drm/amd/display: Speed up DML fast_validate path

[Why]
Iterating over every voltage state when we need to validate thousands of
configurations all at once (i.e. display hotplug) can take a significant
amount of time.

[How]
Check just the highest voltage state when fast_validate is true to
verify whether the configuration can work at all, then do a proper
validation including all voltage states later when fast_validate is false.
Reviewed-by: Nevenko Stupar <Nevenko.Stupar@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Ilya Bakoulin <Ilya.Bakoulin@amd.com>
Signed-off-by: Dillon Varone <Dillon.Varone@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/dml/dcn32/dcn32_fpu.c

@@ -1603,6 +1603,7 @@ bool dcn32_internal_validate_bw(struct dc *dc,
 		context->bw_ctx.dml.soc.allow_for_pstate_or_stutter_in_vblank_final =
 			dm_prefetch_support_fclk_and_stutter;
 
+		context->bw_ctx.dml.validate_max_state = fast_validate;
 		vlevel = dml_get_voltage_level(&context->bw_ctx.dml, pipes, pipe_cnt);
 
 		/* Last attempt with Prefetch mode 2 (dm_prefetch_support_stutter == 3) */
@@ -1611,6 +1612,7 @@ bool dcn32_internal_validate_bw(struct dc *dc,
 				dm_prefetch_support_stutter;
 			vlevel = dml_get_voltage_level(&context->bw_ctx.dml, pipes, pipe_cnt);
 		}
+		context->bw_ctx.dml.validate_max_state = false;
 
 		if (vlevel < context->bw_ctx.dml.soc.num_states) {
 			memset(split, 0, sizeof(split));

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

@@ -1639,9 +1639,14 @@ static void DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerforman
 static void mode_support_configuration(struct vba_vars_st *v,
 				  struct display_mode_lib *mode_lib)
 {
-	int i, j;
+	int i, j, start_state;
 
-	for (i = v->soc.num_states - 1; i >= 0; i--) {
+	if (mode_lib->validate_max_state)
+		start_state = v->soc.num_states - 1;
+	else
+		start_state = 0;
+
+	for (i = v->soc.num_states - 1; i >= start_state; i--) {
 		for (j = 0; j < 2; j++) {
 			if (mode_lib->vba.ScaleRatioAndTapsSupport == true
 				&& mode_lib->vba.SourceFormatPixelAndScanSupport == true
@@ -1710,7 +1715,7 @@ static void mode_support_configuration(struct vba_vars_st *v,
 void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_lib)
 {
 	struct vba_vars_st *v = &mode_lib->vba;
-	int i, j;
+	int i, j, start_state;
 	unsigned int k, m;
 	unsigned int MaximumMPCCombine;
 	unsigned int NumberOfNonCombinedSurfaceOfMaximumBandwidth;
@@ -1723,6 +1728,10 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 #endif
 
 	/*MODE SUPPORT, VOLTAGE STATE AND SOC CONFIGURATION*/
+	if (mode_lib->validate_max_state)
+		start_state = v->soc.num_states - 1;
+	else
+		start_state = 0;
 
 	/*Scale Ratio, taps Support Check*/
 
@@ -2012,7 +2021,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 	mode_lib->vba.MPCCombineMethodIncompatible = v->dummy_vars.dml32_ModeSupportAndSystemConfigurationFull.MPCCombineMethodAsNeededForPStateChangeAndVoltage
 			&& v->dummy_vars.dml32_ModeSupportAndSystemConfigurationFull.MPCCombineMethodAsPossible;
 
-	for (i = 0; i < v->soc.num_states; i++) {
+	for (i = start_state; i < v->soc.num_states; i++) {
 		for (j = 0; j < 2; j++) {
 			mode_lib->vba.TotalNumberOfActiveDPP[i][j] = 0;
 			mode_lib->vba.TotalAvailablePipesSupport[i][j] = true;
@@ -2289,7 +2298,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 		}
 	}
 
-	for (i = 0; i < v->soc.num_states; ++i) {
+	for (i = start_state; i < v->soc.num_states; ++i) {
 		mode_lib->vba.ExceededMultistreamSlots[i] = false;
 		for (k = 0; k < mode_lib->vba.NumberOfActiveSurfaces; ++k) {
 			if (mode_lib->vba.OutputMultistreamEn[k] == true && mode_lib->vba.OutputMultistreamId[k] == k) {
@@ -2389,7 +2398,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 		}
 	}
 
-	for (i = 0; i < v->soc.num_states; ++i) {
+	for (i = start_state; i < v->soc.num_states; ++i) {
 		mode_lib->vba.DTBCLKRequiredMoreThanSupported[i] = false;
 		for (k = 0; k < mode_lib->vba.NumberOfActiveSurfaces; ++k) {
 			if (mode_lib->vba.BlendingAndTiming[k] == k
@@ -2406,7 +2415,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 		}
 	}
 
-	for (i = 0; i < v->soc.num_states; ++i) {
+	for (i = start_state; i < v->soc.num_states; ++i) {
 		mode_lib->vba.ODMCombine2To1SupportCheckOK[i] = true;
 		mode_lib->vba.ODMCombine4To1SupportCheckOK[i] = true;
 		for (k = 0; k < mode_lib->vba.NumberOfActiveSurfaces; ++k) {
@@ -2424,7 +2433,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 		}
 	}
 
-	for (i = 0; i < v->soc.num_states; i++) {
+	for (i = start_state; i < v->soc.num_states; i++) {
 		mode_lib->vba.DSCCLKRequiredMoreThanSupported[i] = false;
 		for (k = 0; k <= mode_lib->vba.NumberOfActiveSurfaces - 1; k++) {
 			if (mode_lib->vba.BlendingAndTiming[k] == k) {
@@ -2461,7 +2470,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 	/* Check DSC Unit and Slices Support */
 	v->dummy_vars.dml32_ModeSupportAndSystemConfigurationFull.TotalDSCUnitsRequired = 0;
 
-	for (i = 0; i < v->soc.num_states; ++i) {
+	for (i = start_state; i < v->soc.num_states; ++i) {
 		mode_lib->vba.NotEnoughDSCUnits[i] = false;
 		mode_lib->vba.NotEnoughDSCSlices[i] = false;
 		v->dummy_vars.dml32_ModeSupportAndSystemConfigurationFull.TotalDSCUnitsRequired = 0;
@@ -2496,7 +2505,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 	}
 
 	/*DSC Delay per state*/
-	for (i = 0; i < v->soc.num_states; ++i) {
+	for (i = start_state; i < v->soc.num_states; ++i) {
 		for (k = 0; k < mode_lib->vba.NumberOfActiveSurfaces; ++k) {
 			mode_lib->vba.DSCDelayPerState[i][k] = dml32_DSCDelayRequirement(
 					mode_lib->vba.RequiresDSC[i][k], mode_lib->vba.ODMCombineEnablePerState[i][k],
@@ -2523,7 +2532,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 
 	//Calculate Swath, DET Configuration, DCFCLKDeepSleep
 	//
-	for (i = 0; i < (int) v->soc.num_states; ++i) {
+	for (i = start_state; i < (int) v->soc.num_states; ++i) {
 		for (j = 0; j <= 1; ++j) {
 			for (k = 0; k < mode_lib->vba.NumberOfActiveSurfaces; ++k) {
 				mode_lib->vba.RequiredDPPCLKThisState[k] = mode_lib->vba.RequiredDPPCLK[i][j][k];
@@ -2661,7 +2670,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 			mode_lib->vba.SurfaceSizeInMALL,
 			&mode_lib->vba.ExceededMALLSize);
 
-	for (i = 0; i < v->soc.num_states; i++) {
+	for (i = start_state; i < v->soc.num_states; i++) {
 		for (j = 0; j < 2; j++) {
 			for (k = 0; k <= mode_lib->vba.NumberOfActiveSurfaces - 1; k++) {
 				mode_lib->vba.swath_width_luma_ub_this_state[k] =
@@ -2888,7 +2897,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 	}
 
 	//Calculate Return BW
-	for (i = 0; i < (int) v->soc.num_states; ++i) {
+	for (i = start_state; i < (int) v->soc.num_states; ++i) {
 		for (j = 0; j <= 1; ++j) {
 			for (k = 0; k <= mode_lib->vba.NumberOfActiveSurfaces - 1; k++) {
 				if (mode_lib->vba.BlendingAndTiming[k] == k) {
@@ -2967,7 +2976,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 			&mode_lib->vba.MinPrefetchMode,
 			&mode_lib->vba.MaxPrefetchMode);
 
-	for (i = 0; i < (int) v->soc.num_states; ++i) {
+	for (i = start_state; i < (int) v->soc.num_states; ++i) {
 		for (j = 0; j <= 1; ++j)
 			mode_lib->vba.DCFCLKState[i][j] = mode_lib->vba.DCFCLKPerState[i];
 	}
@@ -3089,7 +3098,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 				mode_lib->vba.DCFCLKState);
 	} // UseMinimumRequiredDCFCLK == true
 
-	for (i = 0; i < (int) v->soc.num_states; ++i) {
+	for (i = start_state; i < (int) v->soc.num_states; ++i) {
 		for (j = 0; j <= 1; ++j) {
 			mode_lib->vba.ReturnBWPerState[i][j] = dml32_get_return_bw_mbps(&mode_lib->vba.soc, i,
 					mode_lib->vba.HostVMEnable, mode_lib->vba.DCFCLKState[i][j],
@@ -3098,7 +3107,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 	}
 
 	//Re-ordering Buffer Support Check
-	for (i = 0; i < (int) v->soc.num_states; ++i) {
+	for (i = start_state; i < (int) v->soc.num_states; ++i) {
 		for (j = 0; j <= 1; ++j) {
 			if ((mode_lib->vba.ROBBufferSizeInKByte - mode_lib->vba.PixelChunkSizeInKByte) * 1024
 					/ mode_lib->vba.ReturnBWPerState[i][j]
@@ -3120,7 +3129,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 				+ mode_lib->vba.ReadBandwidthChroma[k];
 	}
 
-	for (i = 0; i < (int) v->soc.num_states; ++i) {
+	for (i = start_state; i < (int) v->soc.num_states; ++i) {
 		for (j = 0; j <= 1; ++j) {
 			mode_lib->vba.MaxTotalVerticalActiveAvailableBandwidth[i][j] =
 				dml_min3(mode_lib->vba.ReturnBusWidth * mode_lib->vba.DCFCLKState[i][j]
@@ -3144,7 +3153,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 
 	/* Prefetch Check */
 
-	for (i = 0; i < (int) v->soc.num_states; ++i) {
+	for (i = start_state; i < (int) v->soc.num_states; ++i) {
 		for (j = 0; j <= 1; ++j) {
 
 			mode_lib->vba.TimeCalc = 24 / mode_lib->vba.ProjectedDCFCLKDeepSleep[i][j];
@@ -3662,7 +3671,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 
 	MaximumMPCCombine = 0;
 
-	for (i = v->soc.num_states; i >= 0; i--) {
+	for (i = v->soc.num_states; i >= start_state; i--) {
 		if (i == v->soc.num_states || mode_lib->vba.ModeSupport[i][0] == true ||
 				mode_lib->vba.ModeSupport[i][1] == true) {
 			mode_lib->vba.VoltageLevel = i;

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

@@ -91,6 +91,7 @@ struct display_mode_lib {
 	struct dal_logger *logger;
 	struct dml_funcs funcs;
 	struct _vcs_dpi_display_e2e_pipe_params_st dml_pipe_state[6];
+	bool validate_max_state;
 };
 
 void dml_init_instance(struct display_mode_lib *lib,