This entire pattern really wants to be a shared helper, to allocate a
shadow linear image from another device and then import it across into
the rendering GPU. Querying the FourCC from the DRIImage makes it easier
to pull out into shared code.
This temporarily makes the implementation more ugly, however it's
already pretty hard on the eyes, so probably no great loss.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27709>
When we create a texture, we allocate a pipe_resource for it with a
given format. When we export the texture to an EGLImage, we go look
up what the GL format was and try to re-derive the pipe_format back
again.
Instead of doing that, just immediately reuse the pipe_format for the
image's format.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27709>
When we create a renderbuffer, we allocate a pipe_resource for it with a
given format. When we export the renderbuffer to an EGLImage, we go look
up what the GL format was and try to re-derive the pipe_format back
again.
Instead of doing that, just immediately reuse the pipe_format for the
image's format.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27709>
When the DRI frontend generates configs to pass to the APIs, store the
pipe_format in the internal gl_config structure. Doing this allows us to
nuke a few places where we try to retcon our way back from the config to
a pipe_format.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27709>
When populating DRI configs, driconf allows us to exclude
10bpc/RGBA/float configs. Rather than having a hardcoded set of formats
which satisfy those properties, just check the actual format
definitions.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27709>
Rather than just passing down the depth/stencil bits, pass the
pipe_format for the depth/stencil modes down to driCreateConfigs so we
retain the explicit information as long as possible.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27709>
When we're building configs, be a bit more explicit about which format
we're using, as a prelude to passing the formats down explicitly to
driCreateConfigs.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27709>
Just get all the information we need from u_format, rather than keeping
it in hand-written tables.
If you have bisected any kind of failure to this commit, please build
the previous commit in debug mode (so that assertions are enabled) and
check if there are any runtime failures. If so, there is a difference
between the u_format descriptions and the handwritten descriptions that
this commit deletes.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27709>
Similar to util_format_get_component_bits, get the bit offset for a
particular channel in a given format.
Use this to calculate the shift/mask sets for formats when creating DRI
configs, as a prelude to ripping out and replacing the hardcoded table.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27709>
Now that we have both the pipe_format and the mesa_format available to
us, prefer the former so we don't need to map between the two worlds.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27709>
These aren't used by any external DRI callers, so unexport them and
redeclare them in terms of PIPE_FORMAT_*, so we can flatten them down
later.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27709>
When we're binding a GL texture to an EGLImage, we have to be sure to
check the right resource. A GL texture resource can contain multiple
pipe_resources on different levels; so once we get the texture object,
we have to chase that down to a texture image before we use the
pipe_resource.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27709>
This doesn't seem supported and introduced regressions on
Pitcairn/Tahiti, but it seems fine on Bonaire (GFX7).
Fixes: 926d9f1cef ("radv: support minmax filter for more formats")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27760>
We have couple of dynamic fields used on >= gfx125, ScratchSpaceBuffer
and TessellationDistributionMode. There are also 2 workarounds dealing
with TessellationDistributionMode. These changes make sure that we
always emit both DS and TE together when the dynamic fields need to be
changed.
This fixes KHR-Single-GL46.arrays_of_arrays_gl.AtomicUsage on MTL.
Cc: mesa-stable
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/10636
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27753>
This adds support for AV1 decode to radv, there are likely some
cleanups necessary.
Co-author: Charlie Turner <cturner@igalia.com>
(I wrote the initial mesa extension, and Charlie ported it to
the KHR extension and kept it up to date and fixed CTS things)
Acked-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27424>
This parallels changes made to the radeonsi code, but since we have uvd
in here as well we still use family in some places.
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27424>
The prior refactor has missed a case that timeline sempahore can be
ping-pong'ed between device signal and host wait.
Fixes: d63432012d ("venus: refactor semaphore feedback")
Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Reported-by: Juston Li <justonli@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27807>
There is more to do but this is able to parse batch buffers, including
fetch and parse other bos like shader programs.
Some functions used by read_i915_data_file() could be shared with Xe
so I have moved those to aubinator_error_decode_lib.c/h.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27661>
To ask Xe KMD to dump BO/VMA content to error state it is necessary to
set DRM_XE_VM_BIND_FLAG_DUMPABLE during VM bind, so places that were
setting 'bo.real.capture = true' after the VM bind were useless for
Xe KMD.
To fix that BO_ALLOC_CAPTURE was added, alloc_fresh_bo() will set
'bo.real.capture = true' automaticacly for new bos before vm bind.
alloc_bo_from_cache() also required to check if capture states matches
between asked bo and bo in cache.
Only slabs might be left out of error dump if DEBUG_CAPTURE_ALL is not
set but that was already happening for i915.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27661>
Move intel_nir_lower_texture just before nir_lower_tex since we need to
operate on the offset and those are getting lowerd.
v2: (Ian)
- Rename variable name to intel_tex_options
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27447>
This allow us to invoke the quad helper.
v2: (Georg)
- Add check for is_gather_implicit_lod
Fixes: 48158636bf ("nir: add is_gather_implicit_lod")
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27447>
Bspec: 56996
Programming note:
When doing cube arrays on gather4_l and gather4_b with SIMD16 or
SIMD32 the AI parameters is combined with the LOD/bias parameter on
the 9 LSBs.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27447>
On Xe2+, we need to pack LOD with array index for cube array surfaces,
with that mlod parameter gets adjusted to different indices based on the
layout.
So track if we are packing LOD with array index in fs_inst and propogate
that to sampler lowering code to adjust param location.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27447>
This fixes the off-by one mistake, we want to trim the vector till the
array index not till the coordinates last component.
v2: (Ian)
- Use tex->coord_component - 1 in nir_trim_vector
Fixes: 73a3257968 ("intel/compiler: Add texture operation lowering pass")
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27447>
When we bind a descriptor set with dynamic descriptors, we can't ignore
dynamic descriptors in previously-bound higher descriptor sets. For
example, assume we have descriptor sets A and B, each of which has one
dynamic storage buffer, and we do:
CmdBindDescriptorSets(firstSet=1, descriptorSetCount=1, A)
CmdBindDescriptorSets(firstSet=0, descriptorSetCount=1, B)
and in the first CmdBindDescriptorSets the pipeline layout includes a
descriptor set layout compatible with B in set 0. Then, following
"Pipeline Layout Compatibility," set 0 is disturbed:
When binding a descriptor set to set number N, a previously bound
descriptor set bound with lower index M than N is disturbed if the
pipeline layouts for set M and N are not compatible for set M.
Otherwise, the bound descriptor set in M is not disturbed
When it's disturbed, it's effectively turned into a set with 1 undefined
dynamic storage buffer:
When a descriptor set is disturbed by binding descriptor sets, the
disturbed set is considered to contain undefined descriptors bound
with the same pipeline layout as the disturbing descriptor set.
This disturbed set is compatible with B, so in the second
CmdBindDescriptorSets this clause doesn't apply:
If, additionally, the previously bound descriptor set for set N was
bound using a pipeline layout not compatible for set N, then all
bindings in sets numbered greater than N are disturbed.
and A remains valid to access. The code before 88db7364 worked only if
the pipeline layout when binding B contained a descriptor layout
compatible with A in set 1, because it used the pipeline layout's total
size when allocating the internal dynamic descriptors array, but that
isn't actually a requirement, so the previous code was already broken.
After 88db7364 we only allocate as much space as required by the current
descriptors being bound, because I misread the rules here, which made it
more broken and broke 3DMark Wildlife Extreme that does something like
this.
In order to properly fix this we need to keep track of the maximum ever
seen dynamic descriptor size, similar to what we already do for
descriptor sets, and use that. We have no idea what needs to be
preserved when binding a descriptor set with dynamic descriptors, so we
have to be conservative.
Fixes: 88db7364 ("tu: Rework dynamic offset handling")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27750>
1. move feedback helpers into vn_feedback
2. rename related structs, helpers, etc
3. only recycle wait semaphores is enough for the submission. Later we
can further optimize to only recycle each timeline sempahore once
Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27758>
