On 5/6/21 9:19 AM, Imre Deak wrote:
From: José Roberto de Souza <jose.souza@xxxxxxxxx>
Alderlake-P have a new stride restriction when using DPT and it is used
by non linear framebuffers. Stride needs to be a power of two to take
full DPT rows, but stride is a parameter set by userspace.
What we could do is use a fake stride when doing DPT allocation so
HW requirements are met and userspace don't need to be changed to
met this power of two restrictions but this change will take a while
to be implemented so for now adding this restriction in driver to
reject atomic commits that would cause visual corruptions.
BSpec: 53393
Acked-by: Matt Roper <matthew.d.roper@xxxxxxxxx>
Cc: Matt Roper <matthew.d.roper@xxxxxxxxx>
Cc: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx>
Cc: Stanislav Lisovskiy <stanislav.lisovskiy@xxxxxxxxx>
Signed-off-by: José Roberto de Souza <jose.souza@xxxxxxxxx>
Signed-off-by: Imre Deak <imre.deak@xxxxxxxxx>
---
drivers/gpu/drm/i915/display/intel_display.c | 9 +++++++++
1 file changed, 9 insertions(+)
diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c
index 292396058e75d..70ac197746b1f 100644
--- a/drivers/gpu/drm/i915/display/intel_display.c
+++ b/drivers/gpu/drm/i915/display/intel_display.c
@@ -11566,6 +11566,15 @@ static int intel_framebuffer_init(struct intel_framebuffer *intel_fb,
}
}
+ if (IS_ALDERLAKE_P(dev_priv) &&
+ mode_cmd->modifier[i] != DRM_FORMAT_MOD_LINEAR &&
+ !is_power_of_2(mode_cmd->pitches[i])) {
+ drm_dbg_kms(&dev_priv->drm,
+ "plane %d pitch (%d) must be power of two for tiled buffers\n",
+ i, mode_cmd->pitches[i]);
+ goto err;
+ }
+
Reviewed-by: Clint Taylor <Clinton.A.Taylor@xxxxxxxxx>
fb->obj[i] = &obj->base;
}
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
