drm/amd/display: fix dcn3.1x mode validation on high bandwidth config

[why]
1. correct dram_channel_width (was hard coded to 4 for 32bit)
2. use dm's is_hvm_enable status flag for hostvm_en input for dml.
3. add a function to override to all dcn3.1x.

Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Charlene Liu <Charlene.Liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

9 files changed, 90 insertions, 19 deletions

diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_clk_mgr.c
index c1eaf571407a..1c0569b1dc8f 100644
--- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_clk_mgr.c
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_clk_mgr.c
@@ -609,8 +609,10 @@ static void dcn31_clk_mgr_helper_populate_bw_params(struct clk_mgr_internal *clk
 	}
 
 	bw_params->vram_type = bios_info->memory_type;
-	bw_params->num_channels = bios_info->ma_channel_number;
 
+	bw_params->dram_channel_width_bytes = bios_info->memory_type == 0x22 ? 8 : 4;
+	//bw_params->dram_channel_width_bytes = dc->ctx->asic_id.vram_width;
+	bw_params->num_channels = bios_info->ma_channel_number ? bios_info->ma_channel_number : 4;
 	for (i = 0; i < WM_SET_COUNT; i++) {
 		bw_params->wm_table.entries[i].wm_inst = i;
 
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.c
index 187f5b27fdc8..3edc81e2d417 100644
--- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.c
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.c
@@ -553,6 +553,7 @@ static void dcn316_clk_mgr_helper_populate_bw_params(
 
 	bw_params->vram_type = bios_info->memory_type;
 	bw_params->num_channels = bios_info->ma_channel_number;
+	bw_params->dram_channel_width_bytes = bios_info->memory_type == 0x22 ? 8 : 4;
 
 	for (i = 0; i < WM_SET_COUNT; i++) {
 		bw_params->wm_table.entries[i].wm_inst = i;
diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_resource.c b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_resource.c
index 15e73f4d92ab..98d047233645 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_resource.c
@@ -1639,6 +1639,31 @@ static bool is_dual_plane(enum surface_pixel_format format)
 	return format >= SURFACE_PIXEL_FORMAT_VIDEO_BEGIN || format == SURFACE_PIXEL_FORMAT_GRPH_RGBE_ALPHA;
 }
 
+int dcn31x_populate_dml_pipes_from_context(struct dc *dc,
+					  struct dc_state *context,
+					  display_e2e_pipe_params_st *pipes,
+					  bool fast_validate)
+{
+	uint32_t pipe_cnt;
+	int i;
+
+	dc_assert_fp_enabled();
+
+	pipe_cnt = dcn20_populate_dml_pipes_from_context(dc, context, pipes, fast_validate);
+
+	for (i = 0; i < pipe_cnt; i++) {
+		pipes[i].pipe.src.gpuvm = 1;
+		if (dc->debug.dml_hostvm_override == DML_HOSTVM_NO_OVERRIDE) {
+			//pipes[pipe_cnt].pipe.src.hostvm = dc->res_pool->hubbub->riommu_active;
+			pipes[i].pipe.src.hostvm = dc->vm_pa_config.is_hvm_enabled;
+		} else if (dc->debug.dml_hostvm_override == DML_HOSTVM_OVERRIDE_FALSE)
+			pipes[i].pipe.src.hostvm = false;
+		else if (dc->debug.dml_hostvm_override == DML_HOSTVM_OVERRIDE_TRUE)
+			pipes[i].pipe.src.hostvm = true;
+	}
+	return pipe_cnt;
+}
+
 int dcn31_populate_dml_pipes_from_context(
 	struct dc *dc, struct dc_state *context,
 	display_e2e_pipe_params_st *pipes,
@@ -1650,7 +1675,7 @@ int dcn31_populate_dml_pipes_from_context(
 	bool upscaled = false;
 
 	DC_FP_START();
-	dcn20_populate_dml_pipes_from_context(dc, context, pipes, fast_validate);
+	dcn31x_populate_dml_pipes_from_context(dc, context, pipes, fast_validate);
 	DC_FP_END();
 
 	for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) {
@@ -1680,12 +1705,6 @@ int dcn31_populate_dml_pipes_from_context(
 		dcn31_zero_pipe_dcc_fraction(pipes, pipe_cnt);
 		DC_FP_END();
 
-		if (dc->debug.dml_hostvm_override == DML_HOSTVM_NO_OVERRIDE)
-			pipes[pipe_cnt].pipe.src.hostvm = dc->res_pool->hubbub->riommu_active;
-		else if (dc->debug.dml_hostvm_override == DML_HOSTVM_OVERRIDE_FALSE)
-			pipes[pipe_cnt].pipe.src.hostvm = false;
-		else if (dc->debug.dml_hostvm_override == DML_HOSTVM_OVERRIDE_TRUE)
-			pipes[pipe_cnt].pipe.src.hostvm = true;
 
 		if (pipes[pipe_cnt].dout.dsc_enable) {
 			switch (timing->display_color_depth) {
diff --git a/drivers/gpu/drm/amd/display/dc/dcn315/dcn315_resource.c b/drivers/gpu/drm/amd/display/dc/dcn315/dcn315_resource.c
index 2c68862b0d8e..07a22336b543 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn315/dcn315_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn315/dcn315_resource.c
@@ -1648,7 +1648,7 @@ static int dcn315_populate_dml_pipes_from_context(
 	const int max_usable_det = context->bw_ctx.dml.ip.config_return_buffer_size_in_kbytes - DCN3_15_MIN_COMPBUF_SIZE_KB;
 
 	DC_FP_START();
-	dcn20_populate_dml_pipes_from_context(dc, context, pipes, fast_validate);
+	dcn31x_populate_dml_pipes_from_context(dc, context, pipes, fast_validate);
 	DC_FP_END();
 
 	for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) {
@@ -1667,7 +1667,6 @@ static int dcn315_populate_dml_pipes_from_context(
 		pipes[pipe_cnt].pipe.src.immediate_flip = true;
 
 		pipes[pipe_cnt].pipe.src.unbounded_req_mode = false;
-		pipes[pipe_cnt].pipe.src.gpuvm = true;
 		pipes[pipe_cnt].pipe.dest.vfront_porch = timing->v_front_porch;
 		pipes[pipe_cnt].pipe.src.dcc_rate = 3;
 		pipes[pipe_cnt].dout.dsc_input_bpc = 0;
diff --git a/drivers/gpu/drm/amd/display/dc/dcn316/dcn316_resource.c b/drivers/gpu/drm/amd/display/dc/dcn316/dcn316_resource.c
index 2f643cdaf59f..9a82f8f66f28 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn316/dcn316_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn316/dcn316_resource.c
@@ -1650,7 +1650,7 @@ static int dcn316_populate_dml_pipes_from_context(
 	const int max_usable_det = context->bw_ctx.dml.ip.config_return_buffer_size_in_kbytes - DCN3_16_MIN_COMPBUF_SIZE_KB;
 
 	DC_FP_START();
-	dcn20_populate_dml_pipes_from_context(dc, context, pipes, fast_validate);
+	dcn31x_populate_dml_pipes_from_context(dc, context, pipes, fast_validate);
 	DC_FP_END();
 
 	for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) {
@@ -1669,7 +1669,6 @@ static int dcn316_populate_dml_pipes_from_context(
 		pipes[pipe_cnt].pipe.src.immediate_flip = true;
 
 		pipes[pipe_cnt].pipe.src.unbounded_req_mode = false;
-		pipes[pipe_cnt].pipe.src.gpuvm = true;
 		pipes[pipe_cnt].pipe.dest.vfront_porch = timing->v_front_porch;
 		pipes[pipe_cnt].pipe.src.dcc_rate = 3;
 		pipes[pipe_cnt].dout.dsc_input_bpc = 0;
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.h b/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.h
index fd58b2561ec9..b0b2ba9df57a 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.h
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.h
@@ -47,4 +47,8 @@ void dcn31_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params
 void dcn315_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params);
 void dcn316_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params);
 
+int dcn31x_populate_dml_pipes_from_context(struct dc *dc,
+					  struct dc_state *context,
+					  display_e2e_pipe_params_st *pipes,
+					  bool fast_validate);
 #endif /* __DCN31_FPU_H__*/
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn31/display_mode_vba_31.c b/drivers/gpu/drm/amd/display/dc/dml/dcn31/display_mode_vba_31.c
index 45ab0ce50860..4e45c6d9ecdc 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn31/display_mode_vba_31.c
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn31/display_mode_vba_31.c
@@ -1056,14 +1056,12 @@ static bool CalculatePrefetchSchedule(
 	prefetch_bw_pr = dml_min(1, myPipe->VRatio) * prefetch_bw_pr;
 	max_Tsw = dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC) * LineTime;
 	prefetch_sw_bytes = PrefetchSourceLinesY * swath_width_luma_ub * myPipe->BytePerPixelY + PrefetchSourceLinesC * swath_width_chroma_ub * myPipe->BytePerPixelC;
-	prefetch_bw_oto = dml_max(bytes_pp * myPipe->PixelClock / myPipe->DPPPerPlane, prefetch_sw_bytes / (dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC) * LineTime));
 	prefetch_bw_oto = dml_max(prefetch_bw_pr, prefetch_sw_bytes / max_Tsw);
 
 	min_Lsw = dml_max(1, dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC) / max_vratio_pre);
 	Lsw_oto = dml_ceil(4 * dml_max(prefetch_sw_bytes / prefetch_bw_oto / LineTime, min_Lsw), 1) / 4;
 	Tsw_oto = Lsw_oto * LineTime;
 
-	prefetch_bw_oto = (PrefetchSourceLinesY * swath_width_luma_ub * myPipe->BytePerPixelY + PrefetchSourceLinesC * swath_width_chroma_ub * myPipe->BytePerPixelC) / Tsw_oto;
 
 #ifdef __DML_VBA_DEBUG__
 	dml_print("DML: HTotal: %d\n", myPipe->HTotal);
@@ -5362,6 +5360,58 @@ void dml31_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 				v->ModeSupport[i][j] = true;
 			} else {
 				v->ModeSupport[i][j] = false;
+#ifdef __DML_VBA_DEBUG__
+				if (v->ScaleRatioAndTapsSupport == false)
+					dml_print("DML SUPPORT:     ScaleRatioAndTapsSupport failed");
+				if (v->SourceFormatPixelAndScanSupport == false)
+					dml_print("DML SUPPORT:     SourceFormatPixelAndScanSupport failed");
+				if (v->ViewportSizeSupport[i][j] == false)
+					dml_print("DML SUPPORT:     ViewportSizeSupport failed");
+				if (v->LinkCapacitySupport[i] == false)
+					dml_print("DML SUPPORT:     LinkCapacitySupport failed");
+				if (v->ODMCombine4To1SupportCheckOK[i] == false)
+					dml_print("DML SUPPORT:     DSC422NativeNotSupported failed");
+				if (v->NotEnoughDSCUnits[i] == true)
+					dml_print("DML SUPPORT:     NotEnoughDSCUnits");
+				if (v->DTBCLKRequiredMoreThanSupported[i] == true)
+					dml_print("DML SUPPORT:     DTBCLKRequiredMoreThanSupported");
+				if (v->ROBSupport[i][j] == false)
+					dml_print("DML SUPPORT:     ROBSupport failed");
+				if (v->DISPCLK_DPPCLK_Support[i][j] == false)
+					dml_print("DML SUPPORT:     DISPCLK_DPPCLK_Support failed");
+				if (v->TotalAvailablePipesSupport[i][j] == false)
+					dml_print("DML SUPPORT:     DSC422NativeNotSupported failed");
+				if (EnoughWritebackUnits == false)
+					dml_print("DML SUPPORT:     DSC422NativeNotSupported failed");
+				if (v->WritebackLatencySupport == false)
+					dml_print("DML SUPPORT:     WritebackLatencySupport failed");
+				if (v->WritebackScaleRatioAndTapsSupport == false)
+					dml_print("DML SUPPORT:     DSC422NativeNotSupported ");
+				if (v->CursorSupport == false)
+					dml_print("DML SUPPORT:     DSC422NativeNotSupported failed");
+				if (v->PitchSupport == false)
+					dml_print("DML SUPPORT:     PitchSupport failed");
+				if (ViewportExceedsSurface == true)
+					dml_print("DML SUPPORT:     ViewportExceedsSurface failed");
+				if (v->PrefetchSupported[i][j] == false)
+					dml_print("DML SUPPORT:     PrefetchSupported failed");
+				if (v->DynamicMetadataSupported[i][j] == false)
+					dml_print("DML SUPPORT:     DSC422NativeNotSupported failed");
+				if (v->TotalVerticalActiveBandwidthSupport[i][j] == false)
+					dml_print("DML SUPPORT:     TotalVerticalActiveBandwidthSupport failed");
+				if (v->VRatioInPrefetchSupported[i][j] == false)
+					dml_print("DML SUPPORT:     VRatioInPrefetchSupported failed");
+				if (v->PTEBufferSizeNotExceeded[i][j] == false)
+					dml_print("DML SUPPORT:     PTEBufferSizeNotExceeded failed");
+				if (v->NonsupportedDSCInputBPC == true)
+					dml_print("DML SUPPORT:     NonsupportedDSCInputBPC failed");
+				if (!((v->HostVMEnable == false
+					&& v->ImmediateFlipRequirement[0] != dm_immediate_flip_required)
+							|| v->ImmediateFlipSupportedForState[i][j] == true))
+					dml_print("DML SUPPORT:     ImmediateFlipRequirement failed");
+				if (FMTBufferExceeded == true)
+					dml_print("DML SUPPORT:     FMTBufferExceeded failed");
+#endif
 			}
 		}
 	}
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c
index 796c9d19e671..1dd51c4b6804 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c
@@ -29,6 +29,7 @@
 #include "dcn31/dcn31_hubbub.h"
 #include "dcn314_fpu.h"
 #include "dml/dcn20/dcn20_fpu.h"
+#include "dml/dcn31/dcn31_fpu.h"
 #include "dml/display_mode_vba.h"
 
 struct _vcs_dpi_ip_params_st dcn3_14_ip = {
@@ -288,7 +289,7 @@ int dcn314_populate_dml_pipes_from_context_fpu(struct dc *dc, struct dc_state *c
 
 	dc_assert_fp_enabled();
 
-	dcn20_populate_dml_pipes_from_context(dc, context, pipes, fast_validate);
+	dcn31x_populate_dml_pipes_from_context(dc, context, pipes, fast_validate);
 
 	for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) {
 		struct dc_crtc_timing *timing;
@@ -315,8 +316,6 @@ int dcn314_populate_dml_pipes_from_context_fpu(struct dc *dc, struct dc_state *c
 		pipes[pipe_cnt].pipe.src.immediate_flip = true;
 
 		pipes[pipe_cnt].pipe.src.unbounded_req_mode = false;
-		pipes[pipe_cnt].pipe.src.hostvm = dc->res_pool->hubbub->riommu_active;
-		pipes[pipe_cnt].pipe.src.gpuvm = true;
 		pipes[pipe_cnt].pipe.src.dcc_fraction_of_zs_req_luma = 0;
 		pipes[pipe_cnt].pipe.src.dcc_fraction_of_zs_req_chroma = 0;
 		pipes[pipe_cnt].pipe.dest.vfront_porch = timing->v_front_porch;
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn314/display_mode_vba_314.c b/drivers/gpu/drm/amd/display/dc/dml/dcn314/display_mode_vba_314.c
index 53e3e7364ec6..41f0b4c1c72f 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn314/display_mode_vba_314.c
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn314/display_mode_vba_314.c
@@ -1078,14 +1078,12 @@ static bool CalculatePrefetchSchedule(
 	prefetch_bw_pr = dml_min(1, myPipe->VRatio) * prefetch_bw_pr;
 	max_Tsw = dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC) * LineTime;
 	prefetch_sw_bytes = PrefetchSourceLinesY * swath_width_luma_ub * myPipe->BytePerPixelY + PrefetchSourceLinesC * swath_width_chroma_ub * myPipe->BytePerPixelC;
-	prefetch_bw_oto = dml_max(bytes_pp * myPipe->PixelClock / myPipe->DPPPerPlane, prefetch_sw_bytes / (dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC) * LineTime));
 	prefetch_bw_oto = dml_max(prefetch_bw_pr, prefetch_sw_bytes / max_Tsw);
 
 	min_Lsw = dml_max(1, dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC) / max_vratio_pre);
 	Lsw_oto = dml_ceil(4 * dml_max(prefetch_sw_bytes / prefetch_bw_oto / LineTime, min_Lsw), 1) / 4;
 	Tsw_oto = Lsw_oto * LineTime;
 
-	prefetch_bw_oto = (PrefetchSourceLinesY * swath_width_luma_ub * myPipe->BytePerPixelY + PrefetchSourceLinesC * swath_width_chroma_ub * myPipe->BytePerPixelC) / Tsw_oto;
 
 #ifdef __DML_VBA_DEBUG__
 	dml_print("DML: HTotal: %d\n", myPipe->HTotal);