On Wed, Mar 19, 2025 at 04:02:17PM +0530, Arun R Murthy wrote:
> Hook up the newly added plane function pointer
> format_mod_supported_async to populate the modifiers/formats supported
> by asynchronous flips.
>
> v5: Correct the if condition for modifier support check (Chaitanya)
> v6: Replace uint32_t/uint64_t with u32/u64 (Jani)
> v7: Move plannar check from intel_async_flip_check_hw() to
> intel_plane_format_mod_supported_async() (Ville)
> v8: In case of error print format/modifier (Chaitanya)
> v9: Exclude C8 format as its not supported by hardware
>
> Signed-off-by: Arun R Murthy <arun.r.mur...@intel.com>
> ---
> drivers/gpu/drm/i915/display/i9xx_plane.c | 6 ++--
> drivers/gpu/drm/i915/display/intel_atomic_plane.c | 32
> +++++++++++++++++++++-
> drivers/gpu/drm/i915/display/intel_atomic_plane.h | 6 +++-
> drivers/gpu/drm/i915/display/intel_display.c | 14 +++-------
> drivers/gpu/drm/i915/display/skl_universal_plane.c | 5 +++-
> 5 files changed, 48 insertions(+), 15 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/display/i9xx_plane.c
> b/drivers/gpu/drm/i915/display/i9xx_plane.c
> index
> 013295f66d56ec5e919b3a0c904034bf7985986a..6bd09adb8a30ba002ef334261d7638f398587a3e
> 100644
> --- a/drivers/gpu/drm/i915/display/i9xx_plane.c
> +++ b/drivers/gpu/drm/i915/display/i9xx_plane.c
> @@ -820,7 +820,7 @@ unsigned int vlv_plane_min_alignment(struct intel_plane
> *plane,
> {
> struct intel_display *display = to_intel_display(plane);
>
> - if (intel_plane_can_async_flip(plane, fb->modifier))
> + if (intel_plane_can_async_flip(plane, fb->format->format, fb->modifier))
> return 256 * 1024;
>
> /* FIXME undocumented so not sure what's actually needed */
> @@ -844,7 +844,7 @@ static unsigned int g4x_primary_min_alignment(struct
> intel_plane *plane,
> {
> struct intel_display *display = to_intel_display(plane);
>
> - if (intel_plane_can_async_flip(plane, fb->modifier))
> + if (intel_plane_can_async_flip(plane, fb->format->format, fb->modifier))
> return 256 * 1024;
>
> if (intel_scanout_needs_vtd_wa(display))
> @@ -889,6 +889,7 @@ static const struct drm_plane_funcs i965_plane_funcs = {
> .atomic_duplicate_state = intel_plane_duplicate_state,
> .atomic_destroy_state = intel_plane_destroy_state,
> .format_mod_supported = i965_plane_format_mod_supported,
> + .format_mod_supported_async = intel_plane_format_mod_supported_async,
> };
>
> static const struct drm_plane_funcs i8xx_plane_funcs = {
> @@ -898,6 +899,7 @@ static const struct drm_plane_funcs i8xx_plane_funcs = {
> .atomic_duplicate_state = intel_plane_duplicate_state,
> .atomic_destroy_state = intel_plane_destroy_state,
> .format_mod_supported = i8xx_plane_format_mod_supported,
> + .format_mod_supported_async = intel_plane_format_mod_supported_async,
> };
>
> struct intel_plane *
> diff --git a/drivers/gpu/drm/i915/display/intel_atomic_plane.c
> b/drivers/gpu/drm/i915/display/intel_atomic_plane.c
> index
> 7276179df878658b7053fe6d8dc37b69f19625e3..ce6bf6fe8f241a9517e8f74fb002b835c3f0853a
> 100644
> --- a/drivers/gpu/drm/i915/display/intel_atomic_plane.c
> +++ b/drivers/gpu/drm/i915/display/intel_atomic_plane.c
> @@ -174,11 +174,41 @@ bool intel_plane_needs_physical(struct intel_plane
> *plane)
> DISPLAY_INFO(display)->cursor_needs_physical;
> }
>
> -bool intel_plane_can_async_flip(struct intel_plane *plane, u64 modifier)
> +bool intel_plane_can_async_flip(struct intel_plane *plane, u32 format,
> + u64 modifier)
Please split this change into a separate patch.
> {
> + struct intel_display *display = to_intel_display(plane);
> +
> + if ((DISPLAY_VER(display) <= 14 ?
> + drm_format_info(format)->is_yuv :
> + intel_format_info_is_yuv_semiplanar(drm_format_info(format),
> + modifier)) ||
> + format == DRM_FORMAT_C8) {
> + drm_dbg_kms(plane->base.dev,
> + "[PLANE:%d:%s] Planar formats do not support async
> flips\n",
> + plane->base.base.id, plane->base.name);
> + return false;
This is introducing functional changes. Those should be separate
patches.
I also don't know what this pre-mtl vs. mtl+ stuff is about. I'm
pretty sure planar formats are the only thing that actually can't do
async flips (in the sense that we can't atomically async flip both
luma and chroma).
I suppose we may want to reject C8 as well since it depends on
the LUT and we can't update that atomically with an async flip.
> + }
> +
> return plane->can_async_flip && plane->can_async_flip(modifier);
> }
>
> +bool intel_plane_format_mod_supported_async(struct drm_plane *plane,
> + u32 format,
> + u64 modifier)
> +{
> + if (plane->funcs->format_mod_supported &&
> + !plane->funcs->format_mod_supported(plane, format, modifier)) {
> + drm_dbg_kms(plane->dev,
> + "[PLANE:%d:%s](format %p4cc) modifier 0x%llx not in
> universal list\n",
> + plane->base.id, plane->name, &format, modifier);
> + return false;
> + }
> +
> + return intel_plane_can_async_flip(to_intel_plane(plane),
> + format, modifier);
> +}
> +
> unsigned int intel_adjusted_rate(const struct drm_rect *src,
> const struct drm_rect *dst,
> unsigned int rate)
> diff --git a/drivers/gpu/drm/i915/display/intel_atomic_plane.h
> b/drivers/gpu/drm/i915/display/intel_atomic_plane.h
> index
> 6efac923dcbc757e6f68564cbef2919c920f13cb..512c251cc153753a4808cf177c8bcce2178bb862
> 100644
> --- a/drivers/gpu/drm/i915/display/intel_atomic_plane.h
> +++ b/drivers/gpu/drm/i915/display/intel_atomic_plane.h
> @@ -21,7 +21,8 @@ enum plane_id;
>
> struct intel_plane *
> intel_crtc_get_plane(struct intel_crtc *crtc, enum plane_id plane_id);
> -bool intel_plane_can_async_flip(struct intel_plane *plane, u64 modifier);
> +bool intel_plane_can_async_flip(struct intel_plane *plane, u32 format,
> + u64 modifier);
> unsigned int intel_adjusted_rate(const struct drm_rect *src,
> const struct drm_rect *dst,
> unsigned int rate);
> @@ -87,6 +88,9 @@ void intel_plane_init_cursor_vblank_work(struct
> intel_plane_state *old_plane_sta
> int intel_atomic_add_affected_planes(struct intel_atomic_state *state,
> struct intel_crtc *crtc);
> int intel_atomic_check_planes(struct intel_atomic_state *state);
> +bool intel_plane_format_mod_supported_async(struct drm_plane *plane,
> + u32 format,
> + u64 modifier);
>
> u32 intel_plane_ggtt_offset(const struct intel_plane_state *plane_state);
>
> diff --git a/drivers/gpu/drm/i915/display/intel_display.c
> b/drivers/gpu/drm/i915/display/intel_display.c
> index
> 3afb85fe8536dfffd55dbaa07f6727112cc876b7..5d0bab1f8ff8294716ca5843c856032d2b9ccd5b
> 100644
> --- a/drivers/gpu/drm/i915/display/intel_display.c
> +++ b/drivers/gpu/drm/i915/display/intel_display.c
> @@ -6005,22 +6005,16 @@ static int intel_async_flip_check_hw(struct
> intel_atomic_state *state, struct in
> if (!plane->async_flip)
> continue;
>
> - if (!intel_plane_can_async_flip(plane,
> new_plane_state->hw.fb->modifier)) {
> + if (!intel_plane_can_async_flip(plane,
> new_plane_state->hw.fb->format->format,
> +
> new_plane_state->hw.fb->modifier)) {
> drm_dbg_kms(display->drm,
> - "[PLANE:%d:%s] Modifier 0x%llx does not
> support async flip\n",
> + "[PLANE:%d:%s] Format %p4cc Modifier 0x%llx
> does not support async flip\n",
> plane->base.base.id, plane->base.name,
> + &new_plane_state->hw.fb->format->format,
> new_plane_state->hw.fb->modifier);
> return -EINVAL;
> }
>
> - if
> (intel_format_info_is_yuv_semiplanar(new_plane_state->hw.fb->format,
> -
> new_plane_state->hw.fb->modifier)) {
> - drm_dbg_kms(display->drm,
> - "[PLANE:%d:%s] Planar formats do not
> support async flips\n",
> - plane->base.base.id, plane->base.name);
> - return -EINVAL;
> - }
> -
> /*
> * We turn the first async flip request into a sync flip
> * so that we can reconfigure the plane (eg. change modifier).
> diff --git a/drivers/gpu/drm/i915/display/skl_universal_plane.c
> b/drivers/gpu/drm/i915/display/skl_universal_plane.c
> index
> 70e550539bb21393c7173c7b3904e7790eab25f4..f61e1eff30bb4820ccb17daa5d4b2b073a5d4078
> 100644
> --- a/drivers/gpu/drm/i915/display/skl_universal_plane.c
> +++ b/drivers/gpu/drm/i915/display/skl_universal_plane.c
> @@ -601,7 +601,7 @@ static u32 tgl_plane_min_alignment(struct intel_plane
> *plane,
> * Figure out what's going on here...
> */
> if (display->platform.alderlake_p &&
> - intel_plane_can_async_flip(plane, fb->modifier))
> + intel_plane_can_async_flip(plane, fb->format->format, fb->modifier))
> return mult * 16 * 1024;
>
> switch (fb->modifier) {
> @@ -2666,6 +2666,7 @@ static const struct drm_plane_funcs skl_plane_funcs = {
> .atomic_duplicate_state = intel_plane_duplicate_state,
> .atomic_destroy_state = intel_plane_destroy_state,
> .format_mod_supported = skl_plane_format_mod_supported,
> + .format_mod_supported_async = intel_plane_format_mod_supported_async,
> };
>
> static const struct drm_plane_funcs icl_plane_funcs = {
> @@ -2675,6 +2676,7 @@ static const struct drm_plane_funcs icl_plane_funcs = {
> .atomic_duplicate_state = intel_plane_duplicate_state,
> .atomic_destroy_state = intel_plane_destroy_state,
> .format_mod_supported = icl_plane_format_mod_supported,
> + .format_mod_supported_async = intel_plane_format_mod_supported_async,
> };
>
> static const struct drm_plane_funcs tgl_plane_funcs = {
> @@ -2684,6 +2686,7 @@ static const struct drm_plane_funcs tgl_plane_funcs = {
> .atomic_duplicate_state = intel_plane_duplicate_state,
> .atomic_destroy_state = intel_plane_destroy_state,
> .format_mod_supported = tgl_plane_format_mod_supported,
> + .format_mod_supported_async = intel_plane_format_mod_supported_async,
> };
>
> static void
>
> --
> 2.25.1
--
Ville Syrjälä
Intel
