drm/amd/display: Set max vratio for prefetch to 7.9 for YUV420 MPO

[Description]
- Single 4K60 playing YUV420 MPO video blocks P-State because
  the required VRatio for prefetch is too high (luma plane for
  YUV420 is 1bpe, so swath height is 16 and prefetch requires
  more lines)
- Allow max vratio per plane to be 7.9 for single display YUV420
  MPO video cases
- Ensure that global vratio prefetch (i.e. total prefetch BW vs.
  total active bandwidth) does not excited 4.0
Reviewed-by: Nevenko Stupar <Nevenko.Stupar@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -150,6 +150,8 @@ bool dcn32_allow_subvp_with_active_margin(struct pipe_ctx *pipe);
 
 unsigned int dcn32_calc_num_avail_chans_for_mall(struct dc *dc, int num_chans);
 
+double dcn32_determine_max_vratio_prefetch(struct dc *dc, struct dc_state *context);
+
 /* definitions for run time init of reg offsets */
 
 /* CLK SRC */

drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c

@@ -27,6 +27,7 @@
 #include "dcn32/dcn32_resource.h"
 #include "dcn20/dcn20_resource.h"
 #include "display_mode_vba_util_32.h"
+#include "dml/dcn32/display_mode_vba_32.h"
 // We need this includes for WATERMARKS_* defines
 #include "clk_mgr/dcn32/dcn32_smu13_driver_if.h"
 #include "dcn30/dcn30_resource.h"
@@ -1618,6 +1619,7 @@ bool dcn32_internal_validate_bw(struct dc *dc,
 	}
 
 	dml_log_pipe_params(&context->bw_ctx.dml, pipes, pipe_cnt);
+	context->bw_ctx.dml.soc.max_vratio_pre = dcn32_determine_max_vratio_prefetch(dc, context);
 
 	if (!fast_validate)
 		dcn32_full_validate_bw_helper(dc, context, pipes, &vlevel, split, merge, &pipe_cnt);
@@ -2744,3 +2746,33 @@ bool dcn32_allow_subvp_with_active_margin(struct pipe_ctx *pipe)
 	}
 	return allow;
 }
+
+/**
+ * *******************************************************************************************
+ * dcn32_determine_max_vratio_prefetch: Determine max Vratio for prefetch by driver policy
+ *
+ * @param [in]: dc: Current DC state
+ * @param [in]: context: New DC state to be programmed
+ *
+ * @return: Max vratio for prefetch
+ *
+ * *******************************************************************************************
+ */
+double dcn32_determine_max_vratio_prefetch(struct dc *dc, struct dc_state *context)
+{
+	double max_vratio_pre = __DML_MAX_BW_RATIO_PRE__; // Default value is 4
+	int i;
+
+	/* For single display MPO configs, allow the max vratio to be 8
+	 * if any plane is YUV420 format
+	 */
+	if (context->stream_count == 1 && context->stream_status[0].plane_count > 1) {
+		for (i = 0; i < context->stream_status[0].plane_count; i++) {
+			if (context->stream_status[0].plane_states[i]->format == SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr ||
+					context->stream_status[0].plane_states[i]->format == SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb) {
+				max_vratio_pre = __DML_MAX_VRATIO_PRE__;
+			}
+		}
+	}
+	return max_vratio_pre;
+}

drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.c

@@ -896,8 +896,8 @@ static void DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerforman
 			if (v->DestinationLinesForPrefetch[k] < 2)
 				DestinationLineTimesForPrefetchLessThan2 = true;
 
-			if (v->VRatioPrefetchY[k] > __DML_MAX_VRATIO_PRE__
-					|| v->VRatioPrefetchC[k] > __DML_MAX_VRATIO_PRE__)
+			if (v->VRatioPrefetchY[k] > v->MaxVRatioPre
+					|| v->VRatioPrefetchC[k] > v->MaxVRatioPre)
 				VRatioPrefetchMoreThanMax = true;
 
 			//bool DestinationLinesToRequestVMInVBlankEqualOrMoreThan32 = false;
@@ -3397,8 +3397,8 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 
 				mode_lib->vba.VRatioInPrefetchSupported[i][j] = true;
 				for (k = 0; k <= mode_lib->vba.NumberOfActiveSurfaces - 1; k++) {
-					if (mode_lib->vba.VRatioPreY[i][j][k] > __DML_MAX_VRATIO_PRE__
-							|| mode_lib->vba.VRatioPreC[i][j][k] > __DML_MAX_VRATIO_PRE__
+					if (mode_lib->vba.VRatioPreY[i][j][k] > mode_lib->vba.MaxVRatioPre
+							|| mode_lib->vba.VRatioPreC[i][j][k] > mode_lib->vba.MaxVRatioPre
 							|| mode_lib->vba.NoTimeForPrefetch[i][j][k] == true) {
 						mode_lib->vba.VRatioInPrefetchSupported[i][j] = false;
 					}

drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.h

@@ -44,7 +44,8 @@
 #define __DML_MIN_DCFCLK_FACTOR__   1.15
 
 // Prefetch schedule max vratio
-#define __DML_MAX_VRATIO_PRE__ 4.0
+#define __DML_MAX_VRATIO_PRE__ 7.9
+#define __DML_MAX_BW_RATIO_PRE__ 4.0
 
 #define __DML_VBA_MAX_DST_Y_PRE__    63.75
 

drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.c

@@ -3480,7 +3480,7 @@ bool dml32_CalculatePrefetchSchedule(
 	double  prefetch_sw_bytes;
 	double  bytes_pp;
 	double  dep_bytes;
-	unsigned int max_vratio_pre = __DML_MAX_VRATIO_PRE__;
+	unsigned int max_vratio_pre = v->MaxVRatioPre;
 	double  min_Lsw;
 	double  Tsw_est1 = 0;
 	double  Tsw_est3 = 0;
@@ -6150,7 +6150,9 @@ void dml32_CalculatePrefetchBandwithSupport(unsigned int NumberOfActiveSurfaces,
 		bool *PrefetchBandwidthSupport)
 {
 	unsigned int k;
+	double ActiveBandwidthPerSurface;
 	bool NotEnoughUrgentLatencyHiding = false;
+	double TotalActiveBandwidth = 0;
 	for (k = 0; k < NumberOfActiveSurfaces; ++k) {
 		if (NotUrgentLatencyHiding[k]) {
 			NotEnoughUrgentLatencyHiding = true;
@@ -6159,12 +6161,16 @@ void dml32_CalculatePrefetchBandwithSupport(unsigned int NumberOfActiveSurfaces,
 
 	*PrefetchBandwidth = 0;
 	for (k = 0; k < NumberOfActiveSurfaces; ++k) {
+		ActiveBandwidthPerSurface = ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k] + NumberOfDPP[k] * (meta_row_bandwidth[k] + dpte_row_bandwidth[k]);
+
+		TotalActiveBandwidth += ActiveBandwidthPerSurface;
+
 		*PrefetchBandwidth = *PrefetchBandwidth + dml_max3(NumberOfDPP[k] * prefetch_vmrow_bw[k],
-				ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k] + NumberOfDPP[k] * (meta_row_bandwidth[k] + dpte_row_bandwidth[k]),
+				ActiveBandwidthPerSurface,
 				NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * UrgentBurstFactorLumaPre[k] + PrefetchBandwidthChroma[k] * UrgentBurstFactorChromaPre[k]) + cursor_bw_pre[k] * UrgentBurstFactorCursorPre[k]);
 	}
 
-	*PrefetchBandwidthSupport = (*PrefetchBandwidth <= ReturnBW) && !NotEnoughUrgentLatencyHiding;
+	*PrefetchBandwidthSupport = (*PrefetchBandwidth <= ReturnBW) &&  (*PrefetchBandwidth <= TotalActiveBandwidth * __DML_MAX_BW_RATIO_PRE__) && !NotEnoughUrgentLatencyHiding;
 	*FractionOfUrgentBandwidth = *PrefetchBandwidth / ReturnBW;
 }
 

drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h

@@ -246,6 +246,7 @@ struct _vcs_dpi_soc_bounding_box_st {
 	bool disable_dram_clock_change_vactive_support;
 	bool allow_dram_clock_one_display_vactive;
 	enum self_refresh_affinity allow_dram_self_refresh_or_dram_clock_change_in_vblank;
+	double max_vratio_pre;
 };
 
 /**

drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.c

@@ -412,6 +412,7 @@ static void fetch_socbb_params(struct display_mode_lib *mode_lib)
 		soc->urgent_latency_adjustment_fabric_clock_component_us;
 	mode_lib->vba.UrgentLatencyAdjustmentFabricClockReference =
 		soc->urgent_latency_adjustment_fabric_clock_reference_mhz;
+	mode_lib->vba.MaxVRatioPre = soc->max_vratio_pre;
 }
 
 static void fetch_ip_params(struct display_mode_lib *mode_lib)

drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.h

@@ -263,6 +263,7 @@ struct vba_vars_st {
 	int maxMpcComb;
 	bool UseMaximumVStartup;
 
+	double MaxVRatioPre;
 	double WritebackDISPCLK;
 	double DPPCLKUsingSingleDPPLuma;
 	double DPPCLKUsingSingleDPPChroma;