drm/amd/display: Update MBLK calculation for SubVP

[Description]
Update MBLK calculation according to hardware doc.  For DCC case we were
not allocation enough MALL due to an inaccurate MBLK calculation.
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Pavle Kotarac <Pavle.Kotarac@amd.com>
Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.h

@@ -30,6 +30,9 @@
 
 #define DCN3_2_DET_SEG_SIZE 64
 #define DCN3_2_MALL_MBLK_SIZE_BYTES 65536 // 64 * 1024
+#define DCN3_2_MBLK_WIDTH 128
+#define DCN3_2_MBLK_HEIGHT_4BPE 128
+#define DCN3_2_MBLK_HEIGHT_8BPE 64
 
 #define TO_DCN32_RES_POOL(pool)\
 	container_of(pool, struct dcn32_resource_pool, base)

drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource_helpers.c

@@ -46,7 +46,6 @@
 uint32_t dcn32_helper_calculate_num_ways_for_subvp(struct dc *dc, struct dc_state *context)
 {
 	uint32_t num_ways = 0;
-	uint32_t mall_region_pixels = 0;
 	uint32_t bytes_per_pixel = 0;
 	uint32_t cache_lines_used = 0;
 	uint32_t lines_per_way = 0;
@@ -54,20 +53,64 @@ uint32_t dcn32_helper_calculate_num_ways_for_subvp(struct dc *dc, struct dc_stat
 	uint32_t bytes_in_mall = 0;
 	uint32_t num_mblks = 0;
 	uint32_t cache_lines_per_plane = 0;
-	uint32_t i = 0;
+	uint32_t i = 0, j = 0;
+	uint32_t mblk_width = 0;
+	uint32_t mblk_height = 0;
+	uint32_t full_vp_width_blk_aligned = 0;
+	uint32_t full_vp_height_blk_aligned = 0;
+	uint32_t mall_alloc_width_blk_aligned = 0;
+	uint32_t mall_alloc_height_blk_aligned = 0;
+	uint32_t full_vp_height = 0;
 
 	for (i = 0; i < dc->res_pool->pipe_count; i++) {
 		struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
 
 		// Find the phantom pipes
-		if (pipe->stream && pipe->plane_state && !pipe->top_pipe &&
+		if (pipe->stream && pipe->plane_state && !pipe->top_pipe && !pipe->prev_odm_pipe &&
 				pipe->stream->mall_stream_config.type == SUBVP_PHANTOM) {
-			bytes_per_pixel = pipe->plane_state->format >= SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616 ? 8 : 4;
-			mall_region_pixels = pipe->plane_state->plane_size.surface_pitch * pipe->stream->timing.v_addressable;
+			struct pipe_ctx *main_pipe = NULL;
+
+			/* Get full viewport height from main pipe (required for MBLK calculation) */
+			for (j = 0; j < dc->res_pool->pipe_count; j++) {
+				main_pipe = &context->res_ctx.pipe_ctx[j];
+				if (main_pipe->stream == pipe->stream->mall_stream_config.paired_stream) {
+					full_vp_height = main_pipe->plane_res.scl_data.viewport.height;
+					break;
+				}
+			}
 
-			// For bytes required in MALL, calculate based on number of MBlks required
-			num_mblks = (mall_region_pixels * bytes_per_pixel +
-					DCN3_2_MALL_MBLK_SIZE_BYTES - 1) / DCN3_2_MALL_MBLK_SIZE_BYTES;
+			bytes_per_pixel = pipe->plane_state->format >= SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616 ? 8 : 4;
+			mblk_width = DCN3_2_MBLK_WIDTH;
+			mblk_height = bytes_per_pixel == 4 ? DCN3_2_MBLK_HEIGHT_4BPE : DCN3_2_MBLK_HEIGHT_8BPE;
+
+			/* full_vp_width_blk_aligned = FLOOR(vp_x_start + full_vp_width + blk_width - 1, blk_width) -
+			 * FLOOR(vp_x_start, blk_width)
+			 */
+			full_vp_width_blk_aligned = ((pipe->plane_res.scl_data.viewport.x +
+					pipe->plane_res.scl_data.viewport.width + mblk_width - 1) / mblk_width * mblk_width) +
+					(pipe->plane_res.scl_data.viewport.x / mblk_width * mblk_width);
+
+			/* full_vp_height_blk_aligned = FLOOR(vp_y_start + full_vp_height + blk_height - 1, blk_height) -
+			 * FLOOR(vp_y_start, blk_height)
+			 */
+			full_vp_height_blk_aligned = ((pipe->plane_res.scl_data.viewport.y +
+					full_vp_height + mblk_height - 1) / mblk_height * mblk_height) +
+					(pipe->plane_res.scl_data.viewport.y / mblk_height * mblk_height);
+
+			/* mall_alloc_width_blk_aligned_l/c = full_vp_width_blk_aligned_l/c */
+			mall_alloc_width_blk_aligned = full_vp_width_blk_aligned;
+
+			/* mall_alloc_height_blk_aligned_l/c = CEILING(sub_vp_height_l/c - 1, blk_height_l/c) + blk_height_l/c */
+			mall_alloc_height_blk_aligned = (pipe->stream->timing.v_addressable - 1 + mblk_height - 1) /
+					mblk_height * mblk_height + mblk_height;
+
+			/* full_mblk_width_ub_l/c = mall_alloc_width_blk_aligned_l/c;
+			 * full_mblk_height_ub_l/c = mall_alloc_height_blk_aligned_l/c;
+			 * num_mblk_l/c = (full_mblk_width_ub_l/c / mblk_width_l/c) * (full_mblk_height_ub_l/c / mblk_height_l/c);
+			 * (Should be divisible, but round up if not)
+			 */
+			num_mblks = ((mall_alloc_width_blk_aligned + mblk_width - 1) / mblk_width) *
+					((mall_alloc_height_blk_aligned + mblk_height - 1) / mblk_height);
 			bytes_in_mall = num_mblks * DCN3_2_MALL_MBLK_SIZE_BYTES;
 			// cache lines used is total bytes / cache_line size. Add +2 for worst case alignment
 			// (MALL is 64-byte aligned)