drm/amd/display: Round cursor width up for MALL allocation

[Why & How]
When calculating cursor size for MALL allocation, the cursor width should
be the actual width rounded up to 64 alignment. Additionally, the bit
depth should vary depending on color format.
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Alvin Lee <Alvin.Lee2@amd.com>
Acked-by: Pavle Kotarac <Pavle.Kotarac@amd.com>
Signed-off-by: Taimur Hassan <Syed.Hassan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hubp.c

@@ -103,6 +103,11 @@ void hubp32_cursor_set_attributes(
 	enum cursor_lines_per_chunk lpc = hubp2_get_lines_per_chunk(
 			attr->width, attr->color_format);
 
+	//Round cursor width up to next multiple of 64
+	uint32_t cursor_width = ((attr->width + 63) / 64) * 64;
+	uint32_t cursor_height = attr->height;
+	uint32_t cursor_size = cursor_width * cursor_height;
+
 	hubp->curs_attr = *attr;
 
 	REG_UPDATE(CURSOR_SURFACE_ADDRESS_HIGH,
@@ -126,7 +131,24 @@ void hubp32_cursor_set_attributes(
 			 /* used to shift the cursor chunk request deadline */
 			CURSOR0_CHUNK_HDL_ADJUST, 3);
 
-	if (attr->width * attr->height * 4 > 16384)
+	switch (attr->color_format) {
+	case CURSOR_MODE_MONO:
+		cursor_size /= 2;
+		break;
+	case CURSOR_MODE_COLOR_1BIT_AND:
+	case CURSOR_MODE_COLOR_PRE_MULTIPLIED_ALPHA:
+	case CURSOR_MODE_COLOR_UN_PRE_MULTIPLIED_ALPHA:
+		cursor_size *= 4;
+		break;
+
+	case CURSOR_MODE_COLOR_64BIT_FP_PRE_MULTIPLIED:
+	case CURSOR_MODE_COLOR_64BIT_FP_UN_PRE_MULTIPLIED:
+	default:
+		cursor_size *= 8;
+		break;
+	}
+
+	if (cursor_size > 16384)
 		REG_UPDATE(DCHUBP_MALL_CONFIG, USE_MALL_FOR_CURSOR, true);
 	else
 		REG_UPDATE(DCHUBP_MALL_CONFIG, USE_MALL_FOR_CURSOR, false);

drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.c

@@ -741,7 +741,29 @@ void dcn32_update_mall_sel(struct dc *dc, struct dc_state *context)
 		struct hubp *hubp = pipe->plane_res.hubp;
 
 		if (pipe->stream && pipe->plane_state && hubp && hubp->funcs->hubp_update_mall_sel) {
-			if (hubp->curs_attr.width * hubp->curs_attr.height * 4 > 16384)
+			//Round cursor width up to next multiple of 64
+			int cursor_width = ((hubp->curs_attr.width + 63) / 64) * 64;
+			int cursor_height = hubp->curs_attr.height;
+			int cursor_size = cursor_width * cursor_height;
+
+			switch (hubp->curs_attr.color_format) {
+			case CURSOR_MODE_MONO:
+				cursor_size /= 2;
+				break;
+			case CURSOR_MODE_COLOR_1BIT_AND:
+			case CURSOR_MODE_COLOR_PRE_MULTIPLIED_ALPHA:
+			case CURSOR_MODE_COLOR_UN_PRE_MULTIPLIED_ALPHA:
+				cursor_size *= 4;
+				break;
+
+			case CURSOR_MODE_COLOR_64BIT_FP_PRE_MULTIPLIED:
+			case CURSOR_MODE_COLOR_64BIT_FP_UN_PRE_MULTIPLIED:
+			default:
+				cursor_size *= 8;
+				break;
+			}
+
+			if (cursor_size > 16384)
 				cache_cursor = true;
 
 			if (pipe->stream->mall_stream_config.type == SUBVP_PHANTOM) {