drm/amd/display: fix dcn pipe reset sequence

This change fixes dcn10 front end reset sequence. Previously we
would reset front end during flip which led to issues
in certain MPO and 4k/5k scenarios. We would also never properly
power gate our front end.

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

13 files changed, 155 insertions, 175 deletions

diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_types.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_types.c
index 3711176bd070..a6f27ceaf4af 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_types.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_types.c
@@ -2380,7 +2380,7 @@ static void amdgpu_dm_do_flip(
 	surface_updates->flip_addr = &addr;
 
 
-	dc_update_surfaces_for_stream(adev->dm.dc, surface_updates, 1, acrtc->stream);
+	dc_update_surfaces_and_stream(adev->dm.dc, surface_updates, 1, acrtc->stream, NULL);
 
 	DRM_DEBUG_DRIVER("%s Flipping to hi: 0x%x, low: 0x%x \n",
 			 __func__,
diff --git a/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c b/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c
index 3ec702fecfd1..00961bc37200 100644
--- a/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c
+++ b/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c
@@ -511,12 +511,14 @@ static void split_stream_across_pipes(
 		struct pipe_ctx *primary_pipe,
 		struct pipe_ctx *secondary_pipe)
 {
+	int pipe_idx = secondary_pipe->pipe_idx;
+
 	if (!primary_pipe->surface)
 		return;
 
-	secondary_pipe->stream = primary_pipe->stream;
-	secondary_pipe->tg = primary_pipe->tg;
+	*secondary_pipe = *primary_pipe;
 
+	secondary_pipe->pipe_idx = pipe_idx;
 	secondary_pipe->mpcc = pool->mpcc[secondary_pipe->pipe_idx];
 	secondary_pipe->mi = pool->mis[secondary_pipe->pipe_idx];
 	secondary_pipe->ipp = pool->ipps[secondary_pipe->pipe_idx];
@@ -528,8 +530,6 @@ static void split_stream_across_pipes(
 	}
 	primary_pipe->bottom_pipe = secondary_pipe;
 	secondary_pipe->top_pipe = primary_pipe;
-	secondary_pipe->surface = primary_pipe->surface;
-	secondary_pipe->pipe_dlg_param = primary_pipe->pipe_dlg_param;
 
 	resource_build_scaling_params(primary_pipe);
 	resource_build_scaling_params(secondary_pipe);
@@ -1011,10 +1011,13 @@ bool dcn_validate_bandwidth(
 					dcn_bw_calc_rq_dlg_ttu(dc, v, hsplit_pipe);
 				} else if (hsplit_pipe && hsplit_pipe->surface == pipe->surface) {
 					/* merge previously split pipe */
-					if (pipe->bottom_pipe->bottom_pipe)
-						pipe->bottom_pipe->bottom_pipe->top_pipe = pipe;
-					memset(pipe->bottom_pipe, 0, sizeof(*pipe->bottom_pipe));
-					pipe->bottom_pipe = pipe->bottom_pipe->bottom_pipe;
+					pipe->bottom_pipe = hsplit_pipe->bottom_pipe;
+					if (hsplit_pipe->bottom_pipe)
+						hsplit_pipe->bottom_pipe->top_pipe = pipe;
+					hsplit_pipe->surface = NULL;
+					hsplit_pipe->stream = NULL;
+					hsplit_pipe->top_pipe = NULL;
+					hsplit_pipe->bottom_pipe = NULL;
 					resource_build_scaling_params(pipe);
 				}
 				/* for now important to do this after pipe split for building e2e params */
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c
index 18fde5081742..2f481ef55490 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc.c
@@ -963,11 +963,8 @@ bool dc_post_update_surfaces_to_stream(struct dc *dc)
 
 	for (i = 0; i < core_dc->res_pool->pipe_count; i++)
 		if (context->res_ctx.pipe_ctx[i].stream == NULL
-				|| context->res_ctx.pipe_ctx[i].surface == NULL) {
-			context->res_ctx.pipe_ctx[i].pipe_idx = i;
-			core_dc->hwss.power_down_front_end(
-					core_dc, &context->res_ctx.pipe_ctx[i]);
-		}
+				|| context->res_ctx.pipe_ctx[i].surface == NULL)
+			core_dc->hwss.power_down_front_end(core_dc, i);
 
 	/* 3rd param should be true, temp w/a for RV*/
 #if defined(CONFIG_DRM_AMD_DC_DCN1_0)
@@ -989,7 +986,6 @@ bool dc_commit_surfaces_to_stream(
 	struct dc_plane_info plane_info[MAX_SURFACES];
 	struct dc_scaling_info scaling_info[MAX_SURFACES];
 	int i;
-	bool ret;
 	struct dc_stream_update *stream_update =
 			dm_alloc(sizeof(struct dc_stream_update));
 
@@ -1038,10 +1034,10 @@ bool dc_commit_surfaces_to_stream(
 			new_surface_count,
 			dc_stream, stream_update);
 
-	ret = dc_post_update_surfaces_to_stream(dc);
+	dc_post_update_surfaces_to_stream(dc);
 
 	dm_free(stream_update);
-	return ret;
+	return true;
 }
 
 static bool is_surface_in_context(
@@ -1217,14 +1213,6 @@ enum surface_update_type dc_check_update_surfaces_for_stream(
 	return overall_type;
 }
 
-void dc_update_surfaces_for_stream(struct dc *dc,
-		struct dc_surface_update *surface_updates, int surface_count,
-		const struct dc_stream *dc_stream)
-{
-	dc_update_surfaces_and_stream(dc, surface_updates, surface_count,
-			dc_stream, NULL);
-}
-
 enum surface_update_type update_surface_trace_level = UPDATE_TYPE_FULL;
 
 void dc_update_surfaces_and_stream(struct dc *dc,
@@ -1401,7 +1389,7 @@ void dc_update_surfaces_and_stream(struct dc *dc,
 		}
 	}
 
-	if (!surface_count)  /* reset */
+	if (surface_count == 0)
 		core_dc->hwss.apply_ctx_for_surface(core_dc, NULL, context);
 
 	/* Lock pipes for provided surfaces, or all active if full update*/
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
index f0183d5f0d0d..fb033b69e2bf 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
@@ -1096,6 +1096,10 @@ bool resource_attach_surfaces_to_context(
 		free_pipe->surface = surface;
 
 		if (tail_pipe) {
+			free_pipe->tg = tail_pipe->tg;
+			free_pipe->stream_enc = tail_pipe->stream_enc;
+			free_pipe->audio = tail_pipe->audio;
+			free_pipe->clock_source = tail_pipe->clock_source;
 			free_pipe->top_pipe = tail_pipe;
 			tail_pipe->bottom_pipe = free_pipe;
 		}
@@ -2300,6 +2304,9 @@ bool pipe_need_reprogram(
 		struct pipe_ctx *pipe_ctx_old,
 		struct pipe_ctx *pipe_ctx)
 {
+	if (!pipe_ctx_old->stream)
+		return false;
+
 	if (pipe_ctx_old->stream->sink != pipe_ctx->stream->sink)
 		return true;
 
diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h
index cb70b6d8be10..18f6858e4992 100644
--- a/drivers/gpu/drm/amd/display/dc/dc.h
+++ b/drivers/gpu/drm/amd/display/dc/dc.h
@@ -415,9 +415,6 @@ bool dc_commit_surfaces_to_stream(
 bool dc_post_update_surfaces_to_stream(
 		struct dc *dc);
 
-void dc_update_surfaces_for_stream(struct dc *dc, struct dc_surface_update *updates,
-		int surface_count, const struct dc_stream *stream);
-
 /* Surface update type is used by dc_update_surfaces_and_stream
  * The update type is determined at the very beginning of the function based
  * on parameters passed in and decides how much programming (or updating) is
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
index 0bab85b27a85..fd1cb7fe45bc 100644
--- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
@@ -1185,8 +1185,7 @@ static void disable_vga_and_power_gate_all_controllers(
 		enable_display_pipe_clock_gating(ctx,
 				true);
 
-		dc->hwss.power_down_front_end(
-			dc, &dc->current_context->res_ctx.pipe_ctx[i]);
+		dc->hwss.power_down_front_end(dc, i);
 	}
 }
 
@@ -1340,7 +1339,7 @@ static void reset_single_pipe_hw_ctx(
 	resource_unreference_clock_source(&context->res_ctx, dc->res_pool,
 			 &pipe_ctx->clock_source);
 
-	dc->hwss.power_down_front_end((struct core_dc *)dc, pipe_ctx);
+	dc->hwss.power_down_front_end((struct core_dc *)dc, pipe_ctx->pipe_idx);
 
 	pipe_ctx->stream = NULL;
 }
@@ -2538,17 +2537,17 @@ static void dce110_apply_ctx_for_surface(
 	}
 }
 
-static void dce110_power_down_fe(struct core_dc *dc, struct pipe_ctx *pipe)
+static void dce110_power_down_fe(struct core_dc *dc, int fe_idx)
 {
 	/* Do not power down fe when stream is active on dce*/
-	if (pipe->stream)
+	if (dc->current_context->res_ctx.pipe_ctx[fe_idx].stream)
 		return;
 
 	dc->hwss.enable_display_power_gating(
-		dc, pipe->pipe_idx, dc->ctx->dc_bios, PIPE_GATING_CONTROL_ENABLE);
-	if (pipe->xfm)
-		pipe->xfm->funcs->transform_reset(pipe->xfm);
-	memset(&pipe->scl_data, 0, sizeof(struct scaler_data));
+		dc, fe_idx, dc->ctx->dc_bios, PIPE_GATING_CONTROL_ENABLE);
+
+	dc->res_pool->transforms[fe_idx]->funcs->transform_reset(
+				dc->res_pool->transforms[fe_idx]);
 }
 
 static const struct hw_sequencer_funcs dce110_funcs = {
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
index edcd7369cf9d..d0bddfd64c67 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
@@ -116,18 +116,6 @@ static const struct dcn10_hwseq_reg_offsets reg_offsets[] = {
 		HWSEQ_REG_SET_N(reg, 1, FD(reg##__##field), val)
 
 /* TODO should be moved to OTG */
-static void lock_otg_master_update(
-	struct dc_context *ctx,
-	uint8_t inst)
-{
-	uint32_t inst_offset = reg_offsets[inst].otg;
-
-	HWSEQ_REG_UPDATE(OTG0_OTG_GLOBAL_CONTROL0,
-			OTG_MASTER_UPDATE_LOCK_SEL, inst);
-
-	HWSEQ_REG_UPDATE(OTG0_OTG_MASTER_UPDATE_LOCK,
-			OTG_MASTER_UPDATE_LOCK, 1);
-}
 
 static bool unlock_master_tg_and_wait(
 	struct dc_context *ctx,
@@ -148,19 +136,6 @@ static bool unlock_master_tg_and_wait(
 	return true;
 }
 
-/* TODO: should be  moved to OTG ? */
-static void unlock_otg_master(
-	struct dc_context *ctx,
-	uint8_t inst)
-{
-	uint32_t inst_offset = reg_offsets[inst].otg;
-
-	/* unlock master locker */
-	HWSEQ_REG_UPDATE(OTG0_OTG_MASTER_UPDATE_LOCK,
-			OTG_MASTER_UPDATE_LOCK, 0);
-}
-
-
 static void wait_no_outstanding_request(
 	struct dc_context *ctx,
 	uint8_t plane_id)
@@ -489,8 +464,7 @@ static void hubp_pg_control(
 
 static void power_on_plane(
 	struct dc_context *ctx,
-	uint8_t plane_id,
-	uint8_t inst)
+	int plane_id)
 {
 	uint32_t inst_offset = 0;
 
@@ -500,6 +474,8 @@ static void power_on_plane(
 	hubp_pg_control(ctx, plane_id, true);
 	HWSEQ_REG_SET(DC_IP_REQUEST_CNTL,
 			IP_REQUEST_EN, 0);
+	dm_logger_write(ctx->logger, LOG_DC,
+			"Un-gated front end for pipe %d\n", plane_id);
 }
 
 /* fully check bios enabledisplaypowergating table. dal only need dce init
@@ -596,13 +572,13 @@ static void init_hw(struct core_dc *dc)
 				dc->res_pool->mpcc[i];
 		struct mpcc_cfg mpcc_cfg;
 
-		lock_otg_master_update(dc->ctx, tg->inst);
+		tg->funcs->lock(tg);
 		mpcc_cfg.opp_id = 0xf;
 		mpcc_cfg.top_dpp_id = 0xf;
 		mpcc_cfg.bot_mpcc_id = 0xf;
 		mpcc_cfg.top_of_tree = true;
 		mpcc->funcs->set(mpcc, &mpcc_cfg);
-		unlock_otg_master(dc->ctx, tg->inst);
+		tg->funcs->unlock(tg);
 
 		tg->funcs->disable_vga(tg);
 		/* Blank controller using driver code instead of
@@ -822,56 +798,85 @@ static void reset_back_end_for_pipe(
 		return;
 
 	pipe_ctx->stream = NULL;
+	dm_logger_write(dc->ctx->logger, LOG_DC,
+					"Reset back end for pipe %d, tg:%d\n",
+					pipe_ctx->pipe_idx, pipe_ctx->tg->inst);
 }
 
-static void reset_front_end_for_pipe(
+static void reset_front_end(
 		struct core_dc *dc,
-		struct pipe_ctx *pipe_ctx,
-		struct validate_context *context)
+		int fe_idx)
 {
 	struct mpcc_cfg mpcc_cfg;
+	struct mem_input *mi = dc->res_pool->mis[fe_idx];
+	struct transform *xfm = dc->res_pool->transforms[fe_idx];
+	struct mpcc *mpcc = dc->res_pool->mpcc[fe_idx];
+	struct timing_generator *tg = dc->res_pool->timing_generators[mpcc->opp_id];
 
-	if (!pipe_ctx->surface)
+	/*Already reset*/
+	if (mpcc->opp_id == 0xf)
 		return;
 
-	pipe_ctx->mi->funcs->dcc_control(pipe_ctx->mi, false, false);
-
-	lock_otg_master_update(dc->ctx, pipe_ctx->tg->inst);
+	mi->funcs->dcc_control(mi, false, false);
+	tg->funcs->lock(tg);
 
 	mpcc_cfg.opp_id = 0xf;
 	mpcc_cfg.top_dpp_id = 0xf;
 	mpcc_cfg.bot_mpcc_id = 0xf;
-	mpcc_cfg.top_of_tree = !pipe_ctx->top_pipe;
-	pipe_ctx->mpcc->funcs->set(pipe_ctx->mpcc, &mpcc_cfg);
-
-	pipe_ctx->top_pipe = NULL;
-	pipe_ctx->bottom_pipe = NULL;
+	mpcc_cfg.top_of_tree = tg->inst == mpcc->inst;
+	mpcc->funcs->set(mpcc, &mpcc_cfg);
 
-	unlock_master_tg_and_wait(dc->ctx, pipe_ctx->tg->inst);
+	unlock_master_tg_and_wait(dc->ctx, tg->inst);
+	mpcc->funcs->wait_for_idle(mpcc);
+	mi->funcs->set_blank(mi, true);
+	wait_no_outstanding_request(dc->ctx, fe_idx);
+	disable_clocks(dc->ctx, fe_idx);
 
-	pipe_ctx->mi->funcs->set_blank(pipe_ctx->mi, true);
+	xfm->funcs->transform_reset(xfm);
 
-	wait_no_outstanding_request(dc->ctx, pipe_ctx->pipe_idx);
+	dm_logger_write(dc->ctx->logger, LOG_DC,
+					"Reset front end %d\n",
+					fe_idx);
+}
 
-	disable_clocks(dc->ctx, pipe_ctx->pipe_idx);
+static void dcn10_power_down_fe(struct core_dc *dc, int fe_idx)
+{
+	struct dc_context *ctx = dc->ctx;
+	uint32_t inst_offset = 0;
 
-	pipe_ctx->xfm->funcs->transform_reset(pipe_ctx->xfm);
+	reset_front_end(dc, fe_idx);
 
+	HWSEQ_REG_SET(DC_IP_REQUEST_CNTL,
+			IP_REQUEST_EN, 1);
+	dpp_pg_control(ctx, fe_idx, false);
+	hubp_pg_control(ctx, fe_idx, false);
+	HWSEQ_REG_SET(DC_IP_REQUEST_CNTL,
+			IP_REQUEST_EN, 0);
 	dm_logger_write(dc->ctx->logger, LOG_DC,
-					"Reset front end for pipe %d\n",
-					pipe_ctx->pipe_idx);
-
-	pipe_ctx->surface = NULL;
+			"Power gated front end %d\n", fe_idx);
 }
 
-static void reset_hw_ctx(struct core_dc *dc,
-		struct validate_context *context,
-		void (*reset)(struct core_dc *dc,
-				struct pipe_ctx *pipe_ctx,
-				struct validate_context *context))
+static void reset_hw_ctx_wrap(
+		struct core_dc *dc,
+		struct validate_context *context)
 {
 	int i;
 
+	/* Reset Front End*/
+	for (i = dc->res_pool->pipe_count - 1; i >= 0 ; i--) {
+		struct pipe_ctx *pipe_ctx_old =
+			&dc->current_context->res_ctx.pipe_ctx[i];
+		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
+
+		/*if (!pipe_ctx_old->stream)
+			continue;*/
+
+		if (!pipe_ctx->stream || !pipe_ctx->surface)
+			dcn10_power_down_fe(dc, i);
+		else if (pipe_need_reprogram(pipe_ctx_old, pipe_ctx))
+			reset_front_end(dc, i);
+	}
+	/* Reset Back End*/
 	for (i = dc->res_pool->pipe_count - 1; i >= 0 ; i--) {
 		struct pipe_ctx *pipe_ctx_old =
 			&dc->current_context->res_ctx.pipe_ctx[i];
@@ -882,23 +887,13 @@ static void reset_hw_ctx(struct core_dc *dc,
 
 		if (!pipe_ctx->stream ||
 				pipe_need_reprogram(pipe_ctx_old, pipe_ctx))
-			reset(dc, pipe_ctx_old, dc->current_context);
+			reset_back_end_for_pipe(dc, pipe_ctx_old, dc->current_context);
 	}
 }
 
-static void reset_hw_ctx_wrap(
-		struct core_dc *dc,
-		struct validate_context *context)
-{
-	/* Reset Front End*/
-	reset_hw_ctx(dc, context, reset_front_end_for_pipe);
-	/* Reset Back End*/
-	reset_hw_ctx(dc, context, reset_back_end_for_pipe);
-}
-
 
-static bool patch_address_for_sbs_tb_stereo(struct pipe_ctx *pipe_ctx,
-											PHYSICAL_ADDRESS_LOC *addr)
+static bool patch_address_for_sbs_tb_stereo(
+		struct pipe_ctx *pipe_ctx, PHYSICAL_ADDRESS_LOC *addr)
 {
 	struct core_surface *surface = pipe_ctx->surface;
 	bool sec_split = pipe_ctx->top_pipe &&
@@ -936,8 +931,7 @@ static void update_plane_addr(const struct core_dc *dc, struct pipe_ctx *pipe_ct
 }
 
 static bool dcn10_set_input_transfer_func(
-	struct pipe_ctx *pipe_ctx,
-	const struct core_surface *surface)
+	struct pipe_ctx *pipe_ctx, const struct core_surface *surface)
 {
 	struct input_pixel_processor *ipp = pipe_ctx->ipp;
 	const struct core_transfer_func *tf = NULL;
@@ -950,8 +944,8 @@ static bool dcn10_set_input_transfer_func(
 		tf = DC_TRANSFER_FUNC_TO_CORE(surface->public.in_transfer_func);
 
 	if (surface->public.gamma_correction && dce_use_lut(surface))
-	    ipp->funcs->ipp_program_input_lut(ipp,
-			    surface->public.gamma_correction);
+		ipp->funcs->ipp_program_input_lut(ipp,
+				surface->public.gamma_correction);
 
 	if (tf == NULL)
 		ipp->funcs->ipp_set_degamma(ipp, IPP_DEGAMMA_MODE_BYPASS);
@@ -1333,8 +1327,6 @@ static void dcn10_pipe_control_lock(
 	struct pipe_ctx *pipe,
 	bool lock)
 {
-	struct dce_hwseq *hws = hws = dc->hwseq;
-
 	/* use TG master update lock to lock everything on the TG
 	 * therefore only top pipe need to lock
 	 */
@@ -1342,9 +1334,9 @@ static void dcn10_pipe_control_lock(
 		return;
 
 	if (lock)
-		dcn10_lock(pipe->tg);
+		pipe->tg->funcs->lock(pipe->tg);
 	else
-		dcn10_unlock(pipe->tg);
+		pipe->tg->funcs->unlock(pipe->tg);
 }
 
 static bool wait_for_reset_trigger_to_occur(
@@ -1543,7 +1535,7 @@ static void dcn10_power_on_fe(
 	struct dc_surface *dc_surface = &pipe_ctx->surface->public;
 
 	power_on_plane(dc->ctx,
-		pipe_ctx->pipe_idx, pipe_ctx->tg->inst);
+		pipe_ctx->pipe_idx);
 
 	/* enable DCFCLK current DCHUB */
 	enable_dcfclk(dc->ctx,
@@ -1694,7 +1686,6 @@ static void update_dchubp_dpp(
 	struct default_adjustment ocsc = {0};
 	struct tg_color black_color = {0};
 	struct mpcc_cfg mpcc_cfg;
-	struct pipe_ctx *top_pipe;
 	bool per_pixel_alpha = surface->public.per_pixel_alpha && pipe_ctx->bottom_pipe;
 
 	/* TODO: proper fix once fpga works */
@@ -1734,14 +1725,13 @@ static void update_dchubp_dpp(
 			IPP_OUTPUT_FORMAT_12_BIT_FIX);
 
 	pipe_ctx->scl_data.lb_params.alpha_en = per_pixel_alpha;
-	for (top_pipe = pipe_ctx; top_pipe != NULL; top_pipe = top_pipe->top_pipe)
-		mpcc_cfg.opp_id = top_pipe->opp->inst;
 	mpcc_cfg.top_dpp_id = pipe_ctx->pipe_idx;
 	if (pipe_ctx->bottom_pipe)
 		mpcc_cfg.bot_mpcc_id = pipe_ctx->bottom_pipe->mpcc->inst;
 	else
 		mpcc_cfg.bot_mpcc_id = 0xf;
-	mpcc_cfg.top_of_tree = !pipe_ctx->top_pipe;
+	mpcc_cfg.opp_id = pipe_ctx->tg->inst;
+	mpcc_cfg.top_of_tree = pipe_ctx->pipe_idx == pipe_ctx->tg->inst;
 	mpcc_cfg.per_pixel_alpha = per_pixel_alpha;
 	/* DCN1.0 has output CM before MPC which seems to screw with
 	 * pre-multiplied alpha.
@@ -1749,8 +1739,6 @@ static void update_dchubp_dpp(
 	mpcc_cfg.pre_multiplied_alpha = is_rgb_cspace(
 			pipe_ctx->stream->public.output_color_space)
 					&& per_pixel_alpha;
-	if (!dc->current_context->res_ctx.pipe_ctx[pipe_ctx->pipe_idx].surface)
-		pipe_ctx->mpcc->funcs->wait_for_idle(pipe_ctx->mpcc);
 	pipe_ctx->mpcc->funcs->set(pipe_ctx->mpcc, &mpcc_cfg);
 
 	color_space_to_black_color(
@@ -1796,7 +1784,7 @@ static void program_all_pipe_in_tree(
 		/* watermark is for all pipes */
 		pipe_ctx->mi->funcs->program_watermarks(
 				pipe_ctx->mi, &context->bw.dcn.watermarks, ref_clk_mhz);
-		lock_otg_master_update(dc->ctx, pipe_ctx->tg->inst);
+		pipe_ctx->tg->funcs->lock(pipe_ctx->tg);
 
 		pipe_ctx->tg->dlg_otg_param.vready_offset = pipe_ctx->pipe_dlg_param.vready_offset;
 		pipe_ctx->tg->dlg_otg_param.vstartup_start = pipe_ctx->pipe_dlg_param.vstartup_start;
@@ -1852,13 +1840,41 @@ static void dcn10_apply_ctx_for_surface(
 {
 	int i;
 
+	/* reset unused mpcc */
+	/*for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
+		struct pipe_ctx *old_pipe_ctx =
+				&dc->current_context->res_ctx.pipe_ctx[i];
+
+		if ((!pipe_ctx->surface && old_pipe_ctx->surface)
+				|| (!pipe_ctx->stream && old_pipe_ctx->stream)) {
+			struct mpcc_cfg mpcc_cfg;
+
+			mpcc_cfg.opp_id = 0xf;
+			mpcc_cfg.top_dpp_id = 0xf;
+			mpcc_cfg.bot_mpcc_id = 0xf;
+			mpcc_cfg.top_of_tree = !old_pipe_ctx->top_pipe;
+			old_pipe_ctx->mpcc->funcs->set(old_pipe_ctx->mpcc, &mpcc_cfg);
+
+			old_pipe_ctx->top_pipe = NULL;
+			old_pipe_ctx->bottom_pipe = NULL;
+			old_pipe_ctx->surface = NULL;
+
+			dm_logger_write(dc->ctx->logger, LOG_DC,
+					"Reset mpcc for pipe %d\n",
+					old_pipe_ctx->pipe_idx);
+		}
+	}*/
+
+	if (!surface)
+		return;
+
 	for (i = 0; i < dc->res_pool->pipe_count; i++) {
 		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
 
-		if (!pipe_ctx->surface || pipe_ctx->surface != surface)
+		if (pipe_ctx->surface != surface)
 			continue;
 
-
 		/* looking for top pipe to program */
 		if (!pipe_ctx->top_pipe)
 			program_all_pipe_in_tree(dc, pipe_ctx, context);
@@ -1910,27 +1926,6 @@ static void dcn10_apply_ctx_for_surface(
 			context->bw.dcn.watermarks.d.cstate_pstate.pstate_change_ns,
 			context->bw.dcn.watermarks.d.pte_meta_urgent_ns
 			);
-
-	for (i = 0; i < dc->res_pool->pipe_count; i++) {
-		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
-
-		if (!pipe_ctx->surface || pipe_ctx->top_pipe)
-			continue;
-
-		/* unlock master update lock */
-		unlock_otg_master(dc->ctx, pipe_ctx->tg->inst);
-	}
-
-	/* reset unused pipe */
-	for (i = 0; i < dc->res_pool->pipe_count; i++) {
-		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
-		struct pipe_ctx *old_pipe_ctx =
-				&dc->current_context->res_ctx.pipe_ctx[i];
-
-		if ((!pipe_ctx->surface && old_pipe_ctx->surface)
-				|| (!pipe_ctx->stream && old_pipe_ctx->stream))
-			reset_front_end_for_pipe(dc, old_pipe_ctx, dc->current_context);
-	}
 }
 
 static void dcn10_set_bandwidth(
@@ -1992,23 +1987,6 @@ static void dcn10_set_bandwidth(
 	dcn10_pplib_apply_display_requirements(dc, context);
 }
 
-static void dcn10_power_down_fe(struct core_dc *dc, struct pipe_ctx *pipe)
-{
-	struct dc_context *ctx = dc->ctx;
-	uint32_t inst_offset = 0;
-
-	HWSEQ_REG_SET(DC_IP_REQUEST_CNTL,
-			IP_REQUEST_EN, 1);
-	dpp_pg_control(ctx, pipe->pipe_idx, false);
-	hubp_pg_control(ctx, pipe->pipe_idx, false);
-	HWSEQ_REG_SET(DC_IP_REQUEST_CNTL,
-			IP_REQUEST_EN, 0);
-
-	if (pipe->xfm)
-		pipe->xfm->funcs->transform_reset(pipe->xfm);
-	memset(&pipe->scl_data, 0, sizeof(pipe->scl_data));
-}
-
 static void set_drr(struct pipe_ctx **pipe_ctx,
 		int num_pipes, int vmin, int vmax)
 {
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_mpc.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_mpc.c
index b1c590d4f454..7af04bce3599 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_mpc.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_mpc.c
@@ -65,17 +65,17 @@ void dcn10_mpcc_set_bg_color(
 
 static void set_output_mux(struct dcn10_mpcc *mpcc10, int opp_id, int mpcc_id)
 {
-	ASSERT(mpcc10->opp_id == 0xf || opp_id == mpcc10->opp_id);
-	mpcc10->opp_id = opp_id;
+	ASSERT(mpcc10->base.opp_id == 0xf || opp_id == mpcc10->base.opp_id);
+	mpcc10->base.opp_id = opp_id;
 	REG_UPDATE(OPP_PIPE_CONTROL[opp_id], OPP_PIPE_CLOCK_EN, 1);
 	REG_SET(MUX[opp_id], 0, MPC_OUT_MUX, mpcc_id);
 }
 
 static void reset_output_mux(struct dcn10_mpcc *mpcc10)
 {
-	REG_SET(MUX[mpcc10->opp_id], 0, MPC_OUT_MUX, 0xf);
-	REG_UPDATE(OPP_PIPE_CONTROL[mpcc10->opp_id], OPP_PIPE_CLOCK_EN, 0);
-	mpcc10->opp_id = 0xf;
+	REG_SET(MUX[mpcc10->base.opp_id], 0, MPC_OUT_MUX, 0xf);
+	REG_UPDATE(OPP_PIPE_CONTROL[mpcc10->base.opp_id], OPP_PIPE_CLOCK_EN, 0);
+	mpcc10->base.opp_id = 0xf;
 }
 
 static void dcn10_mpcc_set(struct mpcc *mpcc, struct mpcc_cfg *cfg)
@@ -104,16 +104,17 @@ static void dcn10_mpcc_set(struct mpcc *mpcc, struct mpcc_cfg *cfg)
 	if (cfg->top_of_tree) {
 		if (cfg->opp_id != 0xf)
 			set_output_mux(mpcc10, cfg->opp_id, mpcc->inst);
-		else
+		else if (mpcc->opp_id != 0xf)
 			reset_output_mux(mpcc10);
 	}
+	mpcc10->base.opp_id = cfg->opp_id;
 }
 
 static void dcn10_mpcc_wait_idle(struct mpcc *mpcc)
 {
 	struct dcn10_mpcc *mpcc10 = TO_DCN10_MPCC(mpcc);
 
-	REG_WAIT(MPCC_STATUS, MPCC_IDLE, 1, 1000, 1000);
+	REG_WAIT(MPCC_STATUS, MPCC_BUSY, 0, 1000, 1000);
 }
 
 
@@ -139,5 +140,5 @@ void dcn10_mpcc_construct(struct dcn10_mpcc *mpcc10,
 	mpcc10->mpcc_shift = mpcc_shift;
 	mpcc10->mpcc_mask = mpcc_mask;
 
-	mpcc10->opp_id = inst;
+	mpcc10->base.opp_id = inst;
 }
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_mpc.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_mpc.h
index 0f9f1b97d238..fff267455a3d 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_mpc.h
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_mpc.h
@@ -68,6 +68,7 @@ struct dcn_mpcc_registers {
 	SF(MPCC0_MPCC_CONTROL, MPCC_ALPHA_MULTIPLIED_MODE, mask_sh),\
 	SF(MPCC0_MPCC_CONTROL, MPCC_BLND_ACTIVE_OVERLAP_ONLY, mask_sh),\
 	SF(MPCC0_MPCC_STATUS, MPCC_IDLE, mask_sh),\
+	SF(MPCC0_MPCC_STATUS, MPCC_BUSY, mask_sh),\
 	SF(MPCC0_MPCC_OPP_ID, MPCC_OPP_ID, mask_sh),\
 	SF(MPCC0_MPCC_BG_G_Y, MPCC_BG_G_Y, mask_sh),\
 	SF(MPCC0_MPCC_BG_R_CR, MPCC_BG_R_CR, mask_sh),\
@@ -83,6 +84,7 @@ struct dcn_mpcc_registers {
 	type MPCC_ALPHA_MULTIPLIED_MODE;\
 	type MPCC_BLND_ACTIVE_OVERLAP_ONLY;\
 	type MPCC_IDLE;\
+	type MPCC_BUSY;\
 	type MPCC_OPP_ID;\
 	type MPCC_BG_G_Y;\
 	type MPCC_BG_R_CR;\
@@ -103,8 +105,6 @@ struct dcn10_mpcc {
 	const struct dcn_mpcc_registers *mpcc_regs;
 	const struct dcn_mpcc_shift *mpcc_shift;
 	const struct dcn_mpcc_mask *mpcc_mask;
-
-	int opp_id;
 };
 
 void dcn10_mpcc_construct(struct dcn10_mpcc *mpcc10,
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_timing_generator.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_timing_generator.c
index 58fb29ff5677..802ace2e5212 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_timing_generator.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_timing_generator.c
@@ -575,6 +575,8 @@ void dcn10_lock(struct timing_generator *tg)
 {
 	struct dcn10_timing_generator *tgn10 = DCN10TG_FROM_TG(tg);
 
+	REG_SET(OTG_GLOBAL_CONTROL0, 0,
+			OTG_MASTER_UPDATE_LOCK_SEL, tg->inst);
 	REG_SET(OTG_MASTER_UPDATE_LOCK, 0,
 			OTG_MASTER_UPDATE_LOCK, 1);
 }
@@ -587,9 +589,9 @@ void dcn10_unlock(struct timing_generator *tg)
 			OTG_MASTER_UPDATE_LOCK, 0);
 
 	/* why are we waiting here? */
-	REG_WAIT(OTG_DOUBLE_BUFFER_CONTROL,
+	/*REG_WAIT(OTG_DOUBLE_BUFFER_CONTROL,
 			OTG_UPDATE_PENDING, 0,
-			20000, 200000);
+			20000, 200000);*/
 }
 
 static void dcn10_get_position(struct timing_generator *tg,
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_timing_generator.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_timing_generator.h
index c880fa587790..3b2a20a4b957 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_timing_generator.h
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_timing_generator.h
@@ -37,6 +37,7 @@
 	SRI(OTG_VREADY_PARAM, OTG, inst),\
 	SRI(OTG_BLANK_CONTROL, OTG, inst),\
 	SRI(OTG_MASTER_UPDATE_LOCK, OTG, inst),\
+	SRI(OTG_GLOBAL_CONTROL0, OTG, inst),\
 	SRI(OTG_DOUBLE_BUFFER_CONTROL, OTG, inst),\
 	SRI(OTG_H_TOTAL, OTG, inst),\
 	SRI(OTG_H_BLANK_START_END, OTG, inst),\
@@ -83,6 +84,7 @@ struct dcn_tg_registers {
 	uint32_t OTG_VREADY_PARAM;
 	uint32_t OTG_BLANK_CONTROL;
 	uint32_t OTG_MASTER_UPDATE_LOCK;
+	uint32_t OTG_GLOBAL_CONTROL0;
 	uint32_t OTG_DOUBLE_BUFFER_CONTROL;
 	uint32_t OTG_H_TOTAL;
 	uint32_t OTG_H_BLANK_START_END;
@@ -134,6 +136,7 @@ struct dcn_tg_registers {
 	SF(OTG0_OTG_BLANK_CONTROL, OTG_BLANK_DE_MODE, mask_sh),\
 	SF(OTG0_OTG_BLANK_CONTROL, OTG_CURRENT_BLANK_STATE, mask_sh),\
 	SF(OTG0_OTG_MASTER_UPDATE_LOCK, OTG_MASTER_UPDATE_LOCK, mask_sh),\
+	SF(OTG0_OTG_GLOBAL_CONTROL0, OTG_MASTER_UPDATE_LOCK_SEL, mask_sh),\
 	SF(OTG0_OTG_DOUBLE_BUFFER_CONTROL, OTG_UPDATE_PENDING, mask_sh),\
 	SF(OTG0_OTG_DOUBLE_BUFFER_CONTROL, OTG_BLANK_DATA_DOUBLE_BUFFER_EN, mask_sh),\
 	SF(OTG0_OTG_H_TOTAL, OTG_H_TOTAL, mask_sh),\
@@ -224,6 +227,7 @@ struct dcn_tg_registers {
 	type OTG_CURRENT_BLANK_STATE;\
 	type OTG_MASTER_UPDATE_LOCK;\
 	type OTG_UPDATE_PENDING;\
+	type OTG_MASTER_UPDATE_LOCK_SEL;\
 	type OTG_BLANK_DATA_DOUBLE_BUFFER_EN;\
 	type OTG_H_TOTAL;\
 	type OTG_H_BLANK_START;\
diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h b/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h
index 51bc8ef5ef79..55c9c305a259 100644
--- a/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h
+++ b/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h
@@ -40,6 +40,7 @@ struct mpcc {
 	const struct mpcc_funcs *funcs;
 	struct dc_context *ctx;
 	int inst;
+	int opp_id;
 };
 
 struct mpcc_funcs {
diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h
index 52884815c67b..642ae5e4ee1c 100644
--- a/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h
+++ b/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h
@@ -105,7 +105,7 @@ struct hw_sequencer_funcs {
 					struct dc_bios *dcb,
 					enum pipe_gating_control power_gating);
 
-	void (*power_down_front_end)(struct core_dc *dc, struct pipe_ctx *pipe);
+	void (*power_down_front_end)(struct core_dc *dc, int fe_idx);
 
 	void (*power_on_front_end)(struct core_dc *dc,
 			struct pipe_ctx *pipe,