This reverts commit 3316bc3e88.
These formats were only used by RADV and are no longer needed as we can get the plane dimensions
from the YCbCr table.
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Signed-off-by: Valentine Burley <valentine.burley@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30899>
Instead of having a hardcoded table to convert between sRGB formats and
their linear-gamma equivalents (and vice-versa), generate this from the
information in the format table.
This requires adding a 'sublayout' attribute to differentiate between,
e.g. DXT1 and DXT3, which otherwise appear to be equivalent but for
their name prefix.
As an anonymous union is being used, we also need named initialisers for
the util_format_description entries.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29649>
Instead of having a hardcoded list of endian-independent format aliases
in the header, generate them from the format definitions.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29649>
u_format has always had its format table in CSV. This is kind of nice
for some things, but is a serious pain to extend, especially with
optional fields.
In going through our many (many, many) duplicated tables of format
mappings, it would've been nice to add some descriptions to our central
u_format table, such as mapping to DRM FourCC, to EGLImage mappings, and
to GL internalformats for EGLImage imports. Unfortunately, doing so with
more additional fields would just make the CSV totally unreadable.
Move the CSV table to a YAML-based table and adjust the Python parsers
to suit. The resulting generated files are identical before and after
the transition.
The new parser also has a significant amount of format validation to
make it easier to catch common errors.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29649>
Ensure unsigned integers are used instead of signed ones when performing
left bit shifts.
This has been detected by the Undefined Behaviour Sanitizer (UBSan).
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29772>
Previously, [SU]SCALED formats would hit the integer path and we would
generate:
((uint32_t)CLAMP(src[i], min, max)) & MASK
This is fine for unsigned scaled formats. However, for signed formats,
a negative float value cast to an unsigned integer yields undefined
results. On x86, it implicitly clamps to 0. This change makes us
generate:
((uint32_t)(int32_t)CLAMP(src[i], min, max)) & MASK
hich gets us correct casting.
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28793>
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>
These actually match the Vulkan format enums for single-plane YCbCr
formats, unlike the other PIPE_FORMAT_* enums which put luminance in the
red channel.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24619>
Without this, gallivm doesn't know what to do with the result data of a
texture fetch. We have a bunch of gallivm code to handle these but, as
far as I can tell, none of it works properly without channel types. The
YUYV texturing helpers all consume a 32-bit packed value, unpack it,
shuffle things as needed, and then re-pack into a 32-bit RGBA value,
trusting later code to unpack that as 8-bit UNORM. For the raw formats,
this unpacking never happens.
What saves us is that I also don't think any of this code is ever used.
LLVMpipe supports the actual YUYV formats and we don't use the gallium
lowering pass which lowers to the raw formats.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24619>
When NV21 lowering with hardware sampling and shader CSC was added, the
incorrect PIPE_FORMAT_G8_B8R8_UNORM was used. That format is supposed to
represent vulkan NV12 instead.
This commit introduces PIPE_FORMAT_R8_B8G8_UNORM, which correctly describes the
gallium mapping for YUV CSC, with R as Y, instead of G as Y.
Fixes: 26e3be513d ("gallium/st: add support for PIPE_FORMAT_NV21 and PIPE_FORMAT_G8_B8R8_420")
Signed-off-by: Italo Nicola <italonicola@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24266>
YV12 is the same as DRM_FORMAT_YVU420.
We lower it to PIPE_FORMAT_R8_B8_G8_420, which is equivalent to
PIPE_FORMAT_R8_G8_B8_420 with U/V planes swapped.
This is used for hardware that can sample from YUV but need CSC in shader.
Signed-off-by: Italo Nicola <italonicola@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24266>
This new format is similar to PIPE_FORMAT_G8_B8_R8_420, but with R as Y, G as U
and B as V. The need for two diferent formats here is because gallium maps the
YUV channels differently from vulkan.
Some hardware, e.g. Mali GPUs, can sample from I420 but need CSC in shader,
this patch implements that.
Signed-off-by: Italo Nicola <italonicola@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24266>
Guard util to be standalone by move subdir('util') just behind subdir('android_stub')
texcompress_rgtc_tmp.h is not a standard c header, do not treat it as source
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19522>
As pipe_color_union is used in vulkan drivers, so decouple it from gallium by this move
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19522>
PIPE_MASK_* is used by src/util/*, so do the move to decouple src/util/* from gallium
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19522>
pipe_swizzle is used by src/util/*, so do the move to decouple src/util/* from gallium
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19522>
util_clear_texture implements clear_texture through a memset.
This patch implements u_default_clear_texture, which tries to clear the
given texture using a surface plus clear_render_target or
clear_depth_stencil.
In case this path fails, either because the formats are non-renderable
or for some other reason, we fallback to `util_clear_texture`, which is
guaranteed to work.
This will allow us to make ARB_clear_texture available to every driver,
as well as provide HW acceleration for the clear_texture operation.
If some hardware doesn't want to use it, such as llvmpipe, it can always
just directly point to the software version using pipe->clear_texture.
Signed-off-by: Italo Nicola <italonicola@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23735>
And pipe/p_compiler.h are removed as it not used any more
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: David Heidelberg <david.heidelberg@collabora.com>
Acked-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23577>
These functions are used by gl[Get]TexImage, which imposes no
alignment restructions on the void *pixels parameter.
This fixes an unaligned access in GTK's "gtk:gdk / memorytexture" unit
test on SPARC, which causes the test to fail.
Fixes: 45ae4434b5 ("util: Use bitshift arithmetic to unpack pixels.")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23482>
16K * 16K * 16bpp = 4G, which overflows int32, so layer_stride needs to
have 64 bits. More generally, any "byte_stride * height" computation
can overflow int32.
Use (u)intptr_t in some gallium and st/mesa places where we do CPU access.
Use uint64_t otherwise.
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23389>
This is a poor man's version of MESA_ARRAY_FORMAT.
Implemented based on a gigantic switch-case
with some help from the C preprocessor.
Thank you, preprocessor!
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21692>
This changes the swizzling so that importation of YUYV dmabuf without
dedicated blitter HW can work.
v2: fix the other format too, update test results
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20815>
Because uint comes from pipe/p_compiler.h, and that depends will be removed in future
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19676>
As ubyte comes from p_compiler.h, so do not use it,
and the code calles util_copy_rect may use args with type uint8_t*, ubyte* or unsigned char*,
so use the type void* that consistence with memcpy
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19676>
