We need to move some earlier (so they are <= 255 like the comment says)
and add a few 64-bit ones.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17894>
This is almost always a nir_instr and updating the src of a nir_if will
have to work slightly differently in the future.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Acked-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12910>
This fixes TC's buffer invalidation code for buffers that are shared
between contexts.
TC is unable to notify other TCs in case it replaces a shared buffer's
underlying storage when invalidating, causing those other TCs to use
wrong buffer IDs for busyness tracking, which leads to corruption due
to invalidation fast-paths being triggered when they shouldn't be.
This patch addresses this issue by tracking if a buffer is shared, and
if it is, disabling buffer storage replacement for the affected buffer.
This is achieved by tracking which TC instance first accessed a certain
buffer. If a second instance then accesses it as well, it will realize
that it isn't the only one working on the buffer and mark the buffer
accordingly.
If TC needs to invalidate a buffer for the correctness of an operation
at any point, it will fall back to doing the operation in a synchronous
fashion with this patch if the buffer is shared and currently busy.
It might be possible to later detect that a buffer has become un-shared;
however, this is outside of the scope of this bugfix patch.
v2: Do not disable buffer busyness tracking for shared buffers.
Fixes: e9c41b32 ("gallium/u_threaded: add buffer lists - tracking of buffers referenced by tc")
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17338>
This flag should allow Zink to stop using the interleave-specific API,
and instead just use the normal/full u_transfer_helper API, and start
using some more of these helpers.
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17959>
Whenever we add or remove a flag here, we need to update a bunch of
drivers in a fragile way. Moving to flags here instead should make this
a bit easier to maintain in the future.
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Reviewed-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17959>
The 32-bit per component pipe_formats are a bit different than the 8 and
16 bit formats, in that there's no UNORM or SNORM variants of them. So
let's just omit those variants.
While we're at it, update a comment that was already out-of-date anyway.
Acked-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Soroush Kashani <soroush.kashani@imgtec.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18012>
Multi-mode multi-draws will make it more complicated, so let's start with
simpler code.
I changed the order a little: I put the VBO update next to emit_draw_packets.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18195>
This moves setting those registers from an unconditional place in draw_vbo
into si_set_rasterized_prim (for draw_vbo), si_update_rasterized_prim
(for bind_xx_shader), and si_bind_rs_state.
It's a little more complicated than expected.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18195>
This removes the parameter from si_emit_derived_tess_state and uses
si_context to pass it. This rework is needed for multi-mode draws
where num_patches will be needed much later.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18195>
with other various changes, this should enable the entirety of compute
shader creation to occur in the compile thread
it also enables (slightly) simplifying shader create by moving cube check
out to the caller
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18197>
this got broken in the move to u_live_shader_cache, which does not
handle compute programs and thus does not populate the shader's sha1
cc: mesa-stable
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18197>
when compute shaders can be precompiled, they can be precompiled asynchronously
which allows the implementation of the parallel shader compile hooks
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18197>
for compute shaders that don't need spec constants or cube lowering,
precompiles are possible and can be performed immediately after disk
cache lookup completes
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18197>
if there are no inline uniforms, nonseamless cubes, or local size use,
then this is the "base" pipeline object that can be reused without checking
the hash table
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18197>
this simplifies the whole compute shader/program architecture and
also compiles compute shaders when apps maybe expect them to be compiled
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18197>
