For fast-linking, we really want to upload the binaries directly in
a library to avoid creating and uploading at pipeline creation time.
To achieve that, add a radeon_winsys_bo pointer to radv_shader in
order to indicate that a shader is already uploaded. When a lib is
imported, the pipeline slab BO is also incremented to make sure it's
not freed when the lib is destroyed.
This also allows to free binaries right after they are uploaded.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Tatsuyuki Ishi <ishitatsuyuki@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18860>
The driver no longer bind NULL graphics pipelines, so these checks
are useless.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18873>
inherited from radv_compute_pipeline to contain all RT-related information.
This will make it easier to transition to RT shader functions.
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18755>
v2 by Timur Kristóf:
- Use ALWAYS_INLINE instead of just inline.
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18807>
When compiling the pre-rasterization stages we don't know the primitive
topology, but we still want to enable NGG culling for performance. To
achieve that, NGG culling is enabled unconditionally when the topology
is unknown and disabled dynamically for points or lines.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18776>
This is completely broken because the PS epilog has refcount and
radv_upload_shaders() updates its VA.
This reverts commit 7c34b31db2.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18632>
If we have a VS that needs a prolog without using the dynamic state,
that means that it comes from a library, so we can overwrite the
cmdbuf VS input state.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18519>
GE_CNTL is the equivalent of IA_MULTI_VGT_PARAM on GFX9 and older.
Calling this function for every draw shouldn't really hurt in practice
because only non-NGG pipelines need this.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18344>
The number of tessellation patches that is computed from the number
of patch control points might change dynamically too.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18344>
Also, update list of expected failures.
dEQP-VK.image.sample_texture.*_bit_compressed_format_two_samplers_*
now reliably pass on Polaris10 (GFX8) and Pitcairn (GFX6).
Stoney has new failures but given there is already a lot of
depth/stencil resolve failures, we shouldn't worry about them.
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15587>
With dynamic rendering, the concept of framebuffer dimensions goes away
so this won't make sense. Even with render passes, the render area is
guaranteed to be inside the framebuffer so we may as well clip to the
potentially smaller render area. This commit also moves window scissor
setup to CmdBeginRenderPass2() time. This should be fine, even for meta
ops, as the only meta ops which happen inside a render pass need the
same render area as the render pass itself.
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15587>
There is a new packet DISPATCH_MESH_INDIRECT_MULTI which we use
for indirect mesh shader only draws. This packet allows using
a 3D dispatch (however firstTask doesn't work with this packet).
Otherwise everything else works the same way as
the NV_mesh_shader draw calls.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18367>
This is the standard pattern in the kernel for providing vfunc tables
for C objects. We're using it in the pipeline cache code but we're
about to start adding more stuff and so it really helps if we have it
for command buffers as well.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18324>
Most other init functions follow the Vulkan API convention of putting
the parent object first.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18324>
v2 (Jason Ekstrand):
- Handle rename to vk_command_buffer_set_error
- Add more instances, especially in device generated commands
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16918>
This is allowed by the Vulkan spec when GPL is used.
Fixes dEQP-VK.pipeline.pipeline_library.graphics_library.misc.other.null_descriptor_set_in_monolithic_pipeline
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18313>
Only re-emit the scissor state if viewports or scissors change, and
only re-emit the guardband state if viewports, line width or the
current rasterized primitive change.
This should reduce the number of emitted packets when only the line
width or the rasterized primitive change.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18247>
The guardband state (part of scissors) needs to be re-emitted when
the line width changes. Given this is a dynamic state, it's not
necessary to look at the pipeline line width.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18247>
Color write enable can change CB_TARGET_MASK and emitting a BREAK_BATCH
seems needed for binning. Though, this was broken if this enable bit
changed dynamically for the same pipeline. Split the function to not
increase CPU overhead.
Found by inspection.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18233>
This is probably a leftover when task shader has been reworked, but it
has no effect.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18210>
The following sequence would be broken if we don't re-emit viewports.
vkCmdSetViewport()
VkCmdBindPipeline(negative_one_to_one = false)
vkCmdDraw()
VkCmdBindPipeline(negative_one_to_one = true)
vkCmdDraw()
Found by inspection.
Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18245>
RMW context registers have been removed in RadeonSI a while ago
because they don't seem good for performance.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18234>
Some dynamic states always need to be emitted when the first pipeline
is emitted, some others depend on pipeline state.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18234>
These fields aren't set at pipeline creation, so clearing them is
just useless.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18234>
This assumption is probably valid with render passes but it won't for
dynamic rendering where we may not have a layout for depth at all.
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18084>
This commit fixes
dEQP-VK.rasterization.rasterization_order_attachment_access.format on
GFX9 because changing the layout for Vulkan feedback loops will trigger
a fast-clear eliminate. Though, the root cause is unrelated to that and
it's because the CMASK/FMASK initialization on GFX9 is currently broken
for TC-compatible images (there is a TODO somewhere).
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18084>
