drm/atomic-state-helper: fixes for blend-mode-prop warning
Subsystems (4)
v3 cover drm/atomic-state-helper: fixes for blend-mode-prop warning
PREMULTI was the default DRM blend mode until 9813e158d13d ("drm/drm_blend: allow blend mode property without PREMULTI") introduced the possibility of exposing the blend mode property without PREMULTI being supported. However, __drm_atomic_helper_plane_state_init() still resets pixel_blend_mode to PREMULTI (hardcoded), ignoring the supported mode in this property. In the same series, 860e748bddcc ("drm: ensure blend mode supported if pixel format with alpha exposed") starts warning drivers that supports alpha formats in a given planes but doesn't support blend mode property, which includes AMD primary and cursor planes.
For AMD DCN families, overlay and primary planes support the three blend modes, and therefore PREMULTI is still the default mode. So the warning can be fixed by just creating blend mode properties for primary planes too. PREMULTI is the default/unique mode for cursor planes in DCN and DCE driver. However, looks like PIXEL_NONE is the unique blend mode supported by DCE-generation driver. I don't have the hardware to check it out, but looking at the code I understand that, even if the hardware can do PREMULTI or COVERAGE, the way it programs registers doesn't make these other blend modes actually available.
Bearing this in mind, this series is organized as follow:
- Patch 1 fixes the pixel_blend_mode reset when the blend mode property is advertised without PREMULTI support. It's needed for AMD DCE-generation (patch 3) and I think with this we can also remove a workaround in nouveau nv50_wndw_default_state() caused by the hardcoded PREMULTI default value [1].
- Patch 2 fixes the missing-blend-mode-property warning for DCN primary plane and for DCN+DCE cursor plane. The alpha property keeps only on overlay planes because looks like this is not supported by AMD primary planes and it doesn't affect the blend mode warning we are targetting. Enabling alpha properties to primary was also causing -EINVAL on IGT alpha tests because it started testing primary planes and disabling it (which is not allowed by the AMD display driver).
- Patch 3 fixes the warning for DCE primary plane, but I detached this solution from the previous patch because I don't have hardware to validate if my assumption about PIXEL_NONE-only is correct.
[1] https://lore.kernel.org/dri-devel/[email protected]/
[v1]: https://lore.kernel.org/dri-devel/[email protected]/
Changes:
- new patch for pixel_blend_mode default value other than PREMULTI.
- remove DCE11 which supports per_pixel_alpha but doesn't support DCN_UNIVERSAL_PLANE.
- new patch for the primary plane blend mode on DCE-generations
[v2]: https://lore.kernel.org/dri-devel/[email protected]/
Changes:
- keep alpha property only for overlay planes (looks like primary plane doesn't actually support it) - alpha prop in primary planes causes IGT test failures with RX 7900 XT on a 4k60 HP U27 as reported by Daniel Wheeler.
- improve readability of if conditions (Alex H)
- add Leandro's r-b tags.
Best Regards,
Melissa
Melissa Wen (3):
drm/atomic-state-helper: set pixel_blend_mode to prop default on reset drm/amd/display: fix missing blend-mode-prop warning for DCN drm/amd/display: advertise PIXEL_NONE as blend mode for DCE generations
.../amd/display/amdgpu_dm/amdgpu_dm_plane.c | 25 ++++++++++++++++--- drivers/gpu/drm/drm_atomic_state_helper.c | 7 ++++++ 2 files changed, 29 insertions(+), 3 deletions(-)
Signature · 1 line
2.53.0
PATCH v3 1/3 drm/atomic-state-helper: set pixel_blend_mode to prop default on reset
In __drm_atomic_helper_plane_state_init(), pixel_blend_mode is always reset to DRM_MODE_BLEND_PREMULTI. That was consistent while drm_plane_create_blend_mode_property() required PREMULTI in the supported modes, but it now falls back to COVERAGE or PIXEL_NONE when the driver doesn't support PREMULTI. The hardcoded default may therefore not be a blend mode the hardware can do, nor one the property advertises.
Initialize pixel_blend_mode from the blend mode property default instead, keeping DRM_MODE_BLEND_PREMULTI for planes without the property.
--- drivers/gpu/drm/drm_atomic_state_helper.c | 7 +++++++ 1 file changed, 7 insertions(+)
drivers/gpu/drm/drm_atomic_state_helper.c +7 -0
@@ -278,7 +278,14 @@ void __drm_atomic_helper_plane_state_init(struct drm_plane_state *plane_state, plane_state->rotation =DRM_MODE_ROTATE_0 ; plane_state->alpha =DRM_BLEND_ALPHA_OPAQUE ; + plane_state->pixel_blend_mode =DRM_MODE_BLEND_PREMULTI ; +if (plane->blend_mode_property) { +if (!drm_object_property_get_default_value (&plane->base, + plane->blend_mode_property, + &val)) + plane_state->pixel_blend_mode = val; + } if (plane->color_encoding_property) { if (!drm_object_property_get_default_value(&plane->base,
Signature · 1 line
2.53.0
PATCH v3 2/3 drm/amd/display: fix missing blend-mode-prop warning for DCN
In reply to Melissa Wenvalidate_blend_mode_for_alpha_formats() warns when a plane supports formats with alpha but doesn't expose the blend mode property. Fix this by adding the same overlay plane blend modes to primary plane, since they are all universal planes in DCN-generation. Cursor planes support ARGB8888 format and CURSOR_MODE_COLOR_PRE_MULTIPLIED_ALPHA is set by default (other color formats are not implemented), so only expose support to PREMULTI, which is the default blend mode on DRM.
---
v3:
- keep primary planes w/o alpha property
- re-order if-conditions for readability
--- .../amd/display/amdgpu_dm/amdgpu_dm_plane.c | 21 ++++++++++++++++--- 1 file changed, 18 insertions(+), 3 deletions(-)
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +18 -3
@@ -1923,16 +1923,31 @@ int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm,if (res)return res; -if (plane->type ==DRM_PLANE_TYPE_OVERLAY && - plane_cap && plane_cap->per_pixel_alpha) { +/* TODO: Check which blend modes are supported in DCE-generation +* planes, i.e. DC_PLANE_TYPE_DCE_RGB/UNDERLAY and expose blend mode +* property accordingly. +*/ +if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || + plane->type ==DRM_PLANE_TYPE_PRIMARY ) && + plane_cap && plane_cap->per_pixel_alpha && + plane_cap->type ==DC_PLANE_TYPE_DCN_UNIVERSAL ) {unsigned int blend_caps =BIT (DRM_MODE_BLEND_PIXEL_NONE ) |BIT (DRM_MODE_BLEND_PREMULTI ) |BIT (DRM_MODE_BLEND_COVERAGE ); -drm_plane_create_alpha_property (plane);drm_plane_create_blend_mode_property (plane, blend_caps); + +if (plane->type ==DRM_PLANE_TYPE_OVERLAY ) +drm_plane_create_alpha_property (plane); } +/* Cursor color format is set to CURSOR_MODE_COLOR_PRE_MULTIPLIED_ALPHA +* by default, so only advertise DRM_MODE_BLEND_PREMULTI blend mode for +* this type of plane. +*/ +if (plane->type ==DRM_PLANE_TYPE_CURSOR ) +drm_plane_create_blend_mode_property (plane,BIT (DRM_MODE_BLEND_PREMULTI )); +if (plane->type ==DRM_PLANE_TYPE_PRIMARY ) {/* * Allow OVERLAY planes to be used as underlays by assigning an
Signature · 1 line
2.53.0
Re: [PATCH v3 2/3] drm/amd/display: fix missing blend-mode-prop warning for DCN
I can confirm IGT's "kms_plane_alpha_blend --run-subtest alpha-basic" no longer fails with v3.
Quoted 59 lines · Melissa Wen
validate_blend_mode_for_alpha_formats()warns when a plane supports formats with alpha but doesn't expose the blend mode property. Fix this by adding the same overlay plane blend modes to primary plane, since they are all universal planes in DCN-generation. Cursor planes support ARGB8888 format andCURSOR_MODE_COLOR_PRE_MULTIPLIED_ALPHAis set by default (other color formats are not implemented), so only expose support to PREMULTI, which is the default blend mode on DRM.Fixes 860e748bddcc ("drm: ensure blend mode supported if pixel format with alpha exposed")Reviewed-by Leandro Ribeiro <[email protected]>Signed-off-by Melissa Wen <[email protected]>---
v3:
- keep primary planes w/o alpha property
- re-order if-conditions for readability--- .../amd/display/amdgpu_dm/amdgpu_dm_plane.c | 21 ++++++++++++++++--- 1 file changed, 18 insertions(+), 3 deletions(-)drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +18
-3@@ -1923,16 +1923,31 @@ int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm,if (res)return res; -if (plane->type ==DRM_PLANE_TYPE_OVERLAY && - plane_cap && plane_cap->per_pixel_alpha) { +/* TODO: Check which blend modes are supported in DCE-generation +* planes, i.e. DC_PLANE_TYPE_DCE_RGB/UNDERLAY and expose blend mode +* property accordingly. +*/ +if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || + plane->type ==DRM_PLANE_TYPE_PRIMARY ) && + plane_cap && plane_cap->per_pixel_alpha && + plane_cap->type ==DC_PLANE_TYPE_DCN_UNIVERSAL ) {unsigned int blend_caps =BIT (DRM_MODE_BLEND_PIXEL_NONE ) |BIT (DRM_MODE_BLEND_PREMULTI ) |BIT (DRM_MODE_BLEND_COVERAGE ); -drm_plane_create_alpha_property (plane);drm_plane_create_blend_mode_property (plane, blend_caps); + +if (plane->type ==DRM_PLANE_TYPE_OVERLAY ) +drm_plane_create_alpha_property (plane); } +/* Cursor color format is set to CURSOR_MODE_COLOR_PRE_MULTIPLIED_ALPHA +* by default, so only advertise DRM_MODE_BLEND_PREMULTI blend mode for +* this type of plane. +*/ +if (plane->type ==DRM_PLANE_TYPE_CURSOR ) +drm_plane_create_blend_mode_property (plane,BIT (DRM_MODE_BLEND_PREMULTI )); +if (plane->type ==DRM_PLANE_TYPE_PRIMARY ) {/* * Allow OVERLAY planes to be used as underlays by assigning an
Re: [PATCH v3 2/3] drm/amd/display: fix missing blend-mode-prop warning for DCN
In reply to Melissa WenQuoted 59 lines · Melissa Wen
validate_blend_mode_for_alpha_formats()warns when a plane supports formats with alpha but doesn't expose the blend mode property. Fix this by adding the same overlay plane blend modes to primary plane, since they are all universal planes in DCN-generation. Cursor planes support ARGB8888 format andCURSOR_MODE_COLOR_PRE_MULTIPLIED_ALPHAis set by default (other color formats are not implemented), so only expose support to PREMULTI, which is the default blend mode on DRM.Fixes 860e748bddcc ("drm: ensure blend mode supported if pixel format with alpha exposed")Reviewed-by Leandro Ribeiro <[email protected]>Signed-off-by Melissa Wen <[email protected]>---
v3:
- keep primary planes w/o alpha property
- re-order if-conditions for readability--- .../amd/display/amdgpu_dm/amdgpu_dm_plane.c | 21 ++++++++++++++++--- 1 file changed, 18 insertions(+), 3 deletions(-)drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +18
-3@@ -1923,16 +1923,31 @@ int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm,if (res)return res; -if (plane->type ==DRM_PLANE_TYPE_OVERLAY && - plane_cap && plane_cap->per_pixel_alpha) { +/* TODO: Check which blend modes are supported in DCE-generation +* planes, i.e. DC_PLANE_TYPE_DCE_RGB/UNDERLAY and expose blend mode +* property accordingly. +*/ +if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || + plane->type ==DRM_PLANE_TYPE_PRIMARY ) && + plane_cap && plane_cap->per_pixel_alpha && + plane_cap->type ==DC_PLANE_TYPE_DCN_UNIVERSAL ) {unsigned int blend_caps =BIT (DRM_MODE_BLEND_PIXEL_NONE ) |BIT (DRM_MODE_BLEND_PREMULTI ) |BIT (DRM_MODE_BLEND_COVERAGE ); -drm_plane_create_alpha_property (plane);drm_plane_create_blend_mode_property (plane, blend_caps); + +if (plane->type ==DRM_PLANE_TYPE_OVERLAY ) +drm_plane_create_alpha_property (plane); } +/* Cursor color format is set to CURSOR_MODE_COLOR_PRE_MULTIPLIED_ALPHA +* by default, so only advertise DRM_MODE_BLEND_PREMULTI blend mode for +* this type of plane. +*/ +if (plane->type ==DRM_PLANE_TYPE_CURSOR ) +drm_plane_create_blend_mode_property (plane,BIT (DRM_MODE_BLEND_PREMULTI )); +if (plane->type ==DRM_PLANE_TYPE_PRIMARY ) {/* * Allow OVERLAY planes to be used as underlays by assigning an
PATCH v3 3/3 drm/amd/display: advertise PIXEL_NONE as blend mode for DCE generations
In reply to Melissa WenDCE can support PREMULTI and COVERAGE blend mode depending on its generation, however current driver implementation either doesn't expose more than primary and cursor plane, or doesn't program registers for any blend mode other than PIXEL_NONE. To fix the missing-blend-mode-prop warning according to current DCE plane caps, create blend mode property with PIXEL_NONE only for primary planes, instead of the DRM default PREMULTI.
---
v3:
- fix indentation (Alex H)
--- .../amd/display/amdgpu_dm/amdgpu_dm_plane.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-)
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +11 -7
@@ -1923,14 +1923,18 @@ int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm,if (res)return res; -/* TODO: Check which blend modes are supported in DCE-generation -* planes, i.e. DC_PLANE_TYPE_DCE_RGB/UNDERLAY and expose blend mode -* property accordingly. +/* Blend mode support varies on DCE generations according to HW caps +* and number of planes per CRTC. However, as current driver +* implementation only creates one primary and one cursor plane per +* CRTC for DCE (overlay is only created if +* DC_PLANE_TYPE_DCN_UNIVERSAL), the primary plane blend mode ends up +* being always PIXEL_NONE across DCE versions. */ -if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || - plane->type ==DRM_PLANE_TYPE_PRIMARY ) && - plane_cap && plane_cap->per_pixel_alpha && - plane_cap->type ==DC_PLANE_TYPE_DCN_UNIVERSAL ) { +if (plane_cap && plane_cap->type !=DC_PLANE_TYPE_DCN_UNIVERSAL ) { +drm_plane_create_blend_mode_property (plane,BIT (DRM_MODE_BLEND_PIXEL_NONE )); + }else if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || + plane->type ==DRM_PLANE_TYPE_PRIMARY ) && + plane_cap && plane_cap->per_pixel_alpha) {unsigned int blend_caps =BIT (DRM_MODE_BLEND_PIXEL_NONE ) |BIT (DRM_MODE_BLEND_PREMULTI ) |BIT (DRM_MODE_BLEND_COVERAGE );
Signature · 1 line
2.53.0
[email protected]
AI reviewRe: [PATCH v3 3/3] drm/amd/display: advertise PIXEL_NONE as blend mode for DCE generations
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [High] Calling drm_plane_create_blend_mode_property() without PREMULTI support triggers a kernel warning…
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Calling drm_plane_create_blend_mode_property() without PREMULTI support triggers a kernel warning (WARN_ON) and fails to create the property. --
--- Patch [3]: [PATCH v3 3/3] drm/amd/display: advertise PIXEL_NONE as blend mode for DCE generations --- commit 44681caffc8bcfc7085bcfe1ec288a0eafc8c992
Author: Melissa Wen <[email protected]>
drm/amd/display: advertise PIXEL_NONE as blend mode for DCE generations
This commit changes the blend mode property creation for primary planes on DCE generations. It modifies the driver to create the blend mode property with PIXEL_NONE only, instead of the DRM default PREMULTI mode.
Quoted 4 lines
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +0
-0
[ ... ]
Quoted 1 line
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +0
-0@@ -2208,14 +2208,18 @@ int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm,
[ ... ]
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm, · line 1934
Quoted 6 lines
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +2
-4-if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || - plane->type ==DRM_PLANE_TYPE_PRIMARY ) && - plane_cap && plane_cap->per_pixel_alpha && - plane_cap->type ==DC_PLANE_TYPE_DCN_UNIVERSAL ) { +if (plane_cap && plane_cap->type !=DC_PLANE_TYPE_DCN_UNIVERSAL ) { +drm_plane_create_blend_mode_property (plane,BIT (DRM_MODE_BLEND_PIXEL_NONE ));
[Severity: High]
Does this trigger a kernel warning and fail to create the property?
Looking at drm_plane_create_blend_mode_property() in drm_blend.c, the DRM core explicitly enforces that DRM_MODE_BLEND_PREMULTI must be supported:
int drm_plane_create_blend_mode_property(struct drm_plane *plane,
unsigned int supported_modes)
{
...
if (WARN_ON((supported_modes & ~valid_mode_mask) ||
((supported_modes & BIT(DRM_MODE_BLEND_PREMULTI)) == 0)))
return -EINVAL;
...
}Since amdgpu_dm_plane_init() now passes only BIT(DRM_MODE_BLEND_PIXEL_NONE), this will trigger the WARN_ON and return -EINVAL during device probe, meaning the property is not created at all.
Quoted 6 lines
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +3
-0+ }else if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || + plane->type ==DRM_PLANE_TYPE_PRIMARY ) && + plane_cap && plane_cap->per_pixel_alpha) {unsigned int blend_caps =BIT (DRM_MODE_BLEND_PIXEL_NONE ) |BIT (DRM_MODE_BLEND_PREMULTI ) |BIT (DRM_MODE_BLEND_COVERAGE );
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=3
Re: [PATCH v3 3/3] drm/amd/display: advertise PIXEL_NONE as blend mode for DCE generations
Quoted 43 lines · On 13/08/2026 16:44, [email protected] wrote:
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Callingdrm_plane_create_blend_mode_property()without PREMULTI support triggers a kernel warning (WARN_ON) and fails to create the property. ----- Patch [3]: [PATCH v3 3/3] drm/amd/display: advertise
PIXEL_NONEas blend mode for DCE generations --- commit44681caffc8bcfc7085bcfe1ec288a0eafc8c992
Author: Melissa Wen <[email protected]>drm/amd/display: advertise
PIXEL_NONEas blend mode for DCE generationsThis commit changes the blend mode property creation for primary planes on DCE generations. It modifies the driver to create the blend mode property with
PIXEL_NONEonly, instead of the DRM default PREMULTI mode.Quoted 4 lines
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +0
-0[ ... ]
Quoted 1 line
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +0
-0@@ -2208,14 +2208,18 @@ int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm,[ ... ]
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm, · line 1934
Quoted 6 lines
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +2
-4-if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || - plane->type ==DRM_PLANE_TYPE_PRIMARY ) && - plane_cap && plane_cap->per_pixel_alpha && - plane_cap->type ==DC_PLANE_TYPE_DCN_UNIVERSAL ) { +if (plane_cap && plane_cap->type !=DC_PLANE_TYPE_DCN_UNIVERSAL ) { +drm_plane_create_blend_mode_property (plane,BIT (DRM_MODE_BLEND_PIXEL_NONE ));[Severity: High]
Does this trigger a kernel warning and fail to create the property?Looking at
drm_plane_create_blend_mode_property()indrm_blend.c, the DRM core explicitly enforces thatDRM_MODE_BLEND_PREMULTImust be supported:int drm_plane_create_blend_mode_property(struct
drm_plane*plane,
unsigned intsupported_modes){ ... if (WARN_ON((supported_modes & ~valid_mode_mask) || ((supported_modes & BIT(DRM_MODE_BLEND_PREMULTI)) == 0))) return -EINVAL; ... }
Quoted 3 lines
Since
amdgpu_dm_plane_init()now passes onlyBIT(DRM_MODE_BLEND_PIXEL_NONE), this will trigger theWARN_ONand return-EINVALduring device probe, meaning the property is not created at all.
No, you are probably looking at an outdated codebase.
Other blend modes are supported since:
9813e158d13d5 ("drm/drm_blend: allow blend mode property without PREMULTI") in drm-misc-next.
Melissa
Quoted 7 lines
Quoted 7 lines
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +3
-0+ }else if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || + plane->type ==DRM_PLANE_TYPE_PRIMARY ) && + plane_cap && plane_cap->per_pixel_alpha) {unsigned int blend_caps =BIT (DRM_MODE_BLEND_PIXEL_NONE ) |BIT (DRM_MODE_BLEND_PREMULTI ) |BIT (DRM_MODE_BLEND_COVERAGE );
Re: [PATCH v3 3/3] drm/amd/display: advertise PIXEL_NONE as blend mode for DCE generations
In reply to Melissa WenQuoted 49 lines · Melissa Wen
DCE can support PREMULTI and COVERAGE blend mode depending on its generation, however current driver implementation either doesn't expose more than primary and cursor plane, or doesn't program registers for any blend mode other than
PIXEL_NONE. To fix the missing-blend-mode-prop warning according to current DCE plane caps, create blend mode property withPIXEL_NONEonly for primary planes, instead of the DRM default PREMULTI.Fixes 860e748bddcc ("drm: ensure blend mode supported if pixel format with alpha exposed")Reviewed-by Leandro Ribeiro <[email protected]>Signed-off-by Melissa Wen <[email protected]>---
v3:
- fix indentation (Alex H)--- .../amd/display/amdgpu_dm/amdgpu_dm_plane.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-)drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +11
-7@@ -1923,14 +1923,18 @@ int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm,if (res)return res; -/* TODO: Check which blend modes are supported in DCE-generation -* planes, i.e. DC_PLANE_TYPE_DCE_RGB/UNDERLAY and expose blend mode -* property accordingly. +/* Blend mode support varies on DCE generations according to HW caps +* and number of planes per CRTC. However, as current driver +* implementation only creates one primary and one cursor plane per +* CRTC for DCE (overlay is only created if +* DC_PLANE_TYPE_DCN_UNIVERSAL), the primary plane blend mode ends up +* being always PIXEL_NONE across DCE versions. */ -if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || - plane->type ==DRM_PLANE_TYPE_PRIMARY ) && - plane_cap && plane_cap->per_pixel_alpha && - plane_cap->type ==DC_PLANE_TYPE_DCN_UNIVERSAL ) { +if (plane_cap && plane_cap->type !=DC_PLANE_TYPE_DCN_UNIVERSAL ) { +drm_plane_create_blend_mode_property (plane,BIT (DRM_MODE_BLEND_PIXEL_NONE )); + }else if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || + plane->type ==DRM_PLANE_TYPE_PRIMARY ) && + plane_cap && plane_cap->per_pixel_alpha) {unsigned int blend_caps =BIT (DRM_MODE_BLEND_PIXEL_NONE ) |BIT (DRM_MODE_BLEND_PREMULTI ) |BIT (DRM_MODE_BLEND_COVERAGE );
Re: [PATCH v3 3/3] drm/amd/display: advertise PIXEL_NONE as blend mode for DCE generations
In reply to Melissa Wen[ Adding Timur ]
Quoted 39 lines · Melissa Wen
DCE can support PREMULTI and COVERAGE blend mode depending on its generation, however current driver implementation either doesn't expose more than primary and cursor plane, or doesn't program registers for any blend mode other than
PIXEL_NONE. To fix the missing-blend-mode-prop warning according to current DCE plane caps, create blend mode property withPIXEL_NONEonly for primary planes, instead of the DRM default PREMULTI.Fixes 860e748bddcc ("drm: ensure blend mode supported if pixel format with alpha exposed")Reviewed-by Leandro Ribeiro <[email protected]>Signed-off-by Melissa Wen <[email protected]>---
v3:
- fix indentation (Alex H)--- .../amd/display/amdgpu_dm/amdgpu_dm_plane.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-)drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +6
-5@@ -1923,14 +1923,18 @@ int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm,if (res)return res; -/* TODO: Check which blend modes are supported in DCE-generation -* planes, i.e. DC_PLANE_TYPE_DCE_RGB/UNDERLAY and expose blend mode -* property accordingly. +/* Blend mode support varies on DCE generations according to HW caps +* and number of planes per CRTC. However, as current driver +* implementation only creates one primary and one cursor plane per +* CRTC for DCE (overlay is only created if +* DC_PLANE_TYPE_DCN_UNIVERSAL), the primary plane blend mode ends up +* being always PIXEL_NONE across DCE versions. */ - if ((plane->type == DRM_PLANE_TYPE_OVERLAY || - plane->type == DRM_PLANE_TYPE_PRIMARY) &&
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm, · line 1940
Quoted 10 lines
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +5
-2- plane_cap && plane_cap->per_pixel_alpha && - plane_cap->type == DC_PLANE_TYPE_DCN_UNIVERSAL) { +if (plane_cap && plane_cap->type !=DC_PLANE_TYPE_DCN_UNIVERSAL ) { +drm_plane_create_blend_mode_property (plane,BIT (DRM_MODE_BLEND_PIXEL_NONE )); + }else if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || + plane->type ==DRM_PLANE_TYPE_PRIMARY ) && + plane_cap && plane_cap->per_pixel_alpha) {unsigned int blend_caps =BIT (DRM_MODE_BLEND_PIXEL_NONE ) |BIT (DRM_MODE_BLEND_PREMULTI ) |BIT (DRM_MODE_BLEND_COVERAGE );
With this patch applied to the mainline kernel, my display freezes when Kwin is started (either by SDDM or by launching a KDE Plasma session). GDM and Gnome work fine. The journal doesn't have much information, it's mainly this:
kwin_wayland[8402]: Applying output configuration failed!
and after that several KDE processes with:
<kde_process>[xxxx]: There are no outputs - creating placeholder screen
Any idea how I can debug this further? Maybe ask the KDE developers?
With only the first two patches of this series applied, Kwin works as before, but I get the warnings of course.
Viktor
Re: [PATCH v3 3/3] drm/amd/display: advertise PIXEL_NONE as blend mode for DCE generations
Quoted 60 lines · Viktor Jägersküpper
[ Adding Timur ]
Quoted 39 lines · Melissa Wen
DCE can support PREMULTI and COVERAGE blend mode depending on its generation, however current driver implementation either doesn't expose more than primary and cursor plane, or doesn't program registers for any blend mode other than
PIXEL_NONE. To fix the missing-blend-mode-prop warning according to current DCE plane caps, create blend mode property withPIXEL_NONEonly for primary planes, instead of the DRM default PREMULTI.Fixes 860e748bddcc ("drm: ensure blend mode supported if pixel format with alpha exposed")Reviewed-by Leandro Ribeiro <[email protected]>Signed-off-by Melissa Wen <[email protected]>---
v3:
- fix indentation (Alex H)--- .../amd/display/amdgpu_dm/amdgpu_dm_plane.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-)drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +6
-5@@ -1923,14 +1923,18 @@ int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm,if (res)return res; -/* TODO: Check which blend modes are supported in DCE-generation -* planes, i.e. DC_PLANE_TYPE_DCE_RGB/UNDERLAY and expose blend mode -* property accordingly. +/* Blend mode support varies on DCE generations according to HW caps +* and number of planes per CRTC. However, as current driver +* implementation only creates one primary and one cursor plane per +* CRTC for DCE (overlay is only created if +* DC_PLANE_TYPE_DCN_UNIVERSAL), the primary plane blend mode ends up +* being always PIXEL_NONE across DCE versions. */ - if ((plane->type == DRM_PLANE_TYPE_OVERLAY || - plane->type == DRM_PLANE_TYPE_PRIMARY) &&drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm, · line 1940
Quoted 10 lines
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +5
-2- plane_cap && plane_cap->per_pixel_alpha && - plane_cap->type == DC_PLANE_TYPE_DCN_UNIVERSAL) { +if (plane_cap && plane_cap->type !=DC_PLANE_TYPE_DCN_UNIVERSAL ) { +drm_plane_create_blend_mode_property (plane,BIT (DRM_MODE_BLEND_PIXEL_NONE )); + }else if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || + plane->type ==DRM_PLANE_TYPE_PRIMARY ) && + plane_cap && plane_cap->per_pixel_alpha) {unsigned int blend_caps =BIT (DRM_MODE_BLEND_PIXEL_NONE ) |BIT (DRM_MODE_BLEND_PREMULTI ) |BIT (DRM_MODE_BLEND_COVERAGE );With this patch applied to the mainline kernel, my display freezes when Kwin is started (either by SDDM or by launching a KDE Plasma session). GDM and Gnome work fine. The journal doesn't have much information, it's mainly this:
kwin_wayland[8402]: Applying output configuration failed!
Quoted 8 lines
and after that several KDE processes with:
<kde_process>[xxxx]: There are no outputs - creating placeholder screen
Any idea how I can debug this further? Maybe ask the KDE developers?
With only the first two patches of this series applied, Kwin works as before, but I get the warnings of course.
I completely forgot that I'm observing this with AMD Pitcairn running KDE 6.7.4 on Arch Linux. The kernel is built from source (mainline as of today) with this patch series applied on top.
Viktor
Melissa Wen
Reviewed-by Leandro RibeiroRe: [PATCH v3 3/3] drm/amd/display: advertise PIXEL_NONE as blend mode for DCE generations
Quoted 49 lines · Viktor Jägersküpper
Quoted 48 lines · Viktor Jägersküpper
[ Adding Timur ]
Quoted 45 lines · Melissa Wen
DCE can support PREMULTI and COVERAGE blend mode depending on its generation, however current driver implementation either doesn't expose more than primary and cursor plane, or doesn't program registers for any blend mode other than
PIXEL_NONE. To fix the missing-blend-mode-prop warning according to current DCE plane caps, create blend mode property withPIXEL_NONEonly for primary planes, instead of the DRM default PREMULTI.Fixes 860e748bddcc ("drm: ensure blend mode supported if pixel format with alpha exposed")Reviewed-by Leandro Ribeiro <[email protected]>Signed-off-by Melissa Wen <[email protected]>---
v3:
- fix indentation (Alex H)--- .../amd/display/amdgpu_dm/amdgpu_dm_plane.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-)drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +10
-7@@ -1923,14 +1923,18 @@ int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm,if (res)return res; -/* TODO: Check which blend modes are supported in DCE-generation -* planes, i.e. DC_PLANE_TYPE_DCE_RGB/UNDERLAY and expose blend mode -* property accordingly. +/* Blend mode support varies on DCE generations according to HW caps +* and number of planes per CRTC. However, as current driver +* implementation only creates one primary and one cursor plane per +* CRTC for DCE (overlay is only created if +* DC_PLANE_TYPE_DCN_UNIVERSAL), the primary plane blend mode ends up +* being always PIXEL_NONE across DCE versions. */ -if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || - plane->type ==DRM_PLANE_TYPE_PRIMARY ) && - plane_cap && plane_cap->per_pixel_alpha && - plane_cap->type ==DC_PLANE_TYPE_DCN_UNIVERSAL ) { +if (plane_cap && plane_cap->type !=DC_PLANE_TYPE_DCN_UNIVERSAL ) { +drm_plane_create_blend_mode_property (plane,BIT (DRM_MODE_BLEND_PIXEL_NONE )); + } else if ((plane->type == DRM_PLANE_TYPE_OVERLAY || + plane->type == DRM_PLANE_TYPE_PRIMARY) &&
Quoted 8 lines
Quoted 8 lines
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm, · line 1940
Quoted 4 lines
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +1
-0+ plane_cap && plane_cap->per_pixel_alpha) {unsigned int blend_caps =BIT (DRM_MODE_BLEND_PIXEL_NONE ) |BIT (DRM_MODE_BLEND_PREMULTI ) |BIT (DRM_MODE_BLEND_COVERAGE );With this patch applied to the mainline kernel, my display freezes when Kwin is started (either by SDDM or by launching a KDE Plasma session). GDM and Gnome work fine. The journal doesn't have much information, it's mainly this:
I suspect KDE is not interpreting correctly the blend mode, but I think we can keep it working as before by adding PREMULTI. In the end there is no overlay plane for DCE, and PIXEL_NONE and PREMULTI are equivalent when one primary is composed on top of a black background.
Can you replace this commit by the patch attached?
It might solve the issue as long as background continues being black. I didn't apply it to drm-misc, so let me know how does this new patch works on your side.
/cc Xaver to take a look at the KDE part.
Melissa
Quoted 9 lines
Quoted 9 lines
kwin_wayland[8402]: Applying output configuration failed!
and after that several KDE processes with:
<kde_process>[xxxx]: There are no outputs - creating placeholder screen
Any idea how I can debug this further? Maybe ask the KDE developers?
Quoted 7 lines
Quoted 2 lines
With only the first two patches of this series applied, Kwin works as before, but I get the warnings of course.
I completely forgot that I'm observing this with AMD Pitcairn running KDE 6.7.4 on Arch Linux. The kernel is built from source (mainline as of today) with this patch series applied on top.
Viktor
From 9d097db70ee2bcfa35d307673c91c39f9dcad5f2 Mon Sep 17 00:00:00 2001
From: Melissa Wen <[email protected]>
Date: Tue, 4 Aug 2026 12:44:28 +0200
Subject: [PATCH] drm/amd/display: advertise PIXEL_NONE and PREMULTI blend mode
for DCE
DCE can support PREMULTI and COVERAGE blend mode depending on its generation, however current driver implementation either doesn't expose more than primary and cursor plane, or doesn't program registers for any blend mode other than PIXEL_NONE. To fix the missing-blend-mode-prop warning according to current DCE plane caps, create blend mode property with PIXEL_NONE and PREMULTI for primary planes. As long as the background is black and there is no overlay plane, PIXEL_NONE and PREMULTI are equivalent, and PREMULTI has been the mandatory/default mode for years, so keep it to avoid regressions.
---
v3:
- fix indentation (Alex H)
v4:
- add PREMULTI to avoid regression (reported by Viktor)
--- .../amd/display/amdgpu_dm/amdgpu_dm_plane.c | 24 +++++++++++++------ 1 file changed, 17 insertions(+), 7 deletions(-)
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +17 -7
@@ -1923,14 +1923,24 @@ int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm,if (res)return res; -/* TODO: Check which blend modes are supported in DCE-generation -* planes, i.e. DC_PLANE_TYPE_DCE_RGB/UNDERLAY and expose blend mode -* property accordingly. +/* Blend mode support varies on DCE generations according to HW caps +* and number of planes per CRTC. However, as current driver +* implementation only creates one primary and one cursor plane per +* CRTC for DCE (overlay is only created if +* DC_PLANE_TYPE_DCN_UNIVERSAL), the primary plane blend mode is +* ignored across DCE versions. Keep PREMULTI to avoid uAPI +* regressions: it was the default/mandatory mode for many years and, +* with no overlay plane, primary composes on top of a black +* background, where PREMULTI and PIXEL_NONE are equivalent. */ -if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || - plane->type ==DRM_PLANE_TYPE_PRIMARY ) && - plane_cap && plane_cap->per_pixel_alpha && - plane_cap->type ==DC_PLANE_TYPE_DCN_UNIVERSAL ) { +if (plane_cap && plane_cap->type !=DC_PLANE_TYPE_DCN_UNIVERSAL ) { +unsigned int blend_caps =BIT (DRM_MODE_BLEND_PIXEL_NONE ) | +BIT (DRM_MODE_BLEND_PREMULTI ); + +drm_plane_create_blend_mode_property (plane, blend_caps); + }else if ((plane->type ==DRM_PLANE_TYPE_OVERLAY || + plane->type ==DRM_PLANE_TYPE_PRIMARY ) && + plane_cap && plane_cap->per_pixel_alpha) {unsigned int blend_caps =BIT (DRM_MODE_BLEND_PIXEL_NONE ) |BIT (DRM_MODE_BLEND_PREMULTI ) |BIT (DRM_MODE_BLEND_COVERAGE );
Signature · 1 line
2.53.0
Re: [PATCH v3 3/3] drm/amd/display: advertise PIXEL_NONE as blend mode for DCE generations
Quoted 1 line
I suspect KDE is not interpreting correctly the blend mode
Yeah, KWin made the assumption that pre-multiplied is always supported, since the documentation said so in the past. Thanks for notifying me, I'll have that fixed in the next bugfix update.
Quoted 1 line
I think we can keep it working as before by adding PREMULTI. In the end there is no overlay plane for DCE, and
PIXEL_NONEand PREMULTI are equivalent when one primary is composed on top of a black background.
Makes sense to me.
- Xaver
Viktor Jägersküpper
Tested-byRe: [PATCH v3 3/3] drm/amd/display: advertise PIXEL_NONE as blend mode for DCE generations
In reply to Melissa WenQuoted 77 lines · Melissa Wen
Quoted 58 lines · Viktor Jägersküpper
Quoted 57 lines · Viktor Jägersküpper
[ Adding Timur ]
Quoted 54 lines · Melissa Wen
DCE can support PREMULTI and COVERAGE blend mode depending on its generation, however current driver implementation either doesn't expose more than primary and cursor plane, or doesn't program registers for any
blend mode other thanPIXEL_NONE. To fix the missing-blend-mode-prop warning according to current DCE plane caps, create blend mode property withPIXEL_NONEonly for primary planes, instead of the DRM default PREMULTI.Fixes 860e748bddcc ("drm: ensure blend mode supported if pixelformat with alpha exposed")
Reviewed-by Leandro Ribeiro <[email protected]>Signed-off-by Melissa Wen <[email protected]>---
v3:
- fix indentation (Alex H)
---
.../amd/display/amdgpu_dm/amdgpu_dm_plane.c | 18 ++++++++++
+-------1 file changed, 11 insertions(+), 7 deletions(-)drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +0
-0@@ -1923,14 +1923,18 @@ int amdgpu_dm_plane_init(structamdgpu_display_manager *dm, if (res) return res;- /* TODO: Check which blend modes are supported in DCE-generation - * planes, i.e. DC_PLANE_TYPE_DCE_RGB/UNDERLAY and expose blend mode
- * property accordingly.
+ /* Blend mode support varies on DCE generations according to HW caps
+ * and number of planes per CRTC. However, as current driver + * implementation only creates one primary and one cursor plane per
+ * CRTC for DCE (overlay is only created if
+ *DC_PLANE_TYPE_DCN_UNIVERSAL), the primary plane blend mode ends up
+ * being alwaysPIXEL_NONEacross DCE versions.
*/
- if ((plane->type==DRM_PLANE_TYPE_OVERLAY||
-plane->type==DRM_PLANE_TYPE_PRIMARY) &&
-plane_cap&&plane_cap->per_pixel_alpha&&
-plane_cap->type==DC_PLANE_TYPE_DCN_UNIVERSAL) {
+ if (plane_cap&&plane_cap->type!=DC_PLANE_TYPE_DCN_UNIVERSAL) { +drm_plane_create_blend_mode_property(plane, BIT(DRM_MODE_BLEND_PIXEL_NONE));
+ } else if ((plane->type==DRM_PLANE_TYPE_OVERLAY||
+plane->type==DRM_PLANE_TYPE_PRIMARY) &&Quoted 8 lines
Quoted 8 lines
Quoted 4 lines
+
plane_cap&&plane_cap->per_pixel_alpha) { unsigned intblend_caps=BIT(DRM_MODE_BLEND_PIXEL_NONE)|BIT(DRM_MODE_BLEND_PREMULTI)|BIT(DRM_MODE_BLEND_COVERAGE);With this patch applied to the mainline kernel, my display freezes when Kwin is started (either by SDDM or by launching a KDE Plasma session). GDM and Gnome work fine. The journal doesn't have much information, it's mainly this:
I suspect KDE is not interpreting correctly the blend mode, but I think we can
keep it working as before by adding PREMULTI. In the end there is no overlay
plane for DCE, andPIXEL_NONEand PREMULTI are equivalent when one primary is composed on top of a black background.
Quoted 4 lines
Can you replace this commit by the patch attached?
It might solve the issue as long as background continues being black. I didn't apply it to drm-misc, so let me know how does this new patch works on your side.
The attached patch solves the issue, I tested both KDE/SDDM and Gnome/GDM, the warnings are gone as expected. You can add:
Thanks,
Viktor
Public
Thanks for the fixes Melissa, we'll re-add them to next week's promotion and re-test.
Thank you,
Dan Wheeler
Sr. Technologist | AMD
SW Display
------------------------------------------------------------------------------------------------------------------
1 Commerce Valley Dr E, Thornhill, ON L3T 7X6
amd.com
-----Original Message-----
From: Melissa Wen <[email protected]>
Sent: Thursday, August 13, 2026 10:30 AM
To: [email protected]; Deucher, Alexander <[email protected]>; Koenig, Christian <[email protected]>; [email protected]; Wentland, Harry <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; Li, Sun peng (Leo) <[email protected]>; [email protected]
Cc: Hung, Alex <[email protected]>; Wheeler, Daniel <[email protected]>; [email protected]; Lyude Paul <[email protected]>; [email protected]; [email protected]
Subject: [PATCH v3 0/3] drm/atomic-state-helper: fixes for blend-mode-prop warning
PREMULTI was the default DRM blend mode until 9813e158d13d
("drm/drm_blend: allow blend mode property without PREMULTI") introduced the possibility of exposing the blend mode property without PREMULTI being supported. However, __drm_atomic_helper_plane_state_init() still resets pixel_blend_mode to PREMULTI (hardcoded), ignoring the supported mode in this property. In the same series, 860e748bddcc ("drm: ensure blend mode supported if pixel format with alpha exposed") starts warning drivers that supports alpha formats in a given planes but doesn't support blend mode property, which includes AMD primary and cursor planes.
For AMD DCN families, overlay and primary planes support the three blend modes, and therefore PREMULTI is still the default mode. So the warning can be fixed by just creating blend mode properties for primary planes too. PREMULTI is the default/unique mode for cursor planes in DCN and DCE driver. However, looks like PIXEL_NONE is the unique blend mode supported by DCE-generation driver. I don't have the hardware to check it out, but looking at the code I understand that, even if the hardware can do PREMULTI or COVERAGE, the way it programs registers doesn't make these other blend modes actually available.
Bearing this in mind, this series is organized as follow:
- Patch 1 fixes the pixel_blend_mode reset when the blend mode property is advertised without PREMULTI support. It's needed for AMD DCE-generation (patch 3) and I think with this we can also remove a workaround in nouveau nv50_wndw_default_state() caused by the hardcoded PREMULTI default value [1].
- Patch 2 fixes the missing-blend-mode-property warning for DCN primary plane and for DCN+DCE cursor plane. The alpha property keeps only on overlay planes because looks like this is not supported by AMD primary planes and it doesn't affect the blend mode warning we are targetting. Enabling alpha properties to primary was also causing -EINVAL on IGT alpha tests because it started testing primary planes and disabling it (which is not allowed by the AMD display driver).
- Patch 3 fixes the warning for DCE primary plane, but I detached this solution from the previous patch because I don't have hardware to validate if my assumption about PIXEL_NONE-only is correct.
[1] https://lore.kernel.org/dri-devel/[email protected]/
[v1]: https://lore.kernel.org/dri-devel/[email protected]/
Changes:
- new patch for pixel_blend_mode default value other than PREMULTI.
- remove DCE11 which supports per_pixel_alpha but doesn't support DCN_UNIVERSAL_PLANE.
- new patch for the primary plane blend mode on DCE-generations
[v2]: https://lore.kernel.org/dri-devel/[email protected]/
Changes:
- keep alpha property only for overlay planes (looks like primary plane doesn't actually support it) - alpha prop in primary planes causes IGT test failures with RX 7900 XT on a 4k60 HP U27 as reported by Daniel Wheeler.
- improve readability of if conditions (Alex H)
- add Leandro's r-b tags.
Best Regards,
Melissa
Melissa Wen (3):
drm/atomic-state-helper: set pixel_blend_mode to prop default on reset drm/amd/display: fix missing blend-mode-prop warning for DCN drm/amd/display: advertise PIXEL_NONE as blend mode for DCE generations
.../amd/display/amdgpu_dm/amdgpu_dm_plane.c | 25 ++++++++++++++++--- drivers/gpu/drm/drm_atomic_state_helper.c | 7 ++++++ 2 files changed, 29 insertions(+), 3 deletions(-)
--
2.53.0
Quoted 3 lines · Daniel
Public
Thanks for the fixes Melissa, we'll re-add them to next week's promotion and re-test.
Nice!
Thank you!
Melissa
Quoted 76 lines
Thank you,
Dan Wheeler
Sr. Technologist | AMD
SW Display
------------------------------------------------------------------------------------------------------------------
1 Commerce Valley Dr E, Thornhill, ON L3T 7X6
amd.com-----Original Message-----
From: Melissa Wen <[email protected]>
Sent: Thursday, August 13, 2026 10:30 AM
To: [email protected]; Deucher, Alexander <[email protected]>; Koenig, Christian <[email protected]>; [email protected]; Wentland, Harry <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; Li, Sun peng (Leo) <[email protected]>; [email protected]
Cc: Hung, Alex <[email protected]>; Wheeler, Daniel <[email protected]>; [email protected]; Lyude Paul <[email protected]>; [email protected]; [email protected]
Subject: [PATCH v3 0/3] drm/atomic-state-helper: fixes for blend-mode-prop warningPREMULTI was the default DRM blend mode until
9813e158d13d
("drm/drm_blend: allow blend mode property without PREMULTI") introduced the possibility of exposing the blend mode property without PREMULTI being supported. However,__drm_atomic_helper_plane_state_init()still resetspixel_blend_modeto PREMULTI (hardcoded), ignoring the supported mode in this property. In the same series,860e748bddcc("drm: ensure blend mode supported if pixel format with alpha exposed") starts warning drivers that supports alpha formats in a given planes but doesn't support blend mode property, which includes AMD primary and cursor planes.For AMD DCN families, overlay and primary planes support the three blend modes, and therefore PREMULTI is still the default mode. So the warning can be fixed by just creating blend mode properties for primary planes too. PREMULTI is the default/unique mode for cursor planes in DCN and DCE driver. However, looks like
PIXEL_NONEis the unique blend mode supported by DCE-generation driver. I don't have the hardware to check it out, but looking at the code I understand that, even if the hardware can do PREMULTI or COVERAGE, the way it programs registers doesn't make these other blend modes actually available.Bearing this in mind, this series is organized as follow:
- Patch 1 fixes the
pixel_blend_modereset when the blend mode property is advertised without PREMULTI support. It's needed for AMD DCE-generation (patch 3) and I think with this we can also remove a workaround in nouveaunv50_wndw_default_state()caused by the hardcoded PREMULTI default value [1].- Patch 2 fixes the missing-blend-mode-property warning for DCN primary plane and for DCN+DCE cursor plane. The alpha property keeps only on overlay planes because looks like this is not supported by AMD primary planes and it doesn't affect the blend mode warning we are targetting. Enabling alpha properties to primary was also causing
-EINVALon IGT alpha tests because it started testing primary planes and disabling it (which is not allowed by the AMD display driver).- Patch 3 fixes the warning for DCE primary plane, but I detached this solution from the previous patch because I don't have hardware to validate if my assumption about PIXEL_NONE-only is correct.
[1] https://lore.kernel.org/dri-devel/[email protected]/
[v1]: https://lore.kernel.org/dri-devel/[email protected]/
Changes:
- new patch forpixel_blend_modedefault value other than PREMULTI.
- remove DCE11 which supportsper_pixel_alphabut doesn't supportDCN_UNIVERSAL_PLANE.
- new patch for the primary plane blend mode on DCE-generations[v2]: https://lore.kernel.org/dri-devel/[email protected]/
Changes:
- keep alpha property only for overlay planes (looks like primary plane doesn't actually support it) - alpha prop in primary planes causes IGT test failures with RX 7900 XT on a 4k60 HP U27 as reported by Daniel Wheeler.
- improve readability of if conditions (Alex H)
- add Leandro's r-b tags.Best Regards,
Melissa
Melissa Wen (3):
drm/atomic-state-helper: setpixel_blend_modeto prop default on reset drm/amd/display: fix missing blend-mode-prop warning for DCN drm/amd/display: advertisePIXEL_NONEas blend mode for DCE generations.../amd/display/amdgpu_dm/amdgpu_dm_plane.c | 25 ++++++++++++++++--- drivers/gpu/drm/drm_atomic_state_helper.c | 7 ++++++ 2 files changed, 29 insertions(+), 3 deletions(-)--
2.53.0
Public
Hi Melissa,
The updated patches passed all my testing this week, so the tested-by can be added if you'd like.
Thank you,
Dan Wheeler
Sr. Technologist | AMD
SW Display
------------------------------------------------------------------------------------------------------------------
1 Commerce Valley Dr E, Thornhill, ON L3T 7X6
amd.com
-----Original Message-----
From: Melissa Wen <[email protected]>
Sent: Thursday, August 13, 2026 10:30 AM
To: [email protected]; Deucher, Alexander <[email protected]>; Koenig, Christian <[email protected]>; [email protected]; Wentland, Harry <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; Li, Sun peng (Leo) <[email protected]>; [email protected]
Cc: Hung, Alex <[email protected]>; Wheeler, Daniel <[email protected]>; [email protected]; Lyude Paul <[email protected]>; [email protected]; [email protected]
Subject: [PATCH v3 0/3] drm/atomic-state-helper: fixes for blend-mode-prop warning
PREMULTI was the default DRM blend mode until 9813e158d13d
("drm/drm_blend: allow blend mode property without PREMULTI") introduced the possibility of exposing the blend mode property without PREMULTI being supported. However, __drm_atomic_helper_plane_state_init() still resets pixel_blend_mode to PREMULTI (hardcoded), ignoring the supported mode in this property. In the same series, 860e748bddcc ("drm: ensure blend mode supported if pixel format with alpha exposed") starts warning drivers that supports alpha formats in a given planes but doesn't support blend mode property, which includes AMD primary and cursor planes.
For AMD DCN families, overlay and primary planes support the three blend modes, and therefore PREMULTI is still the default mode. So the warning can be fixed by just creating blend mode properties for primary planes too. PREMULTI is the default/unique mode for cursor planes in DCN and DCE driver. However, looks like PIXEL_NONE is the unique blend mode supported by DCE-generation driver. I don't have the hardware to check it out, but looking at the code I understand that, even if the hardware can do PREMULTI or COVERAGE, the way it programs registers doesn't make these other blend modes actually available.
Bearing this in mind, this series is organized as follow:
- Patch 1 fixes the pixel_blend_mode reset when the blend mode property is advertised without PREMULTI support. It's needed for AMD DCE-generation (patch 3) and I think with this we can also remove a workaround in nouveau nv50_wndw_default_state() caused by the hardcoded PREMULTI default value [1].
- Patch 2 fixes the missing-blend-mode-property warning for DCN primary plane and for DCN+DCE cursor plane. The alpha property keeps only on overlay planes because looks like this is not supported by AMD primary planes and it doesn't affect the blend mode warning we are targetting. Enabling alpha properties to primary was also causing -EINVAL on IGT alpha tests because it started testing primary planes and disabling it (which is not allowed by the AMD display driver).
- Patch 3 fixes the warning for DCE primary plane, but I detached this solution from the previous patch because I don't have hardware to validate if my assumption about PIXEL_NONE-only is correct.
[1] https://lore.kernel.org/dri-devel/[email protected]/
[v1]: https://lore.kernel.org/dri-devel/[email protected]/
Changes:
- new patch for pixel_blend_mode default value other than PREMULTI.
- remove DCE11 which supports per_pixel_alpha but doesn't support DCN_UNIVERSAL_PLANE.
- new patch for the primary plane blend mode on DCE-generations
[v2]: https://lore.kernel.org/dri-devel/[email protected]/
Changes:
- keep alpha property only for overlay planes (looks like primary plane doesn't actually support it) - alpha prop in primary planes causes IGT test failures with RX 7900 XT on a 4k60 HP U27 as reported by Daniel Wheeler.
- improve readability of if conditions (Alex H)
- add Leandro's r-b tags.
Best Regards,
Melissa
Melissa Wen (3):
drm/atomic-state-helper: set pixel_blend_mode to prop default on reset drm/amd/display: fix missing blend-mode-prop warning for DCN drm/amd/display: advertise PIXEL_NONE as blend mode for DCE generations
.../amd/display/amdgpu_dm/amdgpu_dm_plane.c | 25 ++++++++++++++++--- drivers/gpu/drm/drm_atomic_state_helper.c | 7 ++++++ 2 files changed, 29 insertions(+), 3 deletions(-)
--
2.53.0
Hi,
On current mainline every boot on my machine produces this at amdgpu probe:
[PLANE:44:plane-0] pixel format with alpha exposed but blend mode not setup
WARNING: drivers/gpu/drm/drm_mode_config.c:872 at drm_mode_config_validate+0xa16/0x13b0
16 times per boot: 8 per GPU, on the 4 primary and 4 cursor planes of each of the two cards in this box. I bisected it to 860e748bddcc ("drm: ensure blend mode supported if pixel format with alpha exposed") and confirmed it by reverting that commit, before I found this series.
With all three patches of this series applied the warnings are gone.
Base: 66498c75b4f8, kernel 7.2.0-66498c75b4f8+ #154, KASAN and lockdep enabled. Hardware:
0000:03:00.0 Navi 31 (0x1002:0x744C), DCN 3.2
0000:0f:00.0 Raphael iGPU (0x1002:0x164E), DCN 3.1.5
After booting into a normal desktop session:
dmesg | grep -c 'blend mode not setup' -> 0 dmesg | grep -c 'drm_mode_config_validate' -> 0 drm_info | grep -c 'pixel blend mode' -> 20
i.e. all 10 planes on each GPU (4 primary, 2 overlay, 4 cursor) now expose the property. No WARNING, no UBSAN and no lockdep report anywhere in the buffer.
What I did not cover: I have no DCE hardware, so patch 3/3 is untested; and since PREMULTI is present in the blend caps of every plane here, the property default lookup added by 1/3 resolves to the same value that was hardcoded before, so that path is exercised only in its trivial case.
One question about routing: 860e748bddcc landed in the 7.3 merge window, so this is a merge-window regression - it was never in a release, and no stable backport is involved. But the fix should still make 7.3 rather than wait for the 7.4 merge window: right now every amdgpu machine on mainline hits this at probe. 271e90eb5f9f ("drm: use drm_warn() in validate_blend_mode_for_alpha_formats()") is already in drm-misc-next-fixes, while this series is in neither linux-next nor amd-staging-drm-next. Could it go the same way - drm-misc-next-fixes while the window is open, drm-fixes after -rc1 - instead of through amd-staging-drm-next, which is aimed at 7.4 now? Splitting it looks wrong anyway: 3/3 gives DCE planes a property with PIXEL_NONE only, which needs 1/3 to not end up with PREMULTI in the default state.
Signature · 2 lines
Thanks, Mikhail