Some panvk objects need to allocate GPU memory but don't have Pool
objects to get this memory from. Use device-wide mempools with
.owns_bos=false, such that small allocations don't have to pay the 4k
granularity price of private BO allocations.
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/29161>
If we want to be able to allocate private device memory for small
objects and don't have a Pool to allocate from, we'd rather provide
some sort of device-wide heap to avoid the memory overhead incurred
by page-size BO granularity.
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/29161>
This we don't get memory allocations in scopes where failures are not
supposed to happen, and this will also simplify things when we get to
implement memory pools at the device level.
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/29161>
We need a heuristic for handling GS side effects in the least surprising way
possible. Upgrade our previous heuristic to a better one, moving more side
effects into the prepass from the rast shader. This is technically an optimization but mitigates VDM timeouts in absurd Vulkan CTS cases.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29742>
Different APIs have different robustness requirements for VBOs. Add a knob to
select the desired robustness so we can implement rba2 in honeykrisp.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29742>
We need additional lowering to handle negzero properly. fixes float_controls2
fails but strictly the bug was already present!
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29742>
The MCS region to ambiguate needs to shift 4KB from its
starting address. The first 4KB is reserved for hardware.
Signed-off-by: Jianxun Zhang <jianxun.zhang@intel.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28919>
The scaledown rectangle of MSAA fast clear on Xe2 is 8 times
in X and 2 in Y dimension of previous platforms.
Absorb refactoring change suggested by
Nanley Chery <nanley.g.chery@intel.com>
Signed-off-by: Jianxun Zhang <jianxun.zhang@intel.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28919>
It's possible if nouveau_ws_bo_destroy() races with
nouveau_ws_bo_from_dma_buf() for the BO to be found in the cache and
referenced between dropping the final reference and actually invoking
GEM_CLOSE. This would result in us having a closed BO somewhere in our
cache.
Fixes: c370260a8f ("nouveau/winsys: Add dma-buf import support")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29737>
I've been seeing a bunch of read page faults at the end of the
shader allocation, nvk uses a full page at the end to overallocate
so align with that and see if it goes away.
ahulliet and skeggsb both said 2k was used.
Cc: mesa-stable
Reviewed-by: Arthur Huillet <ahuillet@nvidia.com>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29722>
MOV_INDIRECT picks one lane from the src[0] and moves it to all lanes
in the destination. Even if we split the instruction, src[0] should
remain identical.
Noticed this while trying to use this instruction in SIMD32. All
current use cases are limited to SIMD8 shaders (or SIMD16 on Xe2). Or
maybe in SIMD32 but with a uniform src[0]. That's we think we've never
seen the issue so far.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: mesa-stable
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28036>
if mesh and task shaders are bound separately, and if they have different
workgroup sizes, the setting of workgroup size will be broken if
set during shader bind
this must be deferred to draw time to pull the correct values
cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29733>
Turns out the current approach makes implementation of advanced features
like update-after-bind or shader modules quite challenging. Instead of
adding hacks all over the place to support these features, let's use
the Valhall descriptor model.
Each shader now gets its own descriptor tables, which are fed by pilot
shaders copying the descriptors used by the shader from the descriptor
sets currently bound the command buffer.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Co-developped-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29654>
The Bifrost descriptor model is a dead end for advanced features like
shader modules, pipeline libraries or update-after-bind. Let's prepare
the Valhall implementation so we can re-use it on Bifrost.
The implementation itself was pretty generic already. We just need to
map image descriptors to AttributeBuffer instead Texture descriptors,
and the Buffer object had no equivalent on Bifrost, so we just use a
software-defined panvk_ssbo_addr object containing the SSBO address and
size, and make sure this object is padded to 32-bytes.
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/29654>
Valhall descriptor model was loosely based on the Vulkan descriptor
model. Provide a new implementation for the VkDescriptor{Set,Pool}
objects that matches this new model.
Co-developed-by: Mary Guillemard <mary.guillemard@collabora.com>
Co-developed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29654>
Valhall descriptor model was loosely based on the Vulkan descriptor
model. Provide a new implementation for the VkDescriptorSetLayout
object that matches this new model.
Co-developed-by: Mary Guillemard <mary.guillemard@collabora.com>
Co-developed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-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/29654>
Valhall (v9/v10) will be added progressively. In order to allow that,
we need to extend the panvk_per_arch() macro for v9/v10 and tweak
meson.build so it does include valhall sources without compiling
common per-arch files.
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/29654>
The only difference here is the fact image attributes are gone, and
texture descriptors are used instead. We rework the code so it uses
panfrost_new_texture() to emit the texture/plane descriptors, which
leaves us with one less thing to worry about.
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/29654>
There's no image attribute on Valhall, storage images are passed as
texture descriptors.
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/29654>
