drm/amd/display: Improve robustness of FIXED_VS link training at DP1 rates

[WHY]
New sequence for transparent mode DP1.x link training was provided by LTTPR
vendor

[HOW]
Implement new FIXED_VS sequence, increase LT retry count to minimize
any potential intermittent lightup failures

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Michael Strauss <michael.strauss@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

6 files changed, 396 insertions, 3 deletions

diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h
index f51b9c265b51..190e6a2e1334 100644
--- a/drivers/gpu/drm/amd/display/dc/dc.h
+++ b/drivers/gpu/drm/amd/display/dc/dc.h
@@ -405,6 +405,7 @@ struct dc_config {
 	bool force_bios_enable_lttpr;
 	uint8_t force_bios_fixed_vs;
 	int sdpif_request_limit_words_per_umc;
+	bool use_old_fixed_vs_sequence;
 	bool disable_subvp_drr;
 };
 
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 eaaa2e01f6d0..ff8cd5076434 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_resource.c
@@ -1965,6 +1965,8 @@ static bool dcn31_resource_construct(
 	dc->caps.color.mpc.ogam_rom_caps.hlg = 0;
 	dc->caps.color.mpc.ocsc = 1;
 
+	dc->config.use_old_fixed_vs_sequence = true;
+
 	/* Use pipe context based otg sync logic */
 	dc->config.use_pipe_ctx_sync_logic = true;
 
diff --git a/drivers/gpu/drm/amd/display/dc/link/link_dpms.c b/drivers/gpu/drm/amd/display/dc/link/link_dpms.c
index f6c5ee2d639b..027ad1f0144d 100644
--- a/drivers/gpu/drm/amd/display/dc/link/link_dpms.c
+++ b/drivers/gpu/drm/amd/display/dc/link/link_dpms.c
@@ -2035,6 +2035,12 @@ static enum dc_status enable_link_dp(struct dc_state *state,
 	uint32_t post_oui_delay = 30; // 30ms
 	/* Reduce link bandwidth between failed link training attempts. */
 	bool do_fallback = false;
+	int lt_attempts = LINK_TRAINING_ATTEMPTS;
+
+	// Increase retry count if attempting DP1.x on FIXED_VS link
+	if ((link->chip_caps & EXT_DISPLAY_PATH_CAPS__DP_FIXED_VS_EN) &&
+			link_dp_get_encoding_format(link_settings) == DP_8b_10b_ENCODING)
+		lt_attempts = 10;
 
 	// check for seamless boot
 	for (i = 0; i < state->stream_count; i++) {
@@ -2099,7 +2105,7 @@ static enum dc_status enable_link_dp(struct dc_state *state,
 
 	if (perform_link_training_with_retries(link_settings,
 					       skip_video_pattern,
-					       LINK_TRAINING_ATTEMPTS,
+					       lt_attempts,
 					       pipe_ctx,
 					       pipe_ctx->stream->signal,
 					       do_fallback)) {
diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c
index f301c9eaf2f9..70fc0ddf2d7e 100644
--- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c
+++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c
@@ -1496,7 +1496,10 @@ enum link_training_result dp_perform_link_training(
 	 * Non-LT AUX transactions inside training mode.
 	 */
 	if ((link->chip_caps & EXT_DISPLAY_PATH_CAPS__DP_FIXED_VS_EN) && encoding == DP_8b_10b_ENCODING)
-		status = dp_perform_fixed_vs_pe_training_sequence(link, link_res, &lt_settings);
+		if (link->dc->config.use_old_fixed_vs_sequence)
+			status = dp_perform_fixed_vs_pe_training_sequence_legacy(link, link_res, &lt_settings);
+		else
+			status = dp_perform_fixed_vs_pe_training_sequence(link, link_res, &lt_settings);
 	else if (encoding == DP_8b_10b_ENCODING)
 		status = dp_perform_8b_10b_link_training(link, link_res, &lt_settings);
 	else if (encoding == DP_128b_132b_ENCODING)
diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c
index a4071d2959a0..5731c4b61f9f 100644
--- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c
+++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c
@@ -223,7 +223,7 @@ static enum link_training_result perform_fixed_vs_pe_nontransparent_training_seq
 }
 
 
-enum link_training_result dp_perform_fixed_vs_pe_training_sequence(
+enum link_training_result dp_perform_fixed_vs_pe_training_sequence_legacy(
 	struct dc_link *link,
 	const struct link_resource *link_res,
 	struct link_training_settings *lt_settings)
@@ -577,3 +577,379 @@ enum link_training_result dp_perform_fixed_vs_pe_training_sequence(
 
 	return status;
 }
+
+enum link_training_result dp_perform_fixed_vs_pe_training_sequence(
+	struct dc_link *link,
+	const struct link_resource *link_res,
+	struct link_training_settings *lt_settings)
+{
+	const uint8_t vendor_lttpr_write_data_reset[4] = {0x1, 0x50, 0x63, 0xFF};
+	const uint8_t offset = dp_parse_lttpr_repeater_count(
+			link->dpcd_caps.lttpr_caps.phy_repeater_cnt);
+	const uint8_t vendor_lttpr_write_data_intercept_en[4] = {0x1, 0x55, 0x63, 0x0};
+	const uint8_t vendor_lttpr_write_data_intercept_dis[4] = {0x1, 0x55, 0x63, 0x6E};
+	const uint8_t vendor_lttpr_write_data_adicora_eq1[4] = {0x1, 0x55, 0x63, 0x2E};
+	const uint8_t vendor_lttpr_write_data_adicora_eq2[4] = {0x1, 0x55, 0x63, 0x01};
+	const uint8_t vendor_lttpr_write_data_adicora_eq3[4] = {0x1, 0x55, 0x63, 0x68};
+	uint32_t pre_disable_intercept_delay_ms = link->dc->debug.fixed_vs_aux_delay_config_wa;
+	uint8_t vendor_lttpr_write_data_vs[4] = {0x1, 0x51, 0x63, 0x0};
+	uint8_t vendor_lttpr_write_data_pe[4] = {0x1, 0x52, 0x63, 0x0};
+
+	uint32_t vendor_lttpr_write_address = 0xF004F;
+	enum link_training_result status = LINK_TRAINING_SUCCESS;
+	uint8_t lane = 0;
+	union down_spread_ctrl downspread = {0};
+	union lane_count_set lane_count_set = {0};
+	uint8_t toggle_rate;
+	uint8_t rate;
+
+	/* Only 8b/10b is supported */
+	ASSERT(link_dp_get_encoding_format(&lt_settings->link_settings) ==
+			DP_8b_10b_ENCODING);
+
+	if (lt_settings->lttpr_mode == LTTPR_MODE_NON_TRANSPARENT) {
+		status = perform_fixed_vs_pe_nontransparent_training_sequence(link, link_res, lt_settings);
+		return status;
+	}
+
+	if (offset != 0xFF) {
+		vendor_lttpr_write_address +=
+				((DP_REPEATER_CONFIGURATION_AND_STATUS_SIZE) * (offset - 1));
+
+		/* Certain display and cable configuration require extra delay */
+		if (offset > 2)
+			pre_disable_intercept_delay_ms = link->dc->debug.fixed_vs_aux_delay_config_wa * 2;
+	}
+
+	/* Vendor specific: Reset lane settings */
+	core_link_write_dpcd(
+			link,
+			vendor_lttpr_write_address,
+			&vendor_lttpr_write_data_reset[0],
+			sizeof(vendor_lttpr_write_data_reset));
+	core_link_write_dpcd(
+			link,
+			vendor_lttpr_write_address,
+			&vendor_lttpr_write_data_vs[0],
+			sizeof(vendor_lttpr_write_data_vs));
+	core_link_write_dpcd(
+			link,
+			vendor_lttpr_write_address,
+			&vendor_lttpr_write_data_pe[0],
+			sizeof(vendor_lttpr_write_data_pe));
+
+	/* Vendor specific: Enable intercept */
+	core_link_write_dpcd(
+			link,
+			vendor_lttpr_write_address,
+			&vendor_lttpr_write_data_intercept_en[0],
+			sizeof(vendor_lttpr_write_data_intercept_en));
+
+	/* 1. set link rate, lane count and spread. */
+
+	downspread.raw = (uint8_t)(lt_settings->link_settings.link_spread);
+
+	lane_count_set.bits.LANE_COUNT_SET =
+	lt_settings->link_settings.lane_count;
+
+	lane_count_set.bits.ENHANCED_FRAMING = lt_settings->enhanced_framing;
+	lane_count_set.bits.POST_LT_ADJ_REQ_GRANTED = 0;
+
+
+	if (lt_settings->pattern_for_eq < DP_TRAINING_PATTERN_SEQUENCE_4) {
+		lane_count_set.bits.POST_LT_ADJ_REQ_GRANTED =
+				link->dpcd_caps.max_ln_count.bits.POST_LT_ADJ_REQ_SUPPORTED;
+	}
+
+	core_link_write_dpcd(link, DP_DOWNSPREAD_CTRL,
+		&downspread.raw, sizeof(downspread));
+
+	core_link_write_dpcd(link, DP_LANE_COUNT_SET,
+		&lane_count_set.raw, 1);
+
+	rate = get_dpcd_link_rate(&lt_settings->link_settings);
+
+	/* Vendor specific: Toggle link rate */
+	toggle_rate = (rate == 0x6) ? 0xA : 0x6;
+
+	if (link->vendor_specific_lttpr_link_rate_wa == rate) {
+		core_link_write_dpcd(
+				link,
+				DP_LINK_BW_SET,
+				&toggle_rate,
+				1);
+	}
+
+	link->vendor_specific_lttpr_link_rate_wa = rate;
+
+	core_link_write_dpcd(link, DP_LINK_BW_SET, &rate, 1);
+
+	DC_LOG_HW_LINK_TRAINING("%s\n %x rate = %x\n %x lane = %x framing = %x\n %x spread = %x\n",
+		__func__,
+		DP_LINK_BW_SET,
+		lt_settings->link_settings.link_rate,
+		DP_LANE_COUNT_SET,
+		lt_settings->link_settings.lane_count,
+		lt_settings->enhanced_framing,
+		DP_DOWNSPREAD_CTRL,
+		lt_settings->link_settings.link_spread);
+
+	/* 2. Perform link training */
+
+	/* Perform Clock Recovery Sequence */
+	if (status == LINK_TRAINING_SUCCESS) {
+		const uint8_t max_vendor_dpcd_retries = 10;
+		uint32_t retries_cr;
+		uint32_t retry_count;
+		uint32_t wait_time_microsec;
+		enum dc_lane_count lane_count = lt_settings->link_settings.lane_count;
+		union lane_status dpcd_lane_status[LANE_COUNT_DP_MAX];
+		union lane_align_status_updated dpcd_lane_status_updated;
+		union lane_adjust dpcd_lane_adjust[LANE_COUNT_DP_MAX] = {0};
+		enum dc_status dpcd_status = DC_OK;
+		uint8_t i = 0;
+
+		retries_cr = 0;
+		retry_count = 0;
+
+		memset(&dpcd_lane_status, '\0', sizeof(dpcd_lane_status));
+		memset(&dpcd_lane_status_updated, '\0',
+		sizeof(dpcd_lane_status_updated));
+
+		while ((retries_cr < LINK_TRAINING_MAX_RETRY_COUNT) &&
+			(retry_count < LINK_TRAINING_MAX_CR_RETRY)) {
+
+
+			/* 1. call HWSS to set lane settings */
+			dp_set_hw_lane_settings(
+					link,
+					link_res,
+					lt_settings,
+					0);
+
+			/* 2. update DPCD of the receiver */
+			if (!retry_count) {
+				/* EPR #361076 - write as a 5-byte burst,
+				 * but only for the 1-st iteration.
+				 */
+				dpcd_set_lt_pattern_and_lane_settings(
+						link,
+						lt_settings,
+						lt_settings->pattern_for_cr,
+						0);
+				/* Vendor specific: Disable intercept */
+				for (i = 0; i < max_vendor_dpcd_retries; i++) {
+					msleep(pre_disable_intercept_delay_ms);
+					dpcd_status = core_link_write_dpcd(
+							link,
+							vendor_lttpr_write_address,
+							&vendor_lttpr_write_data_intercept_dis[0],
+							sizeof(vendor_lttpr_write_data_intercept_dis));
+
+					if (dpcd_status == DC_OK)
+						break;
+
+					core_link_write_dpcd(
+							link,
+							vendor_lttpr_write_address,
+							&vendor_lttpr_write_data_intercept_en[0],
+							sizeof(vendor_lttpr_write_data_intercept_en));
+				}
+			} else {
+				vendor_lttpr_write_data_vs[3] = 0;
+				vendor_lttpr_write_data_pe[3] = 0;
+
+				for (lane = 0; lane < lane_count; lane++) {
+					vendor_lttpr_write_data_vs[3] |=
+							lt_settings->dpcd_lane_settings[lane].bits.VOLTAGE_SWING_SET << (2 * lane);
+					vendor_lttpr_write_data_pe[3] |=
+							lt_settings->dpcd_lane_settings[lane].bits.PRE_EMPHASIS_SET << (2 * lane);
+				}
+
+				/* Vendor specific: Update VS and PE to DPRX requested value */
+				core_link_write_dpcd(
+						link,
+						vendor_lttpr_write_address,
+						&vendor_lttpr_write_data_vs[0],
+						sizeof(vendor_lttpr_write_data_vs));
+				core_link_write_dpcd(
+						link,
+						vendor_lttpr_write_address,
+						&vendor_lttpr_write_data_pe[0],
+						sizeof(vendor_lttpr_write_data_pe));
+
+				dpcd_set_lane_settings(
+						link,
+						lt_settings,
+						0);
+			}
+
+			/* 3. wait receiver to lock-on*/
+			wait_time_microsec = lt_settings->cr_pattern_time;
+
+			dp_wait_for_training_aux_rd_interval(
+					link,
+					wait_time_microsec);
+
+			/* 4. Read lane status and requested drive
+			 * settings as set by the sink
+			 */
+			dp_get_lane_status_and_lane_adjust(
+					link,
+					lt_settings,
+					dpcd_lane_status,
+					&dpcd_lane_status_updated,
+					dpcd_lane_adjust,
+					0);
+
+			/* 5. check CR done*/
+			if (dp_is_cr_done(lane_count, dpcd_lane_status)) {
+				status = LINK_TRAINING_SUCCESS;
+				break;
+			}
+
+			/* 6. max VS reached*/
+			if (dp_is_max_vs_reached(lt_settings))
+				break;
+
+			/* 7. same lane settings */
+			/* Note: settings are the same for all lanes,
+			 * so comparing first lane is sufficient
+			 */
+			if (lt_settings->dpcd_lane_settings[0].bits.VOLTAGE_SWING_SET ==
+					dpcd_lane_adjust[0].bits.VOLTAGE_SWING_LANE)
+				retries_cr++;
+			else
+				retries_cr = 0;
+
+			/* 8. update VS/PE/PC2 in lt_settings*/
+			dp_decide_lane_settings(lt_settings, dpcd_lane_adjust,
+					lt_settings->hw_lane_settings, lt_settings->dpcd_lane_settings);
+			retry_count++;
+		}
+
+		if (retry_count >= LINK_TRAINING_MAX_CR_RETRY) {
+			ASSERT(0);
+			DC_LOG_ERROR("%s: Link Training Error, could not get CR after %d tries. Possibly voltage swing issue",
+				__func__,
+				LINK_TRAINING_MAX_CR_RETRY);
+
+		}
+
+		status = dp_get_cr_failure(lane_count, dpcd_lane_status);
+	}
+
+	/* Perform Channel EQ Sequence */
+	if (status == LINK_TRAINING_SUCCESS) {
+		enum dc_dp_training_pattern tr_pattern;
+		uint32_t retries_ch_eq;
+		uint32_t wait_time_microsec;
+		enum dc_lane_count lane_count = lt_settings->link_settings.lane_count;
+		union lane_align_status_updated dpcd_lane_status_updated = {0};
+		union lane_status dpcd_lane_status[LANE_COUNT_DP_MAX] = {0};
+		union lane_adjust dpcd_lane_adjust[LANE_COUNT_DP_MAX] = {0};
+
+		core_link_write_dpcd(
+				link,
+				vendor_lttpr_write_address,
+				&vendor_lttpr_write_data_adicora_eq1[0],
+				sizeof(vendor_lttpr_write_data_adicora_eq1));
+		core_link_write_dpcd(
+				link,
+				vendor_lttpr_write_address,
+				&vendor_lttpr_write_data_adicora_eq2[0],
+				sizeof(vendor_lttpr_write_data_adicora_eq2));
+
+		/* Note: also check that TPS4 is a supported feature*/
+		tr_pattern = lt_settings->pattern_for_eq;
+
+		dp_set_hw_training_pattern(link, link_res, tr_pattern, 0);
+
+		status = LINK_TRAINING_EQ_FAIL_EQ;
+
+		for (retries_ch_eq = 0; retries_ch_eq <= LINK_TRAINING_MAX_RETRY_COUNT;
+			retries_ch_eq++) {
+
+			dp_set_hw_lane_settings(link, link_res, lt_settings, 0);
+
+			vendor_lttpr_write_data_vs[3] = 0;
+			vendor_lttpr_write_data_pe[3] = 0;
+
+			for (lane = 0; lane < lane_count; lane++) {
+				vendor_lttpr_write_data_vs[3] |=
+						lt_settings->dpcd_lane_settings[lane].bits.VOLTAGE_SWING_SET << (2 * lane);
+				vendor_lttpr_write_data_pe[3] |=
+						lt_settings->dpcd_lane_settings[lane].bits.PRE_EMPHASIS_SET << (2 * lane);
+			}
+
+			/* Vendor specific: Update VS and PE to DPRX requested value */
+			core_link_write_dpcd(
+					link,
+					vendor_lttpr_write_address,
+					&vendor_lttpr_write_data_vs[0],
+					sizeof(vendor_lttpr_write_data_vs));
+			core_link_write_dpcd(
+					link,
+					vendor_lttpr_write_address,
+					&vendor_lttpr_write_data_pe[0],
+					sizeof(vendor_lttpr_write_data_pe));
+
+			/* 2. update DPCD*/
+			if (!retries_ch_eq) {
+				/* EPR #361076 - write as a 5-byte burst,
+				 * but only for the 1-st iteration
+				 */
+
+				dpcd_set_lt_pattern_and_lane_settings(
+					link,
+					lt_settings,
+					tr_pattern, 0);
+
+				core_link_write_dpcd(
+					link,
+					vendor_lttpr_write_address,
+					&vendor_lttpr_write_data_adicora_eq3[0],
+					sizeof(vendor_lttpr_write_data_adicora_eq3));
+			} else
+				dpcd_set_lane_settings(link, lt_settings, 0);
+
+			/* 3. wait for receiver to lock-on*/
+			wait_time_microsec = lt_settings->eq_pattern_time;
+
+			dp_wait_for_training_aux_rd_interval(
+					link,
+					wait_time_microsec);
+
+			/* 4. Read lane status and requested
+			 * drive settings as set by the sink
+			 */
+			dp_get_lane_status_and_lane_adjust(
+				link,
+				lt_settings,
+				dpcd_lane_status,
+				&dpcd_lane_status_updated,
+				dpcd_lane_adjust,
+				0);
+
+			/* 5. check CR done*/
+			if (!dp_is_cr_done(lane_count, dpcd_lane_status)) {
+				status = LINK_TRAINING_EQ_FAIL_CR;
+				break;
+			}
+
+			/* 6. check CHEQ done*/
+			if (dp_is_ch_eq_done(lane_count, dpcd_lane_status) &&
+					dp_is_symbol_locked(lane_count, dpcd_lane_status) &&
+					dp_is_interlane_aligned(dpcd_lane_status_updated)) {
+				status = LINK_TRAINING_SUCCESS;
+				break;
+			}
+
+			/* 7. update VS/PE/PC2 in lt_settings*/
+			dp_decide_lane_settings(lt_settings, dpcd_lane_adjust,
+					lt_settings->hw_lane_settings, lt_settings->dpcd_lane_settings);
+		}
+	}
+
+	return status;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.h b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.h
index e61970e27661..c0d6ea329504 100644
--- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.h
+++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.h
@@ -28,6 +28,11 @@
 #define __DC_LINK_DP_FIXED_VS_PE_RETIMER_H__
 #include "link_dp_training.h"
 
+enum link_training_result dp_perform_fixed_vs_pe_training_sequence_legacy(
+	struct dc_link *link,
+	const struct link_resource *link_res,
+	struct link_training_settings *lt_settings);
+
 enum link_training_result dp_perform_fixed_vs_pe_training_sequence(
 	struct dc_link *link,
 	const struct link_resource *link_res,