir(), which currently sets up
the const state. However, we need to know the number of user consts
reserved by the driver before setting up the const state, which means
that this information needs to be passed into ir3_shader_from_nir()
somehow rather than being set in the shader.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5500>
This fixes the newly added cubic blit_image tests for A650, by falling back
to the 3D path and setting the filter correctly.
Note: there are still failures with the texture filtering tests.
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5509>
This change accidentally made it into 72d7df40a5, and started causing
blit_image flakes (because of the issue fixed in the previous patch)
Fixes: 72d7df40a5 ("turnip: add layered 3D path clear for CmdClearAttachments")
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5509>
A bit cleaner than open coding the list manipulation. Plus I want to
use it in the next patch, rather than adding more open coded list
futzing.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5280>
When a sequence of same instruction is encoded with repeat flag,
destination registers are written on successive cycles. Teach the
delay calculation about this.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5280>
These two encodings are mutually exclusive. If the instruction is a
vector(ish) `(rptN)` instruction, then we can't fold a `(nopN)` post-
delay into it.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5280>
In the `try_swap_mad_two_srcs()` case, valid_flags() gets called both
for the src that we want to try to fold, and for the other src that we
are trying to swap to make that possible. It can happen in the 2nd case
that a RELATIV src has already been folded. Since `ssa()` returns non-
null in both the `IR3_REG_SSA` and `IR3_REG_ARRAY` cases (in the later
case, it is the dependent array access that the current instruction
cannot be moved ahead of), we need to explicitly check that the src
reg we are looking at is still an SSA src.
Reported-by: Jonathan Marek <jonathan@marek.ca>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5280>
It is better to use `nir_intrinsic_dest_components()` which also handles
the case of intrinsics with a fixed number of dest components.
Somehow this starts showing up with a nir_serialize round-trip with
shader-cache. But we really shouldn't have been relying on
`intr->num_components` directly.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5371>
The only reason for this dependency was the fd_bo used for the uploaded
shader. But this isn't used by turnip. Now that we've unified the
cleanup path from gallium, it isn't hard to pull the fd_bo upload/free
parts into ir3_gallium.
This cleanup has the added benefit that the shader disk-cache will not
have to deal with it.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5476>
ir3_nir_move_varying_inputs is broken when there a load input outside of
the first block which depends on the result of a previous load input.
This simplification/rework avoids the problem, and should also be faster.
Fixes this dEQP-VK test:
dEQP-VK.pipeline.multisample_interpolation.offset_interpolate_at_pixel_center.128_128_1.samples_2
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5465>
This fixes cases where the 3D path is used with layered rendering.
Fixes dEQP-VK.renderpass.suballocation.multisample_resolve.layers* failures
Note the blob's 3D fallback path behaves differently, and uses the
framebuffer information to clear each layer individually (changing the MRT
state each time). But that's not possible in all cases, and the blob fails
to clear properly in dEQP-VK.geometry.layered.*.secondary_cmd_buffer cases.
So this clear path is not based on the blob's behavior.
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5426>
This refactor simplifies things a bit, and will make it easier to share
some logic with tu_clear_blit (see next patches).
This changes the order in which some things are emitted, and emits less
for disabled shader stages. There's also as extra write to SP_GS_PRIM_SIZE
that is removed.
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5426>
Teach RA to setup additional interference to prevent textures fetched
before the FS starts from ending up in a register that is too high to
encode.
Fixes mis-rendering in multiple playcanv.as webgl apps.
Note that the regression was not actually 733bee57eb8's fault, but
that was the commit that exposed the problem.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3108
Fixes: 733bee57eb ("glsl: lower samplers with highp coordinates correctly")
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5431>
This is mainly the "piglit optimization" (ie, since piglit launches an
separate process for for each test). It was never wired up for a6xx,
and makes register class setup unnecessarily complicated. Remove it to
simplify the next patch.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5431>
Refactor a bit the limit checking in the bindless case, and add tex/samp
limit checking for the non-bindless case, to ensure we do not try to
prefetch textures which cannot be encoded in the # of bits available.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5431>
The app is allowed to never bind descriptor sets that are statically
unused by the pipeline, which would've caused a context fault since
CP_LOAD_STATE6 would try to load the descriptors that don't exist. Fix
this by not preloading descriptors from unused descriptor sets. We could
do more fine-grained accounting of which descriptors are used, but this
is enough to fix the problem.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5400>
Replace the various ad-hoc flushes that we've inserted, copied from
freedreno, etc. with a unified system that uses the user-supplied
information via vkCmdPipelineBarrier() and subpass dependencies.
There are a few notable differences in behavior:
- We now move setting RB_CCU_CNTL up a little in the gmem case, but
hopefully that won't matter too much. This matches what the Vulkan blob
does.
- We properly implement delayed setting of events, completing our
implementaton of events.
- Finally, of course, we should be a lot less flush-happy. We won't emit
useless CCU/cache flushes with multiple copies, renderpasses, etc. that
don't depend on each other, and also won't flush/invalidate the cache
around renderpasses unless we actually need to.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4964>
We just dropped the last user which actually cared about the seqno.
This never worked anyway, since the seqno was never reset between
multiple executions of the same command buffer. Turn the part of the
control buffer which used to track the seqno into a dummy dword, and
figure out automatically whether we need to include it. We will
implement seqnos again eventually, with timline semaphores, but that
will likely be totally different.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4964>
The Vulkan blob doesn't do this, and based on my understanding of how
the blob works this is unnecessary. CACHE_FLUSH is already serialized
against all 3d commands so you don't need to wait for rendering commands
to finish before issuing it, and the subsequent wfi + WAIT_FOR_ME will
cause the CP to wait for the CACHE_FLUSH to finish, so there's also no
need to wait for it to complete. The CACHE_INVALIDATE also seems
unnecessary, and also isn't done by the blob.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4964>
Note this changes the value of SP_GS_CTRL_REG0, by using FOUR_QUADS and
setting MERGEDREGS. ir3 expects MERGEDREGS, and using FOUR_QUADS instead
of TWO_QUADS doesn't seem to hurt.
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5370>
Fixes failures with TU_DEBUG=forcebin and geometry shaders, for example:
dEQP-VK.binding_model.*geometry*
dEQP-VK.transform_feedback.simple.query*
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5370>
Otherwise A6XX_TEX_SAMP_1_{MIN,MAX}_LOD silently overflows.
This fixes these tests:
dEQP-VK.texture.explicit_lod.2d.derivatives.*
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5375>
The xml.etree.cElementTree module will be removed in Python 3.9. Since
Python 3.3 the xml.etree.cElementTree module has been deprecated, the
xml.etree.ElementTree module uses a fast implementation whenever
available.
Builds using Python 2.7 can still work but with the slower
implementation.
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Acked-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5349>
The GLES blob on the p3a limits constlen to 512 between VS and FS across
a6xx gpu ids (615, 630, 640, and 650). Experimentally, exceeding that
limit in any one stage results in rendering corruption or GPU hangs
(though my most detailed testing had a loop limit in a uniform, so that
may the cause of the hang). Clamp the limit we use inside of a shader so
we don't exceed it within a stage.
This commit doesn't resovle limiting inter-stage. Experimentally, I've
found that I can push up to a total of ~768 vec4s between VS and FS on
a630, with or without uniform updates between each draw. We'll need to do
some shader key-based limiting of constlen at draw time to respect that
limit, but that's left for future work, and this commit is enough for the
google earth case that initiated this work.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5273>
It turns out the GL uniforms file is larger than the hardware constant
file, so we need to limit how many UBOs we lower to constbuf loads. To do
actual UBO loads, we'll need to be able to upload UBO 0's pointer or
descriptor.
No difference on nohw 1 UBO update drawoverhead case (n=35).
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5273>
Comparing a blob trace using the feature to one not, the difference was
pretty obvious and in the spot you'd expect compared to alphaToCoverage.
The SP_ reg didn't have a corresponding bit set, though it also has an
alphaToCoverage.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5343>
We were remapping the bindings so the HW binding points were consecutive,
which there's no need for. Now that we don't shuffle, we can mostly drop
the dependency on the pipeline for this SDS.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5321>
And document the early-lrz-late-z mode.
Initially I thought this would be two bits to control early-lrz vs
early-z. But having early-z without early-lrz does not make sense,
and the way the values line up makes an enum fit better.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5298>
Unlike other conditions which prevent early-discard of fragments, kill
does not prevent early LRZ test. Split `has_kill` from `no_earlyz` so
we can take advantage of this.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5298>
This allows us to do API specific checks before removing variable
without filling nir_remove_dead_variables() with API specific code.
In the following patches we will use this to support the removal
of dead uniforms in GLSL.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4797>
This fixes render_components being 0 when mrt_count=8, because shift by 32
is UB and in arm64 it ends up shifting by 0. This fixes tests with 8 MRTs.
Fixes the 3d path sysmem CmdClearAttachments to set RENDER_COMPONENTS, as
it was previously relying on tu6_emit_mrt setting it, but it is now part of
the pipeline state.
Also switch back to the previous behavior of not setting render components
for VK_ATTACHMENT_UNUSED attachments: we don't update the MRT state for
such attachments so we definitely don't want to be trying writing to those.
Fixes: 078aa9df8d ("tu: Move RENDER_COMPONENTS setting to pipeline state")
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5292>
This uses a meson builtin to handle -fvisibility=hidden. This is nice
because we don't need to track which languages are used, if C++ is
suddenly added meson just does the right thing.
Acked-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4740>
Doesn't seem to be a big win, although I could still be missing
something in my implementation. But might as well add the
documentation.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5217>
This will be used to import images which have different layout from what
turnip uses by default. For example non-UBWC (linear) images from the video
decoder on some hardware have a 512 pitch alignment.
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4596>
First element is not a scalar. Just initialize the struct like we do
elsewhere.
src/freedreno/ir3/disasm-a3xx.c:958:33: warning: suggest braces around
initialization of subobject [-Wmissing-braces]
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5174>
We're already using the {} syntax elsewhere in turnip.
src/freedreno/vulkan/tu_formats.c:828:71: warning: suggest braces
around initialization of subobject [-Wmissing-braces]
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5174>
This makes it a little more explicit that the values line up.
src/freedreno/vulkan/tu_device.c:2209:75: warning: implicit conversion
from enumeration type 'const VkSamplerReductionMode' (aka 'const enum
VkSamplerReductionMode') to different enumeration type 'enum
a6xx_reduction_mode' [-Wenum-conversion]
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5174>
We're hard-coding this value, so let's use the hw enum and avoid a
warning.
src/freedreno/vulkan/tu_clear_blit.c:2091:19: warning: implicit
conversion from enumeration type 'enum VkStencilOp' to different
enumeration type 'enum adreno_stencil_op' [-Wenum-conversion]
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5174>
This adds RB, VFMT and TFMT NONE values for a3xx-a5xx and FMT6_NONE
for a6xx. Use those values instead of open coded (enum xxx) ~0 or
sometimes even ~0, which triggers out-of-enum range warnings.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5173>
The closed GL driver uses resinfo on images with the writeonly flag (using
the texture-path's getsize only for readonly images). The closed vulkan
driver seems to use resinfo regardless. Using resinfo doesn't need any
fixups after the instruction. It also avoids one of the needs for the
TEX_CONST state for the image, which is awkward to set up in the GL
driver.
The new handler goes into ir3_a6xx to be next to the other current image
code, but the a4xx version is left in place because it wants a bunch of
sampler helpers.
Fixes assertion failure in dEQP-VK.image.image_size.buffer.readonly_32.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3501>