Re: [PATCH v5 00/10] YUV conversion colorop with amdgpu and VKMS

Harry Wentland <[email protected]>
Newsgroups org.freedesktop.lists.dri-devel,org.freedesktop.lists.amd-gfx
Message-ID <[email protected]>

On 2026-08-14 16:03, Robert Mader wrote:
> Hey Harry,
> 
> "[PATCH v5 08/10] drm/amd/display: Check actual state during commit_tail" still fails to build - it modifies fill_plane_color_attributes() in amdgpu_dm.c, but not in amdgpu_dm.h and amdgpu_dm_test.c.
> 

Ah, I didn't have KUNIT enabled for my normal builds. Will send
a v6.

Harry

> Regards
> 
> On 14.08.26 21:33, Harry Wentland wrote:
>>
>> On 2026-08-01 05:42, Robert Mader wrote:
>>> Hi Harry,
>>>
>>> On 31.07.26 20:15, Harry Wentland wrote:
>>>> When we merged the drm_plane color pipeline API the major gap
>>>> that existed was the lack of a YUV to RGB conversion colorop.
>>>> We deprecated any legacy drm_plane color properties, which
>>>> means that the COLOR_RANGE and COLOR_ENCODING properties can't
>>>> be used with the COLOR_PIPELINE property on a drm_plane. In
>>>> practice this means that we can't use a COLOR_PIPELINE on
>>>> YCbCr encoded framebuffers.
>>>>
>>>> This patchset expands on the Fixed Matrix colorop proposed by Chaitanya
>>>> and adds limited range variants of the YCbCr to RGB conversions.
>>>>
>>>> His full patchset can be found at
>>>> https://patchwork.freedesktop.org/patch/709860
>>>>
>>>> This code has been tested with IGT and an experimental KWin branch.
>>>>
>>>> All patches are now reviewed and tested. We have a Weston and
>>>> KWin implementation. IGT patches are missing one review. I
>>>> deem these patches ready to merge once the last IGT patch review
>>>> comes in.
>>>>
>>>> IGT branch:
>>>> https://gitlab.freedesktop.org/hwentland/igt-gpu-tools/-/tree/yuv-fm-colorop
>>>>
>>>> KWin branch used for testing:
>>>> https://invent.kde.org/hwentlan/kwin/-/tree/yuv-fm-colorop
>>>>
>>>> The kernel branch containing these changes, based on drm-misc-next
>>>> can be found at:
>>>> https://gitlab.freedesktop.org/hwentland/linux/-/tree/yuv-fm-colorop
>>> I wanted to give this a quick go with the Weston implementation [1], however unfortunately the branch doesn't build for me and fails with the error below.
>>>
>> I forgot to update the series and still had a bad branch sitting on this branch.
>> I pushed the latest rebase. There should be no conflicts now.
>>
>> The rebase from v5 was trivial (what was sitting on my FDO tree was older) so
>> no need to send a v6.
>>
>> Harry
>>
>>> With that fixed I hope we can land the series - that would be awesome 🤞
>>>
>>> Regards
>>>
>>> 1: https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/2133
>>>
>>> drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm.c:2987:1: error: conflicting types for ‘fill_plane_color_attributes’; have ‘int(struct drm_atomic_commit *, const struct drm_plane_state *, const enum surface_pixel_format,  enum dc_color_space *)’
>>>   2987 | fill_plane_color_attributes(struct drm_atomic_commit *state,
>>>        | ^~~~~~~~~~~~~~~~~~~~~~~~~~~
>>> In file included from ./drivers/gpu/drm/amd/amdgpu/../amdgpu/amdgpu.h:87,
>>>                   from drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm.c:48:
>>> ./drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm.h:1133:5: note: previous declaration of ‘fill_plane_color_attributes’ with type ‘int(const struct drm_plane_state *, const enum surface_pixel_format,  enum dc_color_space *)’
>>>   1133 | int fill_plane_color_attributes(const struct drm_plane_state *plane_state,
>>>        |     ^~~~~~~~~~~~~~~~~~~~~~~~~~~
>>> In file included from ./include/linux/linkage.h:7,
>>>                   from ./include/linux/printk.h:8,
>>>                   from ./include/asm-generic/bug.h:31,
>>>                   from ./arch/x86/include/asm/bug.h:195,
>>>                   from ./include/linux/bug.h:5,
>>>                   from ./include/linux/slab.h:15,
>>>                   from ./drivers/gpu/drm/amd/amdgpu/../display/dc/os_types.h:30,
>>>                   from ./drivers/gpu/drm/amd/amdgpu/../display/dc/dm_services_types.h:29,
>>>                   from drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm.c:30:
>>> drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm.c:3034:17: error: conflicting types for ‘fill_plane_color_attributes’; have ‘int(struct drm_atomic_commit *, const struct drm_plane_state *, const enum surface_pixel_format,  enum dc_color_space *)’
>>>   3034 | EXPORT_IF_KUNIT(fill_plane_color_attributes);
>>>        |                 ^~~~~~~~~~~~~~~~~~~~~~~~~~~
>>> ./include/linux/export.h:76:28: note: in definition of macro ‘__EXPORT_SYMBOL’
>>>     76 |         extern typeof(sym) sym;      \
>>>        |                            ^~~
>>> ./include/linux/export.h:89:41: note: in expansion of macro ‘_EXPORT_SYMBOL’
>>>     89 | #define EXPORT_SYMBOL(sym) _EXPORT_SYMBOL(sym, "")
>>>        |                                         ^~~~~~~~~~~~~~
>>> drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_kunit_helpers.h:12:33: note: in expansion of macro ‘EXPORT_SYMBOL’
>>>     12 | #define EXPORT_IF_KUNIT(symbol) EXPORT_SYMBOL(symbol)
>>>        |                                 ^~~~~~~~~~~~~
>>> drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm.c:3034:1: note: in expansion of macro ‘EXPORT_IF_KUNIT’
>>>   3034 | EXPORT_IF_KUNIT(fill_plane_color_attributes);
>>>        | ^~~~~~~~~~~~~~~
>>> ./drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm.h:1133:5: note: previous declaration of ‘fill_plane_color_attributes’ with type ‘int(const struct drm_plane_state *, const enum surface_pixel_format,  enum dc_color_space *)’
>>>   1133 | int fill_plane_color_attributes(const struct drm_plane_state *plane_state,
>>>        |     ^~~~~~~~~~~~~~~~~~~~~~~~~~~
>>>
>>>
>>>> Further background on this work can be found at:
>>>> https://hwentland.github.io/2026/03/10/plane-color-pipeline-csc-3d-lut-kwin.html
>>>>
>>>> v5:
>>>>    - Drop new VKMS kunit tests for conversion matrices
>>>>    - Added script to show how VKMS kunit test values are computed (Pekka)
>>>>    - Removed fixed-matrix enums for "YCbCr limtied to full" and
>>>>      "RGB709 to RGB2020" as they're currently unused by userspace (Robert)
>>>>
>>>> v4:
>>>>    - Specify matrix entries in docs (Pekka)
>>>>    - Squash limited-range enums into "Add FM" patch (Robert)
>>>>    - Don't reject RGB planes with fixed matrix in VKMS as
>>>>      we don't want or need to make a colorop dependent on
>>>>      the framebuffer's pixel format. (Robert)
>>>>    - Fix conversion matrices in VKMS and implement kunit
>>>>      tests (discovered while documenting the matrices)
>>>>
>>>> v3:
>>>> - base on Chaitanya's updated patch and rename code accordingly
>>>>     to Fixed_Matrix instead of CSC Fixed-Function
>>>>
>>>> v2:
>>>> - use Chaitanya's CSC_FF block for named matrices
>>>>
>>>> Cc: Alex Hung <[email protected]>
>>>> Cc: Daniel Stone <[email protected]>
>>>> Cc: Chaitanya Kumar Borah <[email protected]>
>>>> Cc: Uma Shankar <[email protected]>
>>>> Cc: Louis Chauvet <[email protected]>
>>>> Cc: Melissa Wen <[email protected]>
>>>> Cc: Simon Ser <[email protected]>
>>>> Cc: Robert Mader <[email protected]>
>>>>
>>>> Chaitanya Kumar Borah (1):
>>>>     drm/colorop: Add DRM_COLOROP_FIXED_MATRIX
>>>>
>>>> Harry Wentland (9):
>>>>     drm/vkms: Fix limited-range YCbCr to RGB conversion scaling
>>>>     drm/vkms: Add fixed matrix colorop to color pipeline
>>>>     drm/vkms: Add atomic check and matrix handling for fixed matrix
>>>>       colorop
>>>>     drm/amd/display: Add fixed matrix colorop to color pipeline
>>>>     drm/amd/display: Implement fixed matrix colorop color space mapping
>>>>     drm/amd/display: Use GAMCOR for first TF if YUV conversion is needed
>>>>     drm/amd/display: Check actual state during commit_tail
>>>>     drm/amd/display: Set color_space to plane_infos
>>>>     drm/amd/display: Force GAMCOR for subsampled surfaces with
>>>>       PQ/Gamma22/HLG
>>>>
>>>>    .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c |  14 +-
>>>>    .../amd/display/amdgpu_dm/amdgpu_dm_color.c   |  85 +++++++++++-
>>>>    .../amd/display/amdgpu_dm/amdgpu_dm_colorop.c |  27 +++-
>>>>    .../amd/display/amdgpu_dm/amdgpu_dm_colorop.h |   1 +
>>>>    .../amd/display/modules/color/color_gamma.c   |   3 +-
>>>>    drivers/gpu/drm/drm_atomic.c                  |   4 +
>>>>    drivers/gpu/drm/drm_atomic_uapi.c             |   7 +
>>>>    drivers/gpu/drm/drm_colorop.c                 | 107 +++++++++++++++
>>>>    .../gpu/drm/vkms/tests/gen_yuv_conversion.py  |  87 ++++++++++++
>>>>    drivers/gpu/drm/vkms/tests/vkms_format_test.c |  40 +++---
>>>>    drivers/gpu/drm/vkms/vkms_colorop.c           |  66 ++++++---
>>>>    drivers/gpu/drm/vkms/vkms_composer.c          |   6 +
>>>>    drivers/gpu/drm/vkms/vkms_formats.c           |  64 ++++++---
>>>>    drivers/gpu/drm/vkms/vkms_formats.h           |   2 +-
>>>>    drivers/gpu/drm/vkms/vkms_plane.c             |  55 +++++++-
>>>>    include/drm/drm_colorop.h                     | 127 ++++++++++++++++++
>>>>    include/uapi/drm/drm_mode.h                   |  12 ++
>>>>    17 files changed, 639 insertions(+), 68 deletions(-)
>>>>    create mode 100755 drivers/gpu/drm/vkms/tests/gen_yuv_conversion.py
>>>>
>>>> -- 
>>>> 2.55.0
>>>>
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.