drm/amd/display: Add basic infrastructure for enabling FAMS

We want to enable Firmware Assisted Memory (FAMS) Switching, but first,
we need to add the required code infrastructure in DC before allowing it
in amdgpu_dm.
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@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/core/dc.c

@@ -3842,10 +3842,18 @@ bool dc_is_plane_eligible_for_idle_optimizations(struct dc *dc, struct dc_plane_
 /* cleanup on driver unload */
 void dc_hardware_release(struct dc *dc)
 {
+	dc_mclk_switch_using_fw_based_vblank_stretch_shut_down(dc);
+
 	if (dc->hwss.hardware_release)
 		dc->hwss.hardware_release(dc);
 }
 
+void dc_mclk_switch_using_fw_based_vblank_stretch_shut_down(struct dc *dc)
+{
+	if (dc->current_state)
+		dc->current_state->bw_ctx.bw.dcn.clk.fw_based_mclk_switching_shut_down = true;
+}
+
 /*
  *****************************************************************************
  * Function: dc_is_dmub_outbox_supported -

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

@@ -164,6 +164,7 @@ struct dc_color_caps {
 
 struct dc_dmub_caps {
 	bool psr;
+	bool mclk_sw;
 };
 
 struct dc_caps {
@@ -359,6 +360,8 @@ enum visual_confirm {
 	VISUAL_CONFIRM_HDR = 2,
 	VISUAL_CONFIRM_MPCTREE = 4,
 	VISUAL_CONFIRM_PSR = 5,
+	VISUAL_CONFIRM_SWAPCHAIN = 6,
+	VISUAL_CONFIRM_FAMS = 7,
 	VISUAL_CONFIRM_SWIZZLE = 9,
 };
 
@@ -441,6 +444,7 @@ struct dc_clocks {
 	bool fclk_prev_p_state_change_support;
 	int num_ways;
 	bool fw_based_mclk_switching;
+	bool fw_based_mclk_switching_shut_down;
 	int prev_num_ways;
 	enum dtm_pstate dtm_level;
 	int max_supported_dppclk_khz;
@@ -726,6 +730,7 @@ struct dc_debug_options {
 
 	/* Enable dmub aux for legacy ddc */
 	bool enable_dmub_aux_for_legacy_ddc;
+	bool disable_fams;
 	bool optimize_edp_link_rate; /* eDP ILR */
 	/* FEC/PSR1 sequence enable delay in 100us */
 	uint8_t fec_enable_delay_in100us;
@@ -1452,6 +1457,9 @@ void dc_enable_dcmode_clk_limit(struct dc *dc, bool enable);
 /* cleanup on driver unload */
 void dc_hardware_release(struct dc *dc);
 
+/* disables fw based mclk switch */
+void dc_mclk_switch_using_fw_based_vblank_stretch_shut_down(struct dc *dc);
+
 bool dc_set_psr_allow_active(struct dc *dc, bool enable);
 void dc_z10_restore(const struct dc *dc);
 void dc_z10_save_init(struct dc *dc);

drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c

@@ -252,6 +252,93 @@ void dc_dmub_trace_event_control(struct dc *dc, bool enable)
 	dm_helpers_dmub_outbox_interrupt_control(dc->ctx, enable);
 }
 
+void dc_dmub_srv_drr_update_cmd(struct dc *dc, uint32_t tg_inst, uint32_t vtotal_min, uint32_t vtotal_max)
+{
+	union dmub_rb_cmd cmd = { 0 };
+
+	cmd.drr_update.header.type = DMUB_CMD__FW_ASSISTED_MCLK_SWITCH;
+	cmd.drr_update.header.sub_type = DMUB_CMD__FAMS_DRR_UPDATE;
+	cmd.drr_update.dmub_optc_state_req.v_total_max = vtotal_max;
+	cmd.drr_update.dmub_optc_state_req.v_total_min = vtotal_min;
+	cmd.drr_update.dmub_optc_state_req.tg_inst = tg_inst;
+
+	cmd.drr_update.header.payload_bytes = sizeof(cmd.drr_update) - sizeof(cmd.drr_update.header);
+
+	// Send the command to the DMCUB.
+	dc_dmub_srv_cmd_queue(dc->ctx->dmub_srv, &cmd);
+	dc_dmub_srv_cmd_execute(dc->ctx->dmub_srv);
+	dc_dmub_srv_wait_idle(dc->ctx->dmub_srv);
+}
+
+uint8_t dc_dmub_srv_get_pipes_for_stream(struct dc *dc, struct dc_stream_state *stream)
+{
+	uint8_t pipes = 0;
+	int i = 0;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe->stream == stream && pipe->stream_res.tg)
+			pipes = i;
+	}
+	return pipes;
+}
+
+int dc_dmub_srv_get_timing_generator_offset(struct dc *dc, struct dc_stream_state *stream)
+{
+	int  tg_inst = 0;
+	int i = 0;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		struct pipe_ctx *pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+
+		if (pipe->stream == stream && pipe->stream_res.tg) {
+			tg_inst = pipe->stream_res.tg->inst;
+			break;
+		}
+	}
+	return tg_inst;
+}
+
+bool dc_dmub_srv_p_state_delegate(struct dc *dc, bool should_manage_pstate, struct dc_state *context)
+{
+	union dmub_rb_cmd cmd = { 0 };
+	struct dmub_cmd_fw_assisted_mclk_switch_config *config_data = &cmd.fw_assisted_mclk_switch.config_data;
+	int i = 0;
+	int ramp_up_num_steps = 1; // TODO: Ramp is currently disabled. Reenable it.
+	uint8_t visual_confirm_enabled = dc->debug.visual_confirm == VISUAL_CONFIRM_FAMS;
+
+	if (dc == NULL)
+		return false;
+
+	// Format command.
+	cmd.fw_assisted_mclk_switch.header.type = DMUB_CMD__FW_ASSISTED_MCLK_SWITCH;
+	cmd.fw_assisted_mclk_switch.header.sub_type = DMUB_CMD__FAMS_SETUP_FW_CTRL;
+	cmd.fw_assisted_mclk_switch.config_data.fams_enabled = should_manage_pstate;
+	cmd.fw_assisted_mclk_switch.config_data.visual_confirm_enabled = visual_confirm_enabled;
+
+	for (i = 0; context && i < context->stream_count; i++) {
+		struct dc_stream_state *stream = context->streams[i];
+		uint8_t min_refresh_in_hz = (stream->timing.min_refresh_in_uhz + 999999) / 1000000;
+		int  tg_inst = dc_dmub_srv_get_timing_generator_offset(dc, stream);
+
+		config_data->pipe_data[tg_inst].pix_clk_100hz = stream->timing.pix_clk_100hz;
+		config_data->pipe_data[tg_inst].min_refresh_in_hz = min_refresh_in_hz;
+		config_data->pipe_data[tg_inst].max_ramp_step = ramp_up_num_steps;
+		config_data->pipe_data[tg_inst].pipes = dc_dmub_srv_get_pipes_for_stream(dc, stream);
+	}
+
+	cmd.fw_assisted_mclk_switch.header.payload_bytes =
+		sizeof(cmd.fw_assisted_mclk_switch) - sizeof(cmd.fw_assisted_mclk_switch.header);
+
+	// Send the command to the DMCUB.
+	dc_dmub_srv_cmd_queue(dc->ctx->dmub_srv, &cmd);
+	dc_dmub_srv_cmd_execute(dc->ctx->dmub_srv);
+	dc_dmub_srv_wait_idle(dc->ctx->dmub_srv);
+
+	return true;
+}
+
 void dc_dmub_srv_query_caps_cmd(struct dmub_srv *dmub)
 {
 	union dmub_rb_cmd cmd = { 0 };

drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h

@@ -72,6 +72,10 @@ bool dc_dmub_srv_get_dmub_outbox0_msg(const struct dc *dc, struct dmcub_trace_bu
 
 void dc_dmub_trace_event_control(struct dc *dc, bool enable);
 
+void dc_dmub_srv_drr_update_cmd(struct dc *dc, uint32_t tg_inst, uint32_t vtotal_min, uint32_t vtotal_max);
+
+bool dc_dmub_srv_p_state_delegate(struct dc *dc, bool enable_pstate, struct dc_state *context);
+
 void dc_dmub_srv_query_caps_cmd(struct dmub_srv *dmub);
 void dc_dmub_srv_clear_inbox0_ack(struct dc_dmub_srv *dmub_srv);
 void dc_dmub_srv_wait_for_inbox0_ack(struct dc_dmub_srv *dmub_srv);

drivers/gpu/drm/amd/display/dc/dc_types.h

@@ -196,6 +196,8 @@ struct dc_panel_patch {
 	unsigned int disable_fec;
 	unsigned int extra_t3_ms;
 	unsigned int max_dsc_target_bpp_limit;
+	unsigned int embedded_tiled_slave;
+	unsigned int disable_fams;
 	unsigned int skip_avmute;
 };
 

drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hwseq.c

@@ -641,6 +641,7 @@ void dcn30_init_hw(struct dc *dc)
 	// Get DMCUB capabilities
 	dc_dmub_srv_query_caps_cmd(dc->ctx->dmub_srv->dmub);
 	dc->caps.dmub_caps.psr = dc->ctx->dmub_srv->dmub->feature_caps.psr;
+	dc->caps.dmub_caps.mclk_sw = dc->ctx->dmub_srv->dmub->feature_caps.fw_assisted_mclk_switch;
 }
 
 void dcn30_set_avmute(struct pipe_ctx *pipe_ctx, bool enable)
@@ -938,8 +939,14 @@ bool dcn30_does_plane_fit_in_mall(struct dc *dc, struct dc_plane_state *plane, s
 
 void dcn30_hardware_release(struct dc *dc)
 {
-	/* if pstate unsupported, force it supported */
-	if (!dc->clk_mgr->clks.p_state_change_support &&
+	dc_dmub_srv_p_state_delegate(dc, false, NULL);
+
+	/* If pstate unsupported, or still supported
+	 * by firmware, force it supported by dcn
+	 */
+	if (dc->current_state)
+		if ((!dc->clk_mgr->clks.p_state_change_support ||
+				dc->current_state->bw_ctx.bw.dcn.clk.fw_based_mclk_switching) &&
 				dc->res_pool->hubbub->funcs->force_pstate_change_control)
 			dc->res_pool->hubbub->funcs->force_pstate_change_control(
 					dc->res_pool->hubbub, true, true);
@@ -966,5 +973,8 @@ void dcn30_prepare_bandwidth(struct dc *dc,
 			dc->clk_mgr->funcs->set_max_memclk(dc->clk_mgr, dc->clk_mgr->bw_params->clk_table.entries[dc->clk_mgr->bw_params->clk_table.num_entries - 1].memclk_mhz);
 
 	dcn20_prepare_bandwidth(dc, context);
+
+	dc_dmub_srv_p_state_delegate(dc,
+		context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching, context);
 }
 

drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c

@@ -1899,6 +1899,176 @@ noinline bool dcn30_internal_validate_bw(
 	return out;
 }
 
+int get_refresh_rate(struct dc_state *context)
+{
+	int refresh_rate = 0;
+	int h_v_total = 0;
+	struct dc_crtc_timing *timing = NULL;
+
+	if (context == NULL || context->streams[0] == NULL)
+		return 0;
+
+	/* check if refresh rate at least 120hz */
+	timing = &context->streams[0]->timing;
+	if (timing == NULL)
+		return 0;
+
+	h_v_total = timing->h_total * timing->v_total;
+	if (h_v_total == 0)
+		return 0;
+
+	refresh_rate = ((timing->pix_clk_100hz * 100) / (h_v_total)) + 1;
+	return refresh_rate;
+}
+
+#define MAX_STRETCHED_V_BLANK 500 // in micro-seconds
+/*
+ * Scaling factor for v_blank stretch calculations considering timing in
+ * micro-seconds and pixel clock in 100hz.
+ * Note: the parenthesis are necessary to ensure the correct order of
+ * operation where V_SCALE is used.
+ */
+#define V_SCALE (10000 / MAX_STRETCHED_V_BLANK)
+
+int get_frame_rate_at_max_stretch_100hz(struct dc_state *context)
+{
+	struct dc_crtc_timing *timing = NULL;
+	uint32_t sec_per_100_lines;
+	uint32_t max_v_blank;
+	uint32_t curr_v_blank;
+	uint32_t v_stretch_max;
+	uint32_t stretched_frame_pix_cnt;
+	uint32_t scaled_stretched_frame_pix_cnt;
+	uint32_t scaled_refresh_rate;
+
+	if (context == NULL || context->streams[0] == NULL)
+		return 0;
+
+	/* check if refresh rate at least 120hz */
+	timing = &context->streams[0]->timing;
+	if (timing == NULL)
+		return 0;
+
+	sec_per_100_lines = timing->pix_clk_100hz / timing->h_total + 1;
+	max_v_blank = sec_per_100_lines / V_SCALE + 1;
+	curr_v_blank = timing->v_total - timing->v_addressable;
+	v_stretch_max = (max_v_blank > curr_v_blank) ? (max_v_blank - curr_v_blank) : (0);
+	stretched_frame_pix_cnt = (v_stretch_max + timing->v_total) * timing->h_total;
+	scaled_stretched_frame_pix_cnt = stretched_frame_pix_cnt / 10000;
+	scaled_refresh_rate = (timing->pix_clk_100hz) / scaled_stretched_frame_pix_cnt + 1;
+
+	return scaled_refresh_rate;
+}
+
+bool is_refresh_rate_support_mclk_switch_using_fw_based_vblank_stretch(struct dc_state *context)
+{
+	int refresh_rate_max_stretch_100hz;
+	int min_refresh_100hz;
+
+	if (context == NULL || context->streams[0] == NULL)
+		return false;
+
+	refresh_rate_max_stretch_100hz = get_frame_rate_at_max_stretch_100hz(context);
+	min_refresh_100hz = context->streams[0]->timing.min_refresh_in_uhz / 10000;
+
+	if (refresh_rate_max_stretch_100hz < min_refresh_100hz)
+		return false;
+
+	return true;
+}
+
+bool dcn30_can_support_mclk_switch_using_fw_based_vblank_stretch(struct dc *dc, struct dc_state *context)
+{
+	int refresh_rate = 0;
+	const int minimum_refreshrate_supported = 120;
+
+	if (context == NULL || context->streams[0] == NULL)
+		return false;
+
+	if (context->streams[0]->sink->edid_caps.panel_patch.disable_fams)
+		return false;
+
+	if (dc->debug.disable_fams)
+		return false;
+
+	if (!dc->caps.dmub_caps.mclk_sw)
+		return false;
+
+	if (context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching_shut_down)
+		return false;
+
+	/* more then 1 monitor connected */
+	if (context->stream_count != 1)
+		return false;
+
+	refresh_rate = get_refresh_rate(context);
+	if (refresh_rate < minimum_refreshrate_supported)
+		return false;
+
+	if (!is_refresh_rate_support_mclk_switch_using_fw_based_vblank_stretch(context))
+		return false;
+
+	// check if freesync enabled
+	if (!context->streams[0]->allow_freesync)
+		return false;
+
+	if (context->streams[0]->vrr_active_variable)
+		return false;
+
+	return true;
+}
+
+/*
+ * set up FPO watermarks, pstate, dram latency
+ */
+void dcn30_setup_mclk_switch_using_fw_based_vblank_stretch(struct dc *dc, struct dc_state *context)
+{
+	ASSERT(dc != NULL && context != NULL);
+	if (dc == NULL || context == NULL)
+		return;
+
+	/* Set wm_a.pstate so high natural MCLK switches are impossible: 4 seconds */
+	context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.pstate_change_ns = 4U * 1000U * 1000U * 1000U;
+}
+
+/*
+ * Finds dummy_latency_index when MCLK switching using firmware based
+ * vblank stretch is enabled. This function will iterate through the
+ * table of dummy pstate latencies until the lowest value that allows
+ * dm_allow_self_refresh_and_mclk_switch to happen is found
+ */
+int dcn30_find_dummy_latency_index_for_fw_based_mclk_switch(struct dc *dc, struct dc_state *context,
+		display_e2e_pipe_params_st *pipes, int pipe_cnt, int vlevel)
+{
+	const int max_latency_table_entries = 4;
+	int dummy_latency_index = 0;
+
+	while (dummy_latency_index < max_latency_table_entries) {
+		context->bw_ctx.dml.soc.dram_clock_change_latency_us =
+				dc->clk_mgr->bw_params->dummy_pstate_table[dummy_latency_index].dummy_pstate_latency_us;
+		dcn30_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, false);
+
+		if (context->bw_ctx.dml.soc.allow_dram_self_refresh_or_dram_clock_change_in_vblank ==
+			dm_allow_self_refresh_and_mclk_switch)
+			break;
+
+		dummy_latency_index++;
+	}
+
+	if (dummy_latency_index == max_latency_table_entries) {
+		ASSERT(dummy_latency_index != max_latency_table_entries);
+		/* If the execution gets here, it means dummy p_states are
+		 * not possible. This should never happen and would mean
+		 * something is severely wrong.
+		 * Here we reset dummy_latency_index to 3, because it is
+		 * better to have underflows than system crashes.
+		 */
+		dummy_latency_index = 3;
+	}
+
+	return dummy_latency_index;
+}
+
 void dcn30_update_soc_for_wm_a(struct dc *dc, struct dc_state *context)
 {
 	DC_FP_START();

drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.c

@@ -797,6 +797,12 @@ void dcn20_calculate_dlg_params(
 	context->bw_ctx.bw.dcn.clk.p_state_change_support =
 		context->bw_ctx.dml.vba.DRAMClockChangeSupport[vlevel][context->bw_ctx.dml.vba.maxMpcComb]
 							!= dm_dram_clock_change_unsupported;
+
+	/* Pstate change might not be supported by hardware, but it might be
+	 * possible with firmware driven vertical blank stretching.
+	 */
+	context->bw_ctx.bw.dcn.clk.p_state_change_support |= context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching;
+
 	context->bw_ctx.bw.dcn.clk.dppclk_khz = 0;
 
 	context->bw_ctx.bw.dcn.clk.dtbclk_en = is_dtbclk_required(dc, context);

drivers/gpu/drm/amd/display/dc/dml/dcn30/dcn30_fpu.c

@@ -529,6 +529,8 @@ dc_assert_fp_enabled();
 		context->bw_ctx.dml.soc.dram_clock_change_latency_us =
 				dc->clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.pstate_latency_us;
 
+	if (context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching)
+		dcn30_setup_mclk_switch_using_fw_based_vblank_stretch(dc, context);
 }
 
 void dcn30_fpu_update_dram_channel_width_bytes(struct dc *dc)

drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h

@@ -242,7 +242,8 @@ struct dmub_feature_caps {
 	 * Max PSR version supported by FW.
 	 */
 	uint8_t psr;
-	uint8_t reserved[7];
+	uint8_t fw_assisted_mclk_switch;
+	uint8_t reserved[6];
 };
 
 #if defined(__cplusplus)
@@ -2773,6 +2774,26 @@ struct dmub_rb_cmd_drr_update {
 		struct dmub_optc_state dmub_optc_state_req;
 };
 
+struct dmub_cmd_fw_assisted_mclk_switch_pipe_data {
+	uint32_t pix_clk_100hz;
+	uint8_t max_ramp_step;
+	uint8_t pipes;
+	uint8_t min_refresh_in_hz;
+	uint8_t padding[1];
+};
+
+struct dmub_cmd_fw_assisted_mclk_switch_config {
+	uint8_t fams_enabled;
+	uint8_t visual_confirm_enabled;
+	uint8_t padding[2];
+	struct dmub_cmd_fw_assisted_mclk_switch_pipe_data pipe_data[DMUB_MAX_STREAMS];
+};
+
+struct dmub_rb_cmd_fw_assisted_mclk_switch {
+	struct dmub_cmd_header header;
+	struct dmub_cmd_fw_assisted_mclk_switch_config config_data;
+};
+
 /**
  * enum dmub_cmd_panel_cntl_type - Panel control command.
  */
@@ -3111,6 +3132,8 @@ union dmub_rb_cmd {
 	 */
 	struct dmub_rb_cmd_query_feature_caps query_feature_caps;
 	struct dmub_rb_cmd_drr_update drr_update;
+	struct dmub_rb_cmd_fw_assisted_mclk_switch fw_assisted_mclk_switch;
+
 	/**
 	 * Definition of a DMUB_CMD__VBIOS_LVTMA_CONTROL command.
 	 */