If a meta operation (like a blit or clear) happens while occlusion
queries are active, we temporarily disable the query. Unfortunately
the code for this did not clear out the `syncobj` field. In rare
combinations of circumstances this could cause an attempt to issue
a write back of the occlusion query values, and since we've zeroed
the `ptr` field it writes to a NULL value, causing a bus fault and
device lost error.
Fixes: 61534faf4e ("panvk: Wire occlusion queries to internals")
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/36525>
While the shaderCoreBuiltins feature is only supported on v9+, the
extension itself provides some useful physical device properties and is
supported on all GPUs.
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36019>
Both MSVC/MINGW support this.
For fixes compiling with MINGW/GCC
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36598>
This is for fixes compiling with MINGW/GCC
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36598>
16x MSAA isn't supported at all on certain Xe3 variants, and on its way
out on the rest. Most vendors choose not to support it, and many apps
offer more modern multisampling and upscaling techniques these days.
Only 2/4/8x are supported going forward.
Cc: mesa-stable
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36592>
The mask here was already 0xff (8-bits) because only 8x MSAA is
supported, but the comment was copy-pasted from iris which did 16x.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36592>
No platforms supported by crocus actually support 16X MSAA, so this is
just copy-pasted from iris without actually offering any benefit.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36592>
Since July 2024, *ERROR* is no longer printed in dmesg and this caused
GPU hangs to not be detected with recent kernels. Hopefully, this won't
uncover more problems in CI...
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36611>
Invalidating descriptors at bind time is wrong in a scenario like
trace rays -> dispatch -> trace rays because RT uses compute shaders
and the user SGPRs need to be re-emitted.
Noticed by Konstantin.
Fixes: 40ceece75f ("radv: invalidate compute/rt descriptors at pipeline bind time")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36605>
We don't actually need to put the size of the AS in the descriptor, as
the blob is doing, because we don't need to rely on bounds checking as
the index is controlled by the driver. The only chance to get a
"garbage" index is if ray query functions are called in the wrong order,
which is already undefined behavior. By just always using the maximum
size, we can stop tracking sizes for VK_EXT_descriptor_buffer where only
the AS address is provided.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36502>
