diff options
Diffstat (limited to 'drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c')
| -rw-r--r-- | drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c | 1915 |
1 files changed, 196 insertions, 1719 deletions
diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c index 1ea6d258a20d..9a26d24b579f 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c @@ -88,6 +88,7 @@ #include "dml/dcn30/display_mode_vba_30.h" #include "vm_helper.h" #include "dcn20/dcn20_vmid.h" +#include "dml/dcn32/dcn32_fpu.h" #define DCN_BASE__INST0_SEG1 0x000000C0 #define DCN_BASE__INST0_SEG2 0x000034C0 @@ -114,137 +115,6 @@ static const struct IP_BASE DCN_BASE = { { { { 0x00000012, 0x000000C0, 0x000034C #define DC_LOGGER_INIT(logger) -#define DCN3_2_DEFAULT_DET_SIZE 256 -#define DCN3_2_MAX_DET_SIZE 1152 -#define DCN3_2_MIN_DET_SIZE 128 -#define DCN3_2_MIN_COMPBUF_SIZE_KB 128 - -struct _vcs_dpi_ip_params_st dcn3_2_ip = { - .gpuvm_enable = 1, - .gpuvm_max_page_table_levels = 1, - .hostvm_enable = 0, - .rob_buffer_size_kbytes = 128, - .det_buffer_size_kbytes = DCN3_2_DEFAULT_DET_SIZE, - .config_return_buffer_size_in_kbytes = 1280, - .compressed_buffer_segment_size_in_kbytes = 64, - .meta_fifo_size_in_kentries = 22, - .zero_size_buffer_entries = 512, - .compbuf_reserved_space_64b = 256, - .compbuf_reserved_space_zs = 64, - .dpp_output_buffer_pixels = 2560, - .opp_output_buffer_lines = 1, - .pixel_chunk_size_kbytes = 8, - .alpha_pixel_chunk_size_kbytes = 4, // not appearing in spreadsheet, match c code from hw team - .min_pixel_chunk_size_bytes = 1024, - .dcc_meta_buffer_size_bytes = 6272, - .meta_chunk_size_kbytes = 2, - .min_meta_chunk_size_bytes = 256, - .writeback_chunk_size_kbytes = 8, - .ptoi_supported = false, - .num_dsc = 4, - .maximum_dsc_bits_per_component = 12, - .maximum_pixels_per_line_per_dsc_unit = 6016, - .dsc422_native_support = true, - .is_line_buffer_bpp_fixed = true, - .line_buffer_fixed_bpp = 57, - .line_buffer_size_bits = 1171920, //DPP doc, DCN3_2_DisplayMode_73.xlsm still shows as 986880 bits with 48 bpp - .max_line_buffer_lines = 32, - .writeback_interface_buffer_size_kbytes = 90, - .max_num_dpp = 4, - .max_num_otg = 4, - .max_num_hdmi_frl_outputs = 1, - .max_num_wb = 1, - .max_dchub_pscl_bw_pix_per_clk = 4, - .max_pscl_lb_bw_pix_per_clk = 2, - .max_lb_vscl_bw_pix_per_clk = 4, - .max_vscl_hscl_bw_pix_per_clk = 4, - .max_hscl_ratio = 6, - .max_vscl_ratio = 6, - .max_hscl_taps = 8, - .max_vscl_taps = 8, - .dpte_buffer_size_in_pte_reqs_luma = 64, - .dpte_buffer_size_in_pte_reqs_chroma = 34, - .dispclk_ramp_margin_percent = 1, - .max_inter_dcn_tile_repeaters = 8, - .cursor_buffer_size = 16, - .cursor_chunk_size = 2, - .writeback_line_buffer_buffer_size = 0, - .writeback_min_hscl_ratio = 1, - .writeback_min_vscl_ratio = 1, - .writeback_max_hscl_ratio = 1, - .writeback_max_vscl_ratio = 1, - .writeback_max_hscl_taps = 1, - .writeback_max_vscl_taps = 1, - .dppclk_delay_subtotal = 47, - .dppclk_delay_scl = 50, - .dppclk_delay_scl_lb_only = 16, - .dppclk_delay_cnvc_formatter = 28, - .dppclk_delay_cnvc_cursor = 6, - .dispclk_delay_subtotal = 125, - .dynamic_metadata_vm_enabled = false, - .odm_combine_4to1_supported = false, - .dcc_supported = true, - .max_num_dp2p0_outputs = 2, - .max_num_dp2p0_streams = 4, -}; - -struct _vcs_dpi_soc_bounding_box_st dcn3_2_soc = { - .clock_limits = { - { - .state = 0, - .dcfclk_mhz = 1564.0, - .fabricclk_mhz = 400.0, - .dispclk_mhz = 2150.0, - .dppclk_mhz = 2150.0, - .phyclk_mhz = 810.0, - .phyclk_d18_mhz = 667.0, - .phyclk_d32_mhz = 625.0, - .socclk_mhz = 1200.0, - .dscclk_mhz = 716.667, - .dram_speed_mts = 1600.0, - .dtbclk_mhz = 1564.0, - }, - }, - .num_states = 1, - .sr_exit_time_us = 5.20, - .sr_enter_plus_exit_time_us = 9.60, - .sr_exit_z8_time_us = 285.0, - .sr_enter_plus_exit_z8_time_us = 320, - .writeback_latency_us = 12.0, - .round_trip_ping_latency_dcfclk_cycles = 263, - .urgent_latency_pixel_data_only_us = 4.0, - .urgent_latency_pixel_mixed_with_vm_data_us = 4.0, - .urgent_latency_vm_data_only_us = 4.0, - .fclk_change_latency_us = 20, - .usr_retraining_latency_us = 2, - .smn_latency_us = 2, - .mall_allocated_for_dcn_mbytes = 64, - .urgent_out_of_order_return_per_channel_pixel_only_bytes = 4096, - .urgent_out_of_order_return_per_channel_pixel_and_vm_bytes = 4096, - .urgent_out_of_order_return_per_channel_vm_only_bytes = 4096, - .pct_ideal_sdp_bw_after_urgent = 100.0, - .pct_ideal_fabric_bw_after_urgent = 67.0, - .pct_ideal_dram_sdp_bw_after_urgent_pixel_only = 20.0, - .pct_ideal_dram_sdp_bw_after_urgent_pixel_and_vm = 60.0, // N/A, for now keep as is until DML implemented - .pct_ideal_dram_sdp_bw_after_urgent_vm_only = 30.0, // N/A, for now keep as is until DML implemented - .pct_ideal_dram_bw_after_urgent_strobe = 67.0, - .max_avg_sdp_bw_use_normal_percent = 80.0, - .max_avg_fabric_bw_use_normal_percent = 60.0, - .max_avg_dram_bw_use_normal_strobe_percent = 50.0, - .max_avg_dram_bw_use_normal_percent = 15.0, - .num_chans = 8, - .dram_channel_width_bytes = 2, - .fabric_datapath_to_dcn_data_return_bytes = 64, - .return_bus_width_bytes = 64, - .downspread_percent = 0.38, - .dcn_downspread_percent = 0.5, - .dram_clock_change_latency_us = 400, - .dispclk_dppclk_vco_speed_mhz = 4300.0, - .do_urgent_latency_adjustment = true, - .urgent_latency_adjustment_fabric_clock_component_us = 1.0, - .urgent_latency_adjustment_fabric_clock_reference_mhz = 1000, -}; - enum dcn32_clk_src_array_id { DCN32_CLK_SRC_PLL0, DCN32_CLK_SRC_PLL1, @@ -312,6 +182,7 @@ enum dcn32_clk_src_array_id { .reg_name = NBIO_BASE(regBIF_BX0_ ## reg_name ## _BASE_IDX) + \ regBIF_BX0_ ## reg_name +#undef CTX #define CTX ctx #define REG(reg_name) \ (DCN_BASE.instance[0].segment[reg ## reg_name ## _BASE_IDX] + reg ## reg_name) @@ -695,18 +566,14 @@ static const struct dcn20_dsc_mask dsc_mask = { }; static const struct dcn30_mpc_registers mpc_regs = { - MPC_REG_LIST_DCN3_0(0), - MPC_REG_LIST_DCN3_0(1), - MPC_REG_LIST_DCN3_0(2), - MPC_REG_LIST_DCN3_0(3), + MPC_REG_LIST_DCN3_2(0), + MPC_REG_LIST_DCN3_2(1), + MPC_REG_LIST_DCN3_2(2), + MPC_REG_LIST_DCN3_2(3), MPC_OUT_MUX_REG_LIST_DCN3_0(0), MPC_OUT_MUX_REG_LIST_DCN3_0(1), MPC_OUT_MUX_REG_LIST_DCN3_0(2), MPC_OUT_MUX_REG_LIST_DCN3_0(3), - MPC_MCM_REG_LIST_DCN32(0), - MPC_MCM_REG_LIST_DCN32(1), - MPC_MCM_REG_LIST_DCN32(2), - MPC_MCM_REG_LIST_DCN32(3), MPC_DWB_MUX_REG_LIST_DCN3_0(0), }; @@ -976,8 +843,7 @@ static const struct dc_debug_options debug_defaults_drv = { .timing_trace = false, .clock_trace = true, .disable_pplib_clock_request = false, - .disable_idle_power_optimizations = true, - .pipe_split_policy = MPC_SPLIT_DYNAMIC, + .pipe_split_policy = MPC_SPLIT_AVOID, // Due to CRB, no need to MPC split anymore .force_single_disp_pipe_split = false, .disable_dcc = DCC_ENABLE, .vsr_support = true, @@ -1001,7 +867,10 @@ static const struct dc_debug_options debug_defaults_drv = { } }, .use_max_lb = true, - .force_disable_subvp = true + .force_disable_subvp = true, + .exit_idle_opt_for_cursor_updates = true, + .enable_single_display_2to1_odm_policy = true, + .enable_dp_dig_pixel_rate_div_policy = 1, }; static const struct dc_debug_options debug_defaults_diags = { @@ -1249,6 +1118,7 @@ static const struct encoder_feature_support link_enc_feature = { }; static struct link_encoder *dcn32_link_encoder_create( + struct dc_context *ctx, const struct encoder_init_data *enc_init_data) { struct dcn20_link_encoder *enc20 = @@ -1756,169 +1626,6 @@ bool dcn32_release_post_bldn_3dlut( return ret; } -/** - ******************************************************************************************** - * dcn32_get_num_free_pipes: Calculate number of free pipes - * - * This function assumes that a "used" pipe is a pipe that has - * both a stream and a plane assigned to it. - * - * @param [in] dc: current dc state - * @param [in] context: new dc state - * - * @return: Number of free pipes available in the context - * - ******************************************************************************************** - */ -static unsigned int dcn32_get_num_free_pipes(struct dc *dc, struct dc_state *context) -{ - unsigned int i; - unsigned int free_pipes = 0; - unsigned int num_pipes = 0; - - for (i = 0; i < dc->res_pool->pipe_count; i++) { - struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - - if (pipe->stream && pipe->plane_state && !pipe->top_pipe) { - while (pipe) { - num_pipes++; - pipe = pipe->bottom_pipe; - } - } - } - - free_pipes = dc->res_pool->pipe_count - num_pipes; - return free_pipes; -} - -/** - ******************************************************************************************** - * dcn32_assign_subvp_pipe: Function to decide which pipe will use Sub-VP. - * - * We enter this function if we are Sub-VP capable (i.e. enough pipes available) - * and regular P-State switching (i.e. VACTIVE/VBLANK) is not supported, or if - * we are forcing SubVP P-State switching on the current config. - * - * The number of pipes used for the chosen surface must be less than or equal to the - * number of free pipes available. - * - * In general we choose surfaces that have ActiveDRAMClockChangeLatencyMargin <= 0 first, - * then among those surfaces we choose the one with the smallest VBLANK time. We only consider - * surfaces with ActiveDRAMClockChangeLatencyMargin > 0 if we are forcing a Sub-VP config. - * - * @param [in] dc: current dc state - * @param [in] context: new dc state - * @param [out] index: dc pipe index for the pipe chosen to have phantom pipes assigned - * - * @return: True if a valid pipe assignment was found for Sub-VP. Otherwise false. - * - ******************************************************************************************** - */ - -static bool dcn32_assign_subvp_pipe(struct dc *dc, - struct dc_state *context, - unsigned int *index) -{ - unsigned int i, pipe_idx; - unsigned int min_vblank_us = INT_MAX; - struct vba_vars_st *vba = &context->bw_ctx.dml.vba; - bool valid_assignment_found = false; - unsigned int free_pipes = dcn32_get_num_free_pipes(dc, context); - - for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) { - struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - unsigned int num_pipes = 0; - - if (!pipe->stream) - continue; - - if (pipe->plane_state && !pipe->top_pipe && - pipe->stream->mall_stream_config.type == SUBVP_NONE) { - while (pipe) { - num_pipes++; - pipe = pipe->bottom_pipe; - } - - pipe = &context->res_ctx.pipe_ctx[i]; - if (num_pipes <= free_pipes) { - struct dc_stream_state *stream = pipe->stream; - unsigned int vblank_us = ((stream->timing.v_total - stream->timing.v_addressable) * - stream->timing.h_total / - (double)(stream->timing.pix_clk_100hz * 100)) * 1000000; - if (vba->ActiveDRAMClockChangeLatencyMargin[vba->pipe_plane[pipe_idx]] <= 0 && - vblank_us < min_vblank_us) { - *index = i; - min_vblank_us = vblank_us; - valid_assignment_found = true; - } else if (vba->ActiveDRAMClockChangeLatencyMargin[vba->pipe_plane[pipe_idx]] > 0 && - dc->debug.force_subvp_mclk_switch && !valid_assignment_found) { - // Handle case for forcing Sub-VP config. In this case we can assign - // phantom pipes to a surface that has active margin > 0. - *index = i; - valid_assignment_found = true; - } - } - } - pipe_idx++; - } - return valid_assignment_found; -} - -/** - * *************************************************************************************** - * dcn32_enough_pipes_for_subvp: Function to check if there are "enough" pipes for SubVP. - * - * This function returns true if there are enough free pipes - * to create the required phantom pipes for any given stream - * (that does not already have phantom pipe assigned). - * - * e.g. For a 2 stream config where the first stream uses one - * pipe and the second stream uses 2 pipes (i.e. pipe split), - * this function will return true because there is 1 remaining - * pipe which can be used as the phantom pipe for the non pipe - * split pipe. - * - * @param [in] dc: current dc state - * @param [in] context: new dc state - * - * @return: True if there are enough free pipes to assign phantom pipes to at least one - * stream that does not already have phantom pipes assigned. Otherwise false. - * - * *************************************************************************************** - */ -static bool dcn32_enough_pipes_for_subvp(struct dc *dc, struct dc_state *context) -{ - unsigned int i, split_cnt, free_pipes; - unsigned int min_pipe_split = dc->res_pool->pipe_count + 1; // init as max number of pipes + 1 - bool subvp_possible = false; - - for (i = 0; i < dc->res_pool->pipe_count; i++) { - struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - - // Find the minimum pipe split count for non SubVP pipes - if (pipe->stream && pipe->plane_state && !pipe->top_pipe && - pipe->stream->mall_stream_config.type == SUBVP_NONE) { - split_cnt = 0; - while (pipe) { - split_cnt++; - pipe = pipe->bottom_pipe; - } - - if (split_cnt < min_pipe_split) - min_pipe_split = split_cnt; - } - } - - free_pipes = dcn32_get_num_free_pipes(dc, context); - - // SubVP only possible if at least one pipe is being used (i.e. free_pipes - // should not equal to the pipe_count) - if (free_pipes >= min_pipe_split && free_pipes < dc->res_pool->pipe_count) - subvp_possible = true; - - return subvp_possible; -} - static void dcn32_enable_phantom_plane(struct dc *dc, struct dc_state *context, struct dc_stream_state *phantom_stream, @@ -1960,93 +1667,6 @@ static void dcn32_enable_phantom_plane(struct dc *dc, } } -/** - * *************************************************************************************** - * dcn32_set_phantom_stream_timing: Set timing params for the phantom stream - * - * Set timing params of the phantom stream based on calculated output from DML. - * This function first gets the DML pipe index using the DC pipe index, then - * calls into DML (get_subviewport_lines_needed_in_mall) to get the number of - * lines required for SubVP MCLK switching and assigns to the phantom stream - * accordingly. - * - * - The number of SubVP lines calculated in DML does not take into account - * FW processing delays and required pstate allow width, so we must include - * that separately. - * - * - Set phantom backporch = vstartup of main pipe - * - * @param [in] dc: current dc state - * @param [in] context: new dc state - * @param [in] ref_pipe: Main pipe for the phantom stream - * @param [in] pipes: DML pipe params - * @param [in] pipe_cnt: number of DML pipes - * @param [in] dc_pipe_idx: DC pipe index for the main pipe (i.e. ref_pipe) - * - * @return: void - * - * *************************************************************************************** - */ -static void dcn32_set_phantom_stream_timing(struct dc *dc, - struct dc_state *context, - struct pipe_ctx *ref_pipe, - struct dc_stream_state *phantom_stream, - display_e2e_pipe_params_st *pipes, - unsigned int pipe_cnt, - unsigned int dc_pipe_idx) -{ - unsigned int i, pipe_idx; - struct pipe_ctx *pipe; - uint32_t phantom_vactive, phantom_bp, pstate_width_fw_delay_lines; - unsigned int vlevel = context->bw_ctx.dml.vba.VoltageLevel; - unsigned int dcfclk = context->bw_ctx.dml.vba.DCFCLKState[vlevel][context->bw_ctx.dml.vba.maxMpcComb]; - unsigned int socclk = context->bw_ctx.dml.vba.SOCCLKPerState[vlevel]; - - // Find DML pipe index (pipe_idx) using dc_pipe_idx - for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) { - pipe = &context->res_ctx.pipe_ctx[i]; - - if (!pipe->stream) - continue; - - if (i == dc_pipe_idx) - break; - - pipe_idx++; - } - - // Calculate lines required for pstate allow width and FW processing delays - pstate_width_fw_delay_lines = ((double)(dc->caps.subvp_fw_processing_delay_us + - dc->caps.subvp_pstate_allow_width_us) / 1000000) * - (ref_pipe->stream->timing.pix_clk_100hz * 100) / - (double)ref_pipe->stream->timing.h_total; - - // Update clks_cfg for calling into recalculate - pipes[0].clks_cfg.voltage = vlevel; - pipes[0].clks_cfg.dcfclk_mhz = dcfclk; - pipes[0].clks_cfg.socclk_mhz = socclk; - - // DML calculation for MALL region doesn't take into account FW delay - // and required pstate allow width for multi-display cases - phantom_vactive = get_subviewport_lines_needed_in_mall(&context->bw_ctx.dml, pipes, pipe_cnt, pipe_idx) + - pstate_width_fw_delay_lines; - - // For backporch of phantom pipe, use vstartup of the main pipe - phantom_bp = get_vstartup(&context->bw_ctx.dml, pipes, pipe_cnt, pipe_idx); - - phantom_stream->dst.y = 0; - phantom_stream->dst.height = phantom_vactive; - phantom_stream->src.y = 0; - phantom_stream->src.height = phantom_vactive; - - phantom_stream->timing.v_addressable = phantom_vactive; - phantom_stream->timing.v_front_porch = 1; - phantom_stream->timing.v_total = phantom_stream->timing.v_addressable + - phantom_stream->timing.v_front_porch + - phantom_stream->timing.v_sync_width + - phantom_bp; -} - static struct dc_stream_state *dcn32_enable_phantom_stream(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, @@ -2068,13 +1688,16 @@ static struct dc_stream_state *dcn32_enable_phantom_stream(struct dc *dc, memcpy(&phantom_stream->timing, &ref_pipe->stream->timing, sizeof(phantom_stream->timing)); memcpy(&phantom_stream->src, &ref_pipe->stream->src, sizeof(phantom_stream->src)); memcpy(&phantom_stream->dst, &ref_pipe->stream->dst, sizeof(phantom_stream->dst)); + DC_FP_START(); dcn32_set_phantom_stream_timing(dc, context, ref_pipe, phantom_stream, pipes, pipe_cnt, dc_pipe_idx); + DC_FP_END(); dc_add_stream_to_ctx(dc, context, phantom_stream); return phantom_stream; } -void dcn32_remove_phantom_pipes(struct dc *dc, struct dc_state *context) +// return true if removed piped from ctx, false otherwise +bool dcn32_remove_phantom_pipes(struct dc *dc, struct dc_state *context) { int i; bool removed_pipe = false; @@ -2094,8 +1717,7 @@ void dcn32_remove_phantom_pipes(struct dc *dc, struct dc_state *context) pipe->stream->mall_stream_config.paired_stream = NULL; } } - if (removed_pipe) - dc->hwss.apply_ctx_to_hw(dc, context); + return removed_pipe; } /* TODO: Input to this function should indicate which pipe indexes (or streams) @@ -2132,772 +1754,6 @@ void dcn32_add_phantom_pipes(struct dc *dc, struct dc_state *context, } } -static bool dcn32_split_stream_for_mpc_or_odm( - const struct dc *dc, - struct resource_context *res_ctx, - struct pipe_ctx *pri_pipe, - struct pipe_ctx *sec_pipe, - bool odm) -{ - int pipe_idx = sec_pipe->pipe_idx; - const struct resource_pool *pool = dc->res_pool; - - if (pri_pipe->plane_state) { - /* ODM + window MPO, where MPO window is on left half only */ - if (pri_pipe->plane_state->clip_rect.x + pri_pipe->plane_state->clip_rect.width <= - pri_pipe->stream->src.x + pri_pipe->stream->src.width/2) - return true; - - /* ODM + window MPO, where MPO window is on right half only */ - if (pri_pipe->plane_state->clip_rect.x >= pri_pipe->stream->src.width/2) - return true; - } - - *sec_pipe = *pri_pipe; - - sec_pipe->pipe_idx = pipe_idx; - sec_pipe->plane_res.mi = pool->mis[pipe_idx]; - sec_pipe->plane_res.hubp = pool->hubps[pipe_idx]; - sec_pipe->plane_res.ipp = pool->ipps[pipe_idx]; - sec_pipe->plane_res.xfm = pool->transforms[pipe_idx]; - sec_pipe->plane_res.dpp = pool->dpps[pipe_idx]; - sec_pipe->plane_res.mpcc_inst = pool->dpps[pipe_idx]->inst; - sec_pipe->stream_res.dsc = NULL; - if (odm) { - if (pri_pipe->next_odm_pipe) { - ASSERT(pri_pipe->next_odm_pipe != sec_pipe); - sec_pipe->next_odm_pipe = pri_pipe->next_odm_pipe; - sec_pipe->next_odm_pipe->prev_odm_pipe = sec_pipe; - } - if (pri_pipe->top_pipe && pri_pipe->top_pipe->next_odm_pipe) { - pri_pipe->top_pipe->next_odm_pipe->bottom_pipe = sec_pipe; - sec_pipe->top_pipe = pri_pipe->top_pipe->next_odm_pipe; - } - if (pri_pipe->bottom_pipe && pri_pipe->bottom_pipe->next_odm_pipe) { - pri_pipe->bottom_pipe->next_odm_pipe->top_pipe = sec_pipe; - sec_pipe->bottom_pipe = pri_pipe->bottom_pipe->next_odm_pipe; - } - pri_pipe->next_odm_pipe = sec_pipe; - sec_pipe->prev_odm_pipe = pri_pipe; - ASSERT(sec_pipe->top_pipe == NULL); - - if (!sec_pipe->top_pipe) - sec_pipe->stream_res.opp = pool->opps[pipe_idx]; - else - sec_pipe->stream_res.opp = sec_pipe->top_pipe->stream_res.opp; - if (sec_pipe->stream->timing.flags.DSC == 1) { - dcn20_acquire_dsc(dc, res_ctx, &sec_pipe->stream_res.dsc, pipe_idx); - ASSERT(sec_pipe->stream_res.dsc); - if (sec_pipe->stream_res.dsc == NULL) - return false; - } - } else { - if (pri_pipe->bottom_pipe) { - ASSERT(pri_pipe->bottom_pipe != sec_pipe); - sec_pipe->bottom_pipe = pri_pipe->bottom_pipe; - sec_pipe->bottom_pipe->top_pipe = sec_pipe; - } - pri_pipe->bottom_pipe = sec_pipe; - sec_pipe->top_pipe = pri_pipe; - - ASSERT(pri_pipe->plane_state); - } - - return true; -} - -static struct pipe_ctx *dcn32_find_split_pipe( - struct dc *dc, - struct dc_state *context, - int old_index) -{ - struct pipe_ctx *pipe = NULL; - int i; - - if (old_index >= 0 && context->res_ctx.pipe_ctx[old_index].stream == NULL) { - pipe = &context->res_ctx.pipe_ctx[old_index]; - pipe->pipe_idx = old_index; - } - - if (!pipe) - for (i = dc->res_pool->pipe_count - 1; i >= 0; i--) { - if (dc->current_state->res_ctx.pipe_ctx[i].top_pipe == NULL - && dc->current_state->res_ctx.pipe_ctx[i].prev_odm_pipe == NULL) { - if (context->res_ctx.pipe_ctx[i].stream == NULL) { - pipe = &context->res_ctx.pipe_ctx[i]; - pipe->pipe_idx = i; - break; - } - } - } - - /* - * May need to fix pipes getting tossed from 1 opp to another on flip - * Add for debugging transient underflow during topology updates: - * ASSERT(pipe); - */ - if (!pipe) - for (i = dc->res_pool->pipe_count - 1; i >= 0; i--) { - if (context->res_ctx.pipe_ctx[i].stream == NULL) { - pipe = &context->res_ctx.pipe_ctx[i]; - pipe->pipe_idx = i; - break; - } - } - - return pipe; -} - - -/** - * *************************************************************************************** - * subvp_subvp_schedulable: Determine if SubVP + SubVP config is schedulable - * - * High level algorithm: - * 1. Find longest microschedule length (in us) between the two SubVP pipes - * 2. Check if the worst case overlap (VBLANK in middle of ACTIVE) for both - * pipes still allows for the maximum microschedule to fit in the active - * region for both pipes. - * - * @param [in] dc: current dc state - * @param [in] context: new dc state - * - * @return: bool - True if the SubVP + SubVP config is schedulable, false otherwise - * - * *************************************************************************************** - */ -static bool subvp_subvp_schedulable(struct dc *dc, struct dc_state *context) -{ - struct pipe_ctx *subvp_pipes[2]; - struct dc_stream_state *phantom = NULL; - uint32_t microschedule_lines = 0; - uint32_t index = 0; - uint32_t i; - uint32_t max_microschedule_us = 0; - int32_t vactive1_us, vactive2_us, vblank1_us, vblank2_us; - - for (i = 0; i < dc->res_pool->pipe_count; i++) { - struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - uint32_t time_us = 0; - - /* Loop to calculate the maximum microschedule time between the two SubVP pipes, - * and also to store the two main SubVP pipe pointers in subvp_pipes[2]. - */ - if (pipe->stream && pipe->plane_state && !pipe->top_pipe && - pipe->stream->mall_stream_config.type == SUBVP_MAIN) { - phantom = pipe->stream->mall_stream_config.paired_stream; - microschedule_lines = (phantom->timing.v_total - phantom->timing.v_front_porch) + - phantom->timing.v_addressable; - - // Round up when calculating microschedule time - time_us = ((microschedule_lines * phantom->timing.h_total + - phantom->timing.pix_clk_100hz * 100 - 1) / - (double)(phantom->timing.pix_clk_100hz * 100)) * 1000000 + - dc->caps.subvp_prefetch_end_to_mall_start_us + - dc->caps.subvp_fw_processing_delay_us; - if (time_us > max_microschedule_us) - max_microschedule_us = time_us; - - subvp_pipes[index] = pipe; - index++; - - // Maximum 2 SubVP pipes - if (index == 2) - break; - } - } - vactive1_us = ((subvp_pipes[0]->stream->timing.v_addressable * subvp_pipes[0]->stream->timing.h_total) / - (double)(subvp_pipes[0]->stream->timing.pix_clk_100hz * 100)) * 1000000; - vactive2_us = ((subvp_pipes[1]->stream->timing.v_addressable * subvp_pipes[1]->stream->timing.h_total) / - (double)(subvp_pipes[1]->stream->timing.pix_clk_100hz * 100)) * 1000000; - vblank1_us = (((subvp_pipes[0]->stream->timing.v_total - subvp_pipes[0]->stream->timing.v_addressable) * - subvp_pipes[0]->stream->timing.h_total) / - (double)(subvp_pipes[0]->stream->timing.pix_clk_100hz * 100)) * 1000000; - vblank2_us = (((subvp_pipes[1]->stream->timing.v_total - subvp_pipes[1]->stream->timing.v_addressable) * - subvp_pipes[1]->stream->timing.h_total) / - (double)(subvp_pipes[1]->stream->timing.pix_clk_100hz * 100)) * 1000000; - - if ((vactive1_us - vblank2_us) / 2 > max_microschedule_us && - (vactive2_us - vblank1_us) / 2 > max_microschedule_us) - return true; - - return false; -} - -/** - * *************************************************************************************** - * subvp_drr_schedulable: Determine if SubVP + DRR config is schedulable - * - * High level algorithm: - * 1. Get timing for SubVP pipe, phantom pipe, and DRR pipe - * 2. Determine the frame time for the DRR display when adding required margin for MCLK switching - * (the margin is equal to the MALL region + DRR margin (500us)) - * 3.If (SubVP Active - Prefetch > Stretched DRR frame + max(MALL region, Stretched DRR frame)) - * then report the configuration as supported - * - * @param [in] dc: current dc state - * @param [in] context: new dc state - * @param [in] drr_pipe: DRR pipe_ctx for the SubVP + DRR config - * - * @return: bool - True if the SubVP + DRR config is schedulable, false otherwise - * - * *************************************************************************************** - */ -static bool subvp_drr_schedulable(struct dc *dc, struct dc_state *context, struct pipe_ctx *drr_pipe) -{ - bool schedulable = false; - uint32_t i; - struct pipe_ctx *pipe = NULL; - struct dc_crtc_timing *main_timing = NULL; - struct dc_crtc_timing *phantom_timing = NULL; - struct dc_crtc_timing *drr_timing = NULL; - int16_t prefetch_us = 0; - int16_t mall_region_us = 0; - int16_t drr_frame_us = 0; // nominal frame time - int16_t subvp_active_us = 0; - int16_t stretched_drr_us = 0; - int16_t drr_stretched_vblank_us = 0; - int16_t max_vblank_mallregion = 0; - - // Find SubVP pipe - for (i = 0; i < dc->res_pool->pipe_count; i++) { - pipe = &context->res_ctx.pipe_ctx[i]; - - // We check for master pipe, but it shouldn't matter since we only need - // the pipe for timing info (stream should be same for any pipe splits) - if (!pipe->stream || !pipe->plane_state || pipe->top_pipe || pipe->prev_odm_pipe) - continue; - - // Find the SubVP pipe - if (pipe->stream->mall_stream_config.type == SUBVP_MAIN) - break; - } - - main_timing = &pipe->stream->timing; - phantom_timing = &pipe->stream->mall_stream_config.paired_stream->timing; - drr_timing = &drr_pipe->stream->timing; - prefetch_us = (phantom_timing->v_total - phantom_timing->v_front_porch) * phantom_timing->h_total / - (double)(phantom_timing->pix_clk_100hz * 100) * 1000000 + - dc->caps.subvp_prefetch_end_to_mall_start_us; - subvp_active_us = main_timing->v_addressable * main_timing->h_total / - (double)(main_timing->pix_clk_100hz * 100) * 1000000; - drr_frame_us = drr_timing->v_total * drr_timing->h_total / - (double)(drr_timing->pix_clk_100hz * 100) * 1000000; - // P-State allow width and FW delays already included phantom_timing->v_addressable - mall_region_us = phantom_timing->v_addressable * phantom_timing->h_total / - (double)(phantom_timing->pix_clk_100hz * 100) * 1000000; - stretched_drr_us = drr_frame_us + mall_region_us + SUBVP_DRR_MARGIN_US; - drr_stretched_vblank_us = (drr_timing->v_total - drr_timing->v_addressable) * drr_timing->h_total / - (double)(drr_timing->pix_clk_100hz * 100) * 1000000 + (stretched_drr_us - drr_frame_us); - max_vblank_mallregion = drr_stretched_vblank_us > mall_region_us ? drr_stretched_vblank_us : mall_region_us; - - /* We consider SubVP + DRR schedulable if the stretched frame duration of the DRR display (i.e. the - * highest refresh rate + margin that can support UCLK P-State switch) passes the static analysis - * for VBLANK: (VACTIVE region of the SubVP pipe can fit the MALL prefetch, VBLANK frame time, - * and the max of (VBLANK blanking time, MALL region)). - */ - if (stretched_drr_us < (1 / (double)drr_timing->min_refresh_in_uhz) * 1000000 * 1000000 && - subvp_active_us - prefetch_us - stretched_drr_us - max_vblank_mallregion > 0) - schedulable = true; - - return schedulable; -} - -/** - * *************************************************************************************** - * subvp_vblank_schedulable: Determine if SubVP + VBLANK config is schedulable - * - * High level algorithm: - * 1. Get timing for SubVP pipe, phantom pipe, and VBLANK pipe - * 2. If (SubVP Active - Prefetch > Vblank Frame Time + max(MALL region, Vblank blanking time)) - * then report the configuration as supported - * 3. If the VBLANK display is DRR, then take the DRR static schedulability path - * - * @param [in] dc: current dc state - * @param [in] context: new dc state - * - * @return: bool - True if the SubVP + VBLANK/DRR config is schedulable, false otherwise - * - * *************************************************************************************** - */ -static bool subvp_vblank_schedulable(struct dc *dc, struct dc_state *context) -{ - struct pipe_ctx *pipe = NULL; - struct pipe_ctx *subvp_pipe = NULL; - bool found = false; - bool schedulable = false; - uint32_t i = 0; - uint8_t vblank_index = 0; - int16_t prefetch_us = 0; - int16_t mall_region_us = 0; - int16_t vblank_frame_us = 0; - int16_t subvp_active_us = 0; - int16_t vblank_blank_us = 0; - int16_t max_vblank_mallregion = 0; - struct dc_crtc_timing *main_timing = NULL; - struct dc_crtc_timing *phantom_timing = NULL; - struct dc_crtc_timing *vblank_timing = NULL; - - /* For SubVP + VBLANK/DRR cases, we assume there can only be - * a single VBLANK/DRR display. If DML outputs SubVP + VBLANK - * is supported, it is either a single VBLANK case or two VBLANK - * displays which are synchronized (in which case they have identical - * timings). - */ - for (i = 0; i < dc->res_pool->pipe_count; i++) { - pipe = &context->res_ctx.pipe_ctx[i]; - - // We check for master pipe, but it shouldn't matter since we only need - // the pipe for timing info (stream should be same for any pipe splits) - if (!pipe->stream || !pipe->plane_state || pipe->top_pipe || pipe->prev_odm_pipe) - continue; - - if (!found && pipe->stream->mall_stream_config.type == SUBVP_NONE) { - // Found pipe which is not SubVP or Phantom (i.e. the VBLANK pipe). - vblank_index = i; - found = true; - } - - if (!subvp_pipe && pipe->stream->mall_stream_config.type == SUBVP_MAIN) - subvp_pipe = pipe; - } - // Use ignore_msa_timing_param flag to identify as DRR - if (found && pipe->stream->ignore_msa_timing_param) { - // SUBVP + DRR case - schedulable = subvp_drr_schedulable(dc, context, &context->res_ctx.pipe_ctx[vblank_index]); - } else if (found) { - main_timing = &subvp_pipe->stream->timing; - phantom_timing = &subvp_pipe->stream->mall_stream_config.paired_stream->timing; - vblank_timing = &context->res_ctx.pipe_ctx[vblank_index].stream->timing; - // Prefetch time is equal to VACTIVE + BP + VSYNC of the phantom pipe - // Also include the prefetch end to mallstart delay time - prefetch_us = (phantom_timing->v_total - phantom_timing->v_front_porch) * phantom_timing->h_total / - (double)(phantom_timing->pix_clk_100hz * 100) * 1000000 + - dc->caps.subvp_prefetch_end_to_mall_start_us; - // P-State allow width and FW delays already included phantom_timing->v_addressable - mall_region_us = phantom_timing->v_addressable * phantom_timing->h_total / - (double)(phantom_timing->pix_clk_100hz * 100) * 1000000; - vblank_frame_us = vblank_timing->v_total * vblank_timing->h_total / - (double)(vblank_timing->pix_clk_100hz * 100) * 1000000; - vblank_blank_us = (vblank_timing->v_total - vblank_timing->v_addressable) * vblank_timing->h_total / - (double)(vblank_timing->pix_clk_100hz * 100) * 1000000; - subvp_active_us = main_timing->v_addressable * main_timing->h_total / - (double)(main_timing->pix_clk_100hz * 100) * 1000000; - max_vblank_mallregion = vblank_blank_us > mall_region_us ? vblank_blank_us : mall_region_us; - - // Schedulable if VACTIVE region of the SubVP pipe can fit the MALL prefetch, VBLANK frame time, - // and the max of (VBLANK blanking time, MALL region) - // TODO: Possibly add some margin (i.e. the below conditions should be [...] > X instead of [...] > 0) - if (subvp_active_us - prefetch_us - vblank_frame_us - max_vblank_mallregion > 0) - schedulable = true; - } - return schedulable; -} - -/** - * ******************************************************************************************** - * subvp_validate_static_schedulability: Check which SubVP case is calculated and handle - * static analysis based on the case. - * - * Three cases: - * 1. SubVP + SubVP - * 2. SubVP + VBLANK (DRR checked internally) - * 3. SubVP + VACTIVE (currently unsupported) - * - * @param [in] dc: current dc state - * @param [in] context: new dc state - * @param [in] vlevel: Voltage level calculated by DML - * - * @return: bool - True if statically schedulable, false otherwise - * - * ******************************************************************************************** - */ -static bool subvp_validate_static_schedulability(struct dc *dc, - struct dc_state *context, - int vlevel) -{ - bool schedulable = true; // true by default for single display case - struct vba_vars_st *vba = &context->bw_ctx.dml.vba; - uint32_t i, pipe_idx; - uint8_t subvp_count = 0; - uint8_t vactive_count = 0; - - for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) { - struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - - if (!pipe->stream) - continue; - - if (pipe->plane_state && !pipe->top_pipe && - pipe->stream->mall_stream_config.type == SUBVP_MAIN) - subvp_count++; - - // Count how many planes are capable of VACTIVE switching (SubVP + VACTIVE unsupported) - if (vba->ActiveDRAMClockChangeLatencyMargin[vba->pipe_plane[pipe_idx]] > 0) { - vactive_count++; - } - pipe_idx++; - } - - if (subvp_count == 2) { - // Static schedulability check for SubVP + SubVP case - schedulable = subvp_subvp_schedulable(dc, context); - } else if (vba->DRAMClockChangeSupport[vlevel][vba->maxMpcComb] == dm_dram_clock_change_vblank_w_mall_sub_vp) { - // Static schedulability check for SubVP + VBLANK case. Also handle the case where - // DML outputs SubVP + VBLANK + VACTIVE (DML will report as SubVP + VBLANK) - if (vactive_count > 0) - schedulable = false; - else - schedulable = subvp_vblank_schedulable(dc, context); - } else if (vba->DRAMClockChangeSupport[vlevel][vba->maxMpcComb] == dm_dram_clock_change_vactive_w_mall_sub_vp) { - // SubVP + VACTIVE currently unsupported - schedulable = false; - } - return schedulable; -} - -static void dcn32_full_validate_bw_helper(struct dc *dc, - struct dc_state *context, - display_e2e_pipe_params_st *pipes, - int *vlevel, - int *split, - bool *merge, - int *pipe_cnt) -{ - struct vba_vars_st *vba = &context->bw_ctx.dml.vba; - unsigned int dc_pipe_idx = 0; - bool found_supported_config = false; - struct pipe_ctx *pipe = NULL; - uint32_t non_subvp_pipes = 0; - bool drr_pipe_found = false; - uint32_t drr_pipe_index = 0; - uint32_t i = 0; - - /* - * DML favors voltage over p-state, but we're more interested in - * supporting p-state over voltage. We can't support p-state in - * prefetch mode > 0 so try capping the prefetch mode to start. - */ - context->bw_ctx.dml.soc.allow_for_pstate_or_stutter_in_vblank_final = - dm_prefetch_support_uclk_fclk_and_stutter; - *vlevel = dml_get_voltage_level(&context->bw_ctx.dml, pipes, *pipe_cnt); - /* This may adjust vlevel and maxMpcComb */ - if (*vlevel < context->bw_ctx.dml.soc.num_states) - *vlevel = dcn20_validate_apply_pipe_split_flags(dc, context, *vlevel, split, merge); - - /* Conditions for setting up phantom pipes for SubVP: - * 1. Not force disable SubVP - * 2. Full update (i.e. !fast_validate) - * 3. Enough pipes are available to support SubVP (TODO: Which pipes will use VACTIVE / VBLANK / SUBVP?) - * 4. Display configuration passes validation - * 5. (Config doesn't support MCLK in VACTIVE/VBLANK || dc->debug.force_subvp_mclk_switch) - */ - if (!dc->debug.force_disable_subvp && - (*vlevel == context->bw_ctx.dml.soc.num_states || - vba->DRAMClockChangeSupport[*vlevel][vba->maxMpcComb] == dm_dram_clock_change_unsupported || - dc->debug.force_subvp_mclk_switch)) { - - while (!found_supported_config && dcn32_enough_pipes_for_subvp(dc, context) && - dcn32_assign_subvp_pipe(dc, context, &dc_pipe_idx)) { - - dc->res_pool->funcs->add_phantom_pipes(dc, context, pipes, *pipe_cnt, dc_pipe_idx); - - *pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, false); - *vlevel = dml_get_voltage_level(&context->bw_ctx.dml, pipes, *pipe_cnt); - - if (*vlevel < context->bw_ctx.dml.soc.num_states && - vba->DRAMClockChangeSupport[*vlevel][vba->maxMpcComb] != dm_dram_clock_change_unsupported - && subvp_validate_static_schedulability(dc, context, *vlevel)) { - found_supported_config = true; - } else if (*vlevel < context->bw_ctx.dml.soc.num_states && - vba->DRAMClockChangeSupport[*vlevel][vba->maxMpcComb] == dm_dram_clock_change_unsupported) { - /* Case where 1 SubVP is added, and DML reports MCLK unsupported. This handles - * the case for SubVP + DRR, where the DRR display does not support MCLK switch - * at it's native refresh rate / timing. - */ - for (i = 0; i < dc->res_pool->pipe_count; i++) { - pipe = &context->res_ctx.pipe_ctx[i]; - if (pipe->stream && pipe->plane_state && !pipe->top_pipe && - pipe->stream->mall_stream_config.type == SUBVP_NONE) { - non_subvp_pipes++; - // Use ignore_msa_timing_param flag to identify as DRR - if (pipe->stream->ignore_msa_timing_param) { - drr_pipe_found = true; - drr_pipe_index = i; - } - } - } - // If there is only 1 remaining non SubVP pipe that is DRR, check static - // schedulability for SubVP + DRR. - if (non_subvp_pipes == 1 && drr_pipe_found) { - found_supported_config = subvp_drr_schedulable(dc, - context, &context->res_ctx.pipe_ctx[drr_pipe_index]); - } - } - } - - // If SubVP pipe config is unsupported (or cannot be used for UCLK switching) - // remove phantom pipes and repopulate dml pipes - if (!found_supported_config) { - dc->res_pool->funcs->remove_phantom_pipes(dc, context); - *pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, false); - } else { - // only call dcn20_validate_apply_pipe_split_flags if we found a supported config - memset(split, 0, MAX_PIPES * sizeof(int)); - memset(merge, 0, MAX_PIPES * sizeof(bool)); - *vlevel = dcn20_validate_apply_pipe_split_flags(dc, context, *vlevel, split, merge); - - // If found a supported SubVP config, phantom pipes were added to the context. - // Program timing for the phantom pipes. - dc->hwss.apply_ctx_to_hw(dc, context); - } - } -} - -static bool dcn32_internal_validate_bw( - struct dc *dc, - struct dc_state *context, - display_e2e_pipe_params_st *pipes, - int *pipe_cnt_out, - int *vlevel_out, - bool fast_validate) -{ - bool out = false; - bool repopulate_pipes = false; - int split[MAX_PIPES] = { 0 }; - bool merge[MAX_PIPES] = { false }; - bool newly_split[MAX_PIPES] = { false }; - int pipe_cnt, i, pipe_idx, vlevel; - struct vba_vars_st *vba = &context->bw_ctx.dml.vba; - - ASSERT(pipes); - if (!pipes) - return false; - - // For each full update, remove all existing phantom pipes first - dc->res_pool->funcs->remove_phantom_pipes(dc, context); - - dc->res_pool->funcs->update_soc_for_wm_a(dc, context); - - pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, fast_validate); - - if (!pipe_cnt) { - out = true; - goto validate_out; - } - - dml_log_pipe_params(&context->bw_ctx.dml, pipes, pipe_cnt); - - if (!fast_validate) { - dcn32_full_validate_bw_helper(dc, context, pipes, &vlevel, split, merge, &pipe_cnt); - } - - if (fast_validate || vlevel == context->bw_ctx.dml.soc.num_states || - vba->DRAMClockChangeSupport[vlevel][vba->maxMpcComb] == dm_dram_clock_change_unsupported) { - /* - * If mode is unsupported or there's still no p-state support then - * fall back to favoring voltage. - * - * We don't actually support prefetch mode 2, so require that we - * at least support prefetch mode 1. - */ - context->bw_ctx.dml.soc.allow_for_pstate_or_stutter_in_vblank_final = - dm_prefetch_support_stutter; - - vlevel = dml_get_voltage_level(&context->bw_ctx.dml, pipes, pipe_cnt); - if (vlevel < context->bw_ctx.dml.soc.num_states) { - memset(split, 0, MAX_PIPES * sizeof(int)); - memset(merge, 0, MAX_PIPES * sizeof(bool)); - vlevel = dcn20_validate_apply_pipe_split_flags(dc, context, vlevel, split, merge); - } - } - - dml_log_mode_support_params(&context->bw_ctx.dml); - - if (vlevel == context->bw_ctx.dml.soc.num_states) - goto validate_fail; - - for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) { - struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - struct pipe_ctx *mpo_pipe = pipe->bottom_pipe; - - if (!pipe->stream) - continue; - - /* We only support full screen mpo with ODM */ - if (vba->ODMCombineEnabled[vba->pipe_plane[pipe_idx]] != dm_odm_combine_mode_disabled - && pipe->plane_state && mpo_pipe - && memcmp(&mpo_pipe->plane_res.scl_data.recout, - &pipe->plane_res.scl_data.recout, - sizeof(struct rect)) != 0) { - ASSERT(mpo_pipe->plane_state != pipe->plane_state); - goto validate_fail; - } - pipe_idx++; - } - - /* merge pipes if necessary */ - for (i = 0; i < dc->res_pool->pipe_count; i++) { - struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - - /*skip pipes that don't need merging*/ - if (!merge[i]) - continue; - - /* if ODM merge we ignore mpc tree, mpo pipes will have their own flags */ - if (pipe->prev_odm_pipe) { - /*split off odm pipe*/ - pipe->prev_odm_pipe->next_odm_pipe = pipe->next_odm_pipe; - if (pipe->next_odm_pipe) - pipe->next_odm_pipe->prev_odm_pipe = pipe->prev_odm_pipe; - - pipe->bottom_pipe = NULL; - pipe->next_odm_pipe = NULL; - pipe->plane_state = NULL; - pipe->stream = NULL; - pipe->top_pipe = NULL; - pipe->prev_odm_pipe = NULL; - if (pipe->stream_res.dsc) - dcn20_release_dsc(&context->res_ctx, dc->res_pool, &pipe->stream_res.dsc); - memset(&pipe->plane_res, 0, sizeof(pipe->plane_res)); - memset(&pipe->stream_res, 0, sizeof(pipe->stream_res)); - repopulate_pipes = true; - } else if (pipe->top_pipe && pipe->top_pipe->plane_state == pipe->plane_state) { - struct pipe_ctx *top_pipe = pipe->top_pipe; - struct pipe_ctx *bottom_pipe = pipe->bottom_pipe; - - top_pipe->bottom_pipe = bottom_pipe; - if (bottom_pipe) - bottom_pipe->top_pipe = top_pipe; - - pipe->top_pipe = NULL; - pipe->bottom_pipe = NULL; - pipe->plane_state = NULL; - pipe->stream = NULL; - memset(&pipe->plane_res, 0, sizeof(pipe->plane_res)); - memset(&pipe->stream_res, 0, sizeof(pipe->stream_res)); - repopulate_pipes = true; - } else - ASSERT(0); /* Should never try to merge master pipe */ - - } - - for (i = 0, pipe_idx = -1; i < dc->res_pool->pipe_count; i++) { - struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - struct pipe_ctx *old_pipe = &dc->current_state->res_ctx.pipe_ctx[i]; - struct pipe_ctx *hsplit_pipe = NULL; - bool odm; - int old_index = -1; - - if (!pipe->stream || newly_split[i]) - continue; - - pipe_idx++; - odm = vba->ODMCombineEnabled[vba->pipe_plane[pipe_idx]] != dm_odm_combine_mode_disabled; - - if (!pipe->plane_state && !odm) - continue; - - if (split[i]) { - if (odm) { - if (split[i] == 4 && old_pipe->next_odm_pipe && old_pipe->next_odm_pipe->next_odm_pipe) - old_index = old_pipe->next_odm_pipe->next_odm_pipe->pipe_idx; - else if (old_pipe->next_odm_pipe) - old_index = old_pipe->next_odm_pipe->pipe_idx; - } else { - if (split[i] == 4 && old_pipe->bottom_pipe && old_pipe->bottom_pipe->bottom_pipe && - old_pipe->bottom_pipe->bottom_pipe->plane_state == old_pipe->plane_state) - old_index = old_pipe->bottom_pipe->bottom_pipe->pipe_idx; - else if (old_pipe->bottom_pipe && - old_pipe->bottom_pipe->plane_state == old_pipe->plane_state) - old_index = old_pipe->bottom_pipe->pipe_idx; - } - hsplit_pipe = dcn32_find_split_pipe(dc, context, old_index); - ASSERT(hsplit_pipe); - if (!hsplit_pipe) - goto validate_fail; - - if (!dcn32_split_stream_for_mpc_or_odm( - dc, &context->res_ctx, - pipe, hsplit_pipe, odm)) - goto validate_fail; - - newly_split[hsplit_pipe->pipe_idx] = true; - repopulate_pipes = true; - } - if (split[i] == 4) { - struct pipe_ctx *pipe_4to1; - - if (odm && old_pipe->next_odm_pipe) - old_index = old_pipe->next_odm_pipe->pipe_idx; - else if (!odm && old_pipe->bottom_pipe && - old_pipe->bottom_pipe->plane_state == old_pipe->plane_state) - old_index = old_pipe->bottom_pipe->pipe_idx; - else - old_index = -1; - pipe_4to1 = dcn32_find_split_pipe(dc, context, old_index); - ASSERT(pipe_4to1); - if (!pipe_4to1) - goto validate_fail; - if (!dcn32_split_stream_for_mpc_or_odm( - dc, &context->res_ctx, - pipe, pipe_4to1, odm)) - goto validate_fail; - newly_split[pipe_4to1->pipe_idx] = true; - - if (odm && old_pipe->next_odm_pipe && old_pipe->next_odm_pipe->next_odm_pipe - && old_pipe->next_odm_pipe->next_odm_pipe->next_odm_pipe) - old_index = old_pipe->next_odm_pipe->next_odm_pipe->next_odm_pipe->pipe_idx; - else if (!odm && old_pipe->bottom_pipe && old_pipe->bottom_pipe->bottom_pipe && - old_pipe->bottom_pipe->bottom_pipe->bottom_pipe && - old_pipe->bottom_pipe->bottom_pipe->bottom_pipe->plane_state == old_pipe->plane_state) - old_index = old_pipe->bottom_pipe->bottom_pipe->bottom_pipe->pipe_idx; - else - old_index = -1; - pipe_4to1 = dcn32_find_split_pipe(dc, context, old_index); - ASSERT(pipe_4to1); - if (!pipe_4to1) - goto validate_fail; - if (!dcn32_split_stream_for_mpc_or_odm( - dc, &context->res_ctx, - hsplit_pipe, pipe_4to1, odm)) - goto validate_fail; - newly_split[pipe_4to1->pipe_idx] = true; - } - if (odm) - dcn20_build_mapped_resource(dc, context, pipe->stream); - } - - for (i = 0; i < dc->res_pool->pipe_count; i++) { - struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - - if (pipe->plane_state) { - if (!resource_build_scaling_params(pipe)) - goto validate_fail; - } - } - - /* Actual dsc count per stream dsc validation*/ - if (!dcn20_validate_dsc(dc, context)) { - vba->ValidationStatus[vba->soc.num_states] = DML_FAIL_DSC_VALIDATION_FAILURE; - goto validate_fail; - } - - if (repopulate_pipes) - pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, fast_validate); - *vlevel_out = vlevel; - *pipe_cnt_out = pipe_cnt; - - out = true; - goto validate_out; - -validate_fail: - out = false; - -validate_out: - return out; -} - bool dcn32_validate_bandwidth(struct dc *dc, struct dc_state *context, bool fast_validate) @@ -2913,9 +1769,9 @@ bool dcn32_validate_bandwidth(struct dc *dc, BW_VAL_TRACE_COUNT(); - DC_FP_START(); + DC_FP_START(); out = dcn32_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, fast_validate); - DC_FP_END(); + DC_FP_END(); if (pipe_cnt == 0) goto validate_out; @@ -2965,11 +1821,13 @@ int dcn32_populate_dml_pipes_from_context( int i, pipe_cnt; struct resource_context *res_ctx = &context->res_ctx; struct pipe_ctx *pipe; + bool subvp_in_use = false, is_pipe_split_expected[MAX_PIPES]; + int plane_count = 0; + struct dc_crtc_timing *timing; dcn20_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; if (!res_ctx->pipe_ctx[i].stream) continue; @@ -2987,10 +1845,13 @@ int dcn32_populate_dml_pipes_from_context( switch (pipe->stream->mall_stream_config.type) { case SUBVP_MAIN: pipes[pipe_cnt].pipe.src.use_mall_for_pstate_change = dm_use_mall_pstate_change_sub_viewport; + subvp_in_use = true; break; case SUBVP_PHANTOM: pipes[pipe_cnt].pipe.src.use_mall_for_pstate_change = dm_use_mall_pstate_change_phantom_pipe; - pipes[pipe_cnt].pipe.src.use_mall_for_static_screen = dm_use_mall_static_screen_enable; + pipes[pipe_cnt].pipe.src.use_mall_for_static_screen = dm_use_mall_static_screen_disable; + // Disallow unbounded req for SubVP according to DCHUB programming guide + pipes[pipe_cnt].pipe.src.unbounded_req_mode = false; break; case SUBVP_NONE: pipes[pipe_cnt].pipe.src.use_mall_for_pstate_change = dm_use_mall_pstate_change_disable; @@ -3017,581 +1878,91 @@ int dcn32_populate_dml_pipes_from_context( break; } } - pipe_cnt++; - } - - switch (pipe_cnt) { - case 1: - context->bw_ctx.dml.ip.det_buffer_size_kbytes = DCN3_2_MAX_DET_SIZE; - if (pipe->plane_state && !dc->debug.disable_z9_mpc) { - if (!is_dual_plane(pipe->plane_state->format)) { - context->bw_ctx.dml.ip.det_buffer_size_kbytes = DCN3_2_DEFAULT_DET_SIZE; - pipes[0].pipe.src.unbounded_req_mode = true; - if (pipe->plane_state->src_rect.width >= 5120 && - pipe->plane_state->src_rect.height >= 2880) - context->bw_ctx.dml.ip.det_buffer_size_kbytes = 320; // 5K or higher - } - } - break; - case 2: - context->bw_ctx.dml.ip.det_buffer_size_kbytes = DCN3_2_MAX_DET_SIZE / 2; // 576 KB (9 segments) - break; - case 3: - context->bw_ctx.dml.ip.det_buffer_size_kbytes = DCN3_2_MAX_DET_SIZE / 3; // 384 KB (6 segments) - break; - case 4: - default: - context->bw_ctx.dml.ip.det_buffer_size_kbytes = DCN3_2_DEFAULT_DET_SIZE; // 256 KB (4 segments) - break; - } - - return pipe_cnt; -} -void dcn32_calculate_wm_and_dlg_fp( - struct dc *dc, struct dc_state *context, - display_e2e_pipe_params_st *pipes, - int pipe_cnt, - int vlevel) -{ - int i, pipe_idx, vlevel_temp = 0; - - double dcfclk = dcn3_2_soc.clock_limits[0].dcfclk_mhz; - double dcfclk_from_validation = context->bw_ctx.dml.vba.DCFCLKState[vlevel][context->bw_ctx.dml.vba.maxMpcComb]; - unsigned int min_dram_speed_mts = context->bw_ctx.dml.vba.DRAMSpeed; - bool pstate_en = context->bw_ctx.dml.vba.DRAMClockChangeSupport[vlevel][context->bw_ctx.dml.vba.maxMpcComb] != - dm_dram_clock_change_unsupported; - - /* Set B: - * For Set B calculations use clocks from clock_limits[2] when available i.e. when SMU is present, - * otherwise use arbitrary low value from spreadsheet for DCFCLK as lower is safer for watermark - * calculations to cover bootup clocks. - * DCFCLK: soc.clock_limits[2] when available - * UCLK: soc.clock_limits[2] when available - */ - if (dcn3_2_soc.num_states > 2) { - vlevel_temp = 2; - dcfclk = dcn3_2_soc.clock_limits[2].dcfclk_mhz; - } else - dcfclk = 615; //DCFCLK Vmin_lv + /* Calculate the number of planes we have so we can determine + * whether to apply ODM 2to1 policy or not + */ + if (pipe->stream && !pipe->prev_odm_pipe && + (!pipe->top_pipe || pipe->top_pipe->plane_state != pipe->plane_state)) + ++plane_count; - pipes[0].clks_cfg.voltage = vlevel_temp; - pipes[0].clks_cfg.dcfclk_mhz = dcfclk; - pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[vlevel_temp].socclk_mhz; + DC_FP_START(); + is_pipe_split_expected[i] = dcn32_predict_pipe_split(context, pipes[i].pipe, i); + DC_FP_END(); - if (dc->clk_mgr->bw_params->wm_table.nv_entries[WM_B].valid) { - context->bw_ctx.dml.soc.dram_clock_change_latency_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_B].dml_input.pstate_latency_us; - context->bw_ctx.dml.soc.fclk_change_latency_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_B].dml_input.fclk_change_latency_us; - context->bw_ctx.dml.soc.sr_enter_plus_exit_time_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_B].dml_input.sr_enter_plus_exit_time_us; - context->bw_ctx.dml.soc.sr_exit_time_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_B].dml_input.sr_exit_time_us; + pipe_cnt++; } - context->bw_ctx.bw.dcn.watermarks.b.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.b.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.b.frac_urg_bw_nom = get_fraction_of_urgent_bandwidth(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.b.frac_urg_bw_flip = get_fraction_of_urgent_bandwidth_imm_flip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.b.urgent_latency_ns = get_urgent_latency(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.fclk_pstate_change_ns = get_fclk_watermark(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.b.usr_retraining_ns = get_usr_retraining_watermark(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - - /* Set D: - * All clocks min. - * DCFCLK: Min, as reported by PM FW when available - * UCLK : Min, as reported by PM FW when available - * sr_enter_exit/sr_exit should be lower than used for DRAM (TBD after bringup or later, use as decided in Clk Mgr) - */ - if (dcn3_2_soc.num_states > 2) { - vlevel_temp = 0; - dcfclk = dc->clk_mgr->bw_params->clk_table.entries[0].dcfclk_mhz; - } else - dcfclk = 615; //DCFCLK Vmin_lv - - pipes[0].clks_cfg.voltage = vlevel_temp; - pipes[0].clks_cfg.dcfclk_mhz = dcfclk; - pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[vlevel_temp].socclk_mhz; - - if (dc->clk_mgr->bw_params->wm_table.nv_entries[WM_D].valid) { - context->bw_ctx.dml.soc.dram_clock_change_latency_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_D].dml_input.pstate_latency_us; - context->bw_ctx.dml.soc.fclk_change_latency_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_D].dml_input.fclk_change_latency_us; - context->bw_ctx.dml.soc.sr_enter_plus_exit_time_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_D].dml_input.sr_enter_plus_exit_time_us; - context->bw_ctx.dml.soc.sr_exit_time_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_D].dml_input.sr_exit_time_us; - } - context->bw_ctx.bw.dcn.watermarks.d.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.d.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.d.frac_urg_bw_nom = get_fraction_of_urgent_bandwidth(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.d.frac_urg_bw_flip = get_fraction_of_urgent_bandwidth_imm_flip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.d.urgent_latency_ns = get_urgent_latency(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.fclk_pstate_change_ns = get_fclk_watermark(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.d.usr_retraining_ns = get_usr_retraining_watermark(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - - /* Set C, for Dummy P-State: - * All clocks min. - * DCFCLK: Min, as reported by PM FW, when available - * UCLK : Min, as reported by PM FW, when available - * pstate latency as per UCLK state dummy pstate latency + /* Determine whether we will apply ODM 2to1 policy + * Applies to single display and where the number of planes is less than 3 + * For 3 plane case ( 2 MPO planes ), we will not set the policy for the MPO pipes */ - - if (dc->clk_mgr->bw_params->wm_table.nv_entries[WM_C].valid) { - unsigned int min_dram_speed_mts_margin = 160; - - if ((!pstate_en)) - min_dram_speed_mts = dc->clk_mgr->bw_params->clk_table.entries[dc->clk_mgr->bw_params->clk_table.num_entries - 1].memclk_mhz * 16; - - /* find largest table entry that is lower than dram speed, but lower than DPM0 still uses DPM0 */ - for (i = 3; i > 0; i--) - if (min_dram_speed_mts + min_dram_speed_mts_margin > dc->clk_mgr->bw_params->dummy_pstate_table[i].dram_speed_mts) - break; - - context->bw_ctx.dml.soc.dram_clock_change_latency_us = dc->clk_mgr->bw_params->dummy_pstate_table[i].dummy_pstate_latency_us; - context->bw_ctx.dml.soc.dummy_pstate_latency_us = dc->clk_mgr->bw_params->dummy_pstate_table[i].dummy_pstate_latency_us; - context->bw_ctx.dml.soc.fclk_change_latency_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_C].dml_input.fclk_change_latency_us; - context->bw_ctx.dml.soc.sr_enter_plus_exit_time_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_C].dml_input.sr_enter_plus_exit_time_us; - context->bw_ctx.dml.soc.sr_exit_time_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_C].dml_input.sr_exit_time_us; - } - context->bw_ctx.bw.dcn.watermarks.c.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.c.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.c.frac_urg_bw_nom = get_fraction_of_urgent_bandwidth(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.c.frac_urg_bw_flip = get_fraction_of_urgent_bandwidth_imm_flip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.c.urgent_latency_ns = get_urgent_latency(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.fclk_pstate_change_ns = get_fclk_watermark(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.c.usr_retraining_ns = get_usr_retraining_watermark(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - - if ((!pstate_en) && (dc->clk_mgr->bw_params->wm_table.nv_entries[WM_C].valid)) { - /* The only difference between A and C is p-state latency, if p-state is not supported - * with full p-state latency we want to calculate DLG based on dummy p-state latency, - * Set A p-state watermark set to 0 on DCN32, when p-state unsupported, for now keep as DCN32. - */ - context->bw_ctx.bw.dcn.watermarks.a = context->bw_ctx.bw.dcn.watermarks.c; - context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.pstate_change_ns = 0; - } else { - /* Set A: - * All clocks min. - * DCFCLK: Min, as reported by PM FW, when available - * UCLK: Min, as reported by PM FW, when available - */ - dc->res_pool->funcs->update_soc_for_wm_a(dc, context); - context->bw_ctx.bw.dcn.watermarks.a.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.a.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.a.frac_urg_bw_nom = get_fraction_of_urgent_bandwidth(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.a.frac_urg_bw_flip = get_fraction_of_urgent_bandwidth_imm_flip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.a.urgent_latency_ns = get_urgent_latency(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.fclk_pstate_change_ns = get_fclk_watermark(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.a.usr_retraining_ns = get_usr_retraining_watermark(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - } - - pipes[0].clks_cfg.voltage = vlevel; - pipes[0].clks_cfg.dcfclk_mhz = dcfclk_from_validation; - pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[vlevel].socclk_mhz; - - for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) { - if (!context->res_ctx.pipe_ctx[i].stream) + for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) { + if (!res_ctx->pipe_ctx[i].stream) continue; + pipe = &res_ctx->pipe_ctx[i]; + timing = &pipe->stream->timing; - pipes[pipe_idx].clks_cfg.dispclk_mhz = get_dispclk_calculated(&context->bw_ctx.dml, pipes, pipe_cnt); - pipes[pipe_idx].clks_cfg.dppclk_mhz = get_dppclk_calculated(&context->bw_ctx.dml, pipes, pipe_cnt, pipe_idx); - - if (dc->config.forced_clocks) { - pipes[pipe_idx].clks_cfg.dispclk_mhz = context->bw_ctx.dml.soc.clock_limits[0].dispclk_mhz; - pipes[pipe_idx].clks_cfg.dppclk_mhz = context->bw_ctx.dml.soc.clock_limits[0].dppclk_mhz; + pipes[pipe_cnt].pipe.dest.odm_combine_policy = dm_odm_combine_policy_dal; + res_ctx->pipe_ctx[i].stream->odm_2to1_policy_applied = false; + if (context->stream_count == 1 && timing->dsc_cfg.num_slices_h != 1) { + if (dc->debug.enable_single_display_2to1_odm_policy) { + if (!((plane_count > 2) && pipe->top_pipe)) + pipes[pipe_cnt].pipe.dest.odm_combine_policy = dm_odm_combine_policy_2to1; + } + res_ctx->pipe_ctx[i].stream->odm_2to1_policy_applied = true; } - if (dc->debug.min_disp_clk_khz > pipes[pipe_idx].clks_cfg.dispclk_mhz * 1000) - pipes[pipe_idx].clks_cfg.dispclk_mhz = dc->debug.min_disp_clk_khz / 1000.0; - if (dc->debug.min_dpp_clk_khz > pipes[pipe_idx].clks_cfg.dppclk_mhz * 1000) - pipes[pipe_idx].clks_cfg.dppclk_mhz = dc->debug.min_dpp_clk_khz / 1000.0; - - pipe_idx++; + pipe_cnt++; } - context->perf_params.stutter_period_us = context->bw_ctx.dml.vba.StutterPeriod; + /* For DET allocation, we don't want to use DML policy (not optimal for utilizing all + * the DET available for each pipe). Use the DET override input to maintain our driver + * policy. + */ + if (pipe_cnt == 1 && !is_pipe_split_expected[0]) { + pipes[0].pipe.src.det_size_override = DCN3_2_MAX_DET_SIZE; + if (pipe->plane_state && !dc->debug.disable_z9_mpc) { + if (!is_dual_plane(pipe->plane_state->format)) { + pipes[0].pipe.src.det_size_override = DCN3_2_DEFAULT_DET_SIZE; + pipes[0].pipe.src.unbounded_req_mode = true; + if (pipe->plane_state->src_rect.width >= 5120 && + pipe->plane_state->src_rect.height >= 2880) + pipes[0].pipe.src.det_size_override = 320; // 5K or higher + } + } + } else + dcn32_determine_det_override(context, pipes, is_pipe_split_expected, dc->res_pool->pipe_count); - dcn32_calculate_dlg_params(dc, context, pipes, pipe_cnt, vlevel); + // In general cases we want to keep the dram clock change requirement + // (prefer configs that support MCLK switch). Only override to false + // for SubVP + if (subvp_in_use) + context->bw_ctx.dml.soc.dram_clock_change_requirement_final = false; + else + context->bw_ctx.dml.soc.dram_clock_change_requirement_final = true; - if (!pstate_en) - /* Restore full p-state latency */ - 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; + return pipe_cnt; } static struct dc_cap_funcs cap_funcs = { .get_dcc_compression_cap = dcn20_get_dcc_compression_cap }; - -static void dcn32_get_optimal_dcfclk_fclk_for_uclk(unsigned int uclk_mts, - unsigned int *optimal_dcfclk, - unsigned int *optimal_fclk) -{ - double bw_from_dram, bw_from_dram1, bw_from_dram2; - - bw_from_dram1 = uclk_mts * dcn3_2_soc.num_chans * - dcn3_2_soc.dram_channel_width_bytes * (dcn3_2_soc.max_avg_dram_bw_use_normal_percent / 100); - bw_from_dram2 = uclk_mts * dcn3_2_soc.num_chans * - dcn3_2_soc.dram_channel_width_bytes * (dcn3_2_soc.max_avg_sdp_bw_use_normal_percent / 100); - - bw_from_dram = (bw_from_dram1 < bw_from_dram2) ? bw_from_dram1 : bw_from_dram2; - - if (optimal_fclk) - *optimal_fclk = bw_from_dram / - (dcn3_2_soc.fabric_datapath_to_dcn_data_return_bytes * (dcn3_2_soc.max_avg_sdp_bw_use_normal_percent / 100)); - - if (optimal_dcfclk) - *optimal_dcfclk = bw_from_dram / - (dcn3_2_soc.return_bus_width_bytes * (dcn3_2_soc.max_avg_sdp_bw_use_normal_percent / 100)); -} - -void dcn32_calculate_wm_and_dlg( - struct dc *dc, struct dc_state *context, - display_e2e_pipe_params_st *pipes, - int pipe_cnt, - int vlevel) +void dcn32_calculate_wm_and_dlg(struct dc *dc, struct dc_state *context, + display_e2e_pipe_params_st *pipes, + int pipe_cnt, + int vlevel) { DC_FP_START(); - dcn32_calculate_wm_and_dlg_fp( - dc, context, - pipes, - pipe_cnt, - vlevel); + dcn32_calculate_wm_and_dlg_fpu(dc, context, pipes, pipe_cnt, vlevel); DC_FP_END(); } -static bool is_dtbclk_required(struct dc *dc, struct dc_state *context) -{ - int i; - - for (i = 0; i < dc->res_pool->pipe_count; i++) { - if (!context->res_ctx.pipe_ctx[i].stream) - continue; - if (is_dp_128b_132b_signal(&context->res_ctx.pipe_ctx[i])) - return true; - } - return false; -} - -void dcn32_calculate_dlg_params(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - int pipe_cnt, int vlevel) -{ - int i, pipe_idx; - bool usr_retraining_support = false; - - /* Writeback MCIF_WB arbitration parameters */ - dc->res_pool->funcs->set_mcif_arb_params(dc, context, pipes, pipe_cnt); - - context->bw_ctx.bw.dcn.clk.dispclk_khz = context->bw_ctx.dml.vba.DISPCLK * 1000; - context->bw_ctx.bw.dcn.clk.dcfclk_khz = context->bw_ctx.dml.vba.DCFCLK * 1000; - context->bw_ctx.bw.dcn.clk.socclk_khz = context->bw_ctx.dml.vba.SOCCLK * 1000; - context->bw_ctx.bw.dcn.clk.dramclk_khz = context->bw_ctx.dml.vba.DRAMSpeed * 1000 / 16; - context->bw_ctx.bw.dcn.clk.dcfclk_deep_sleep_khz = context->bw_ctx.dml.vba.DCFCLKDeepSleep * 1000; - context->bw_ctx.bw.dcn.clk.fclk_khz = context->bw_ctx.dml.vba.FabricClock * 1000; - 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; - - /* - * TODO: needs FAMS - * 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); - context->bw_ctx.bw.dcn.clk.ref_dtbclk_khz = context->bw_ctx.dml.vba.DTBCLKPerState[vlevel] * 1000; - if (context->bw_ctx.dml.vba.FCLKChangeSupport[vlevel][context->bw_ctx.dml.vba.maxMpcComb] == dm_fclock_change_unsupported) - context->bw_ctx.bw.dcn.clk.fclk_p_state_change_support = false; - else - context->bw_ctx.bw.dcn.clk.fclk_p_state_change_support = true; - - usr_retraining_support = context->bw_ctx.dml.vba.USRRetrainingSupport[vlevel][context->bw_ctx.dml.vba.maxMpcComb]; - ASSERT(usr_retraining_support); - - if (context->bw_ctx.bw.dcn.clk.dispclk_khz < dc->debug.min_disp_clk_khz) - context->bw_ctx.bw.dcn.clk.dispclk_khz = dc->debug.min_disp_clk_khz; - - for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) { - if (!context->res_ctx.pipe_ctx[i].stream) - continue; - pipes[pipe_idx].pipe.dest.vstartup_start = get_vstartup(&context->bw_ctx.dml, pipes, pipe_cnt, - pipe_idx); - pipes[pipe_idx].pipe.dest.vupdate_offset = get_vupdate_offset(&context->bw_ctx.dml, pipes, pipe_cnt, - pipe_idx); - pipes[pipe_idx].pipe.dest.vupdate_width = get_vupdate_width(&context->bw_ctx.dml, pipes, pipe_cnt, - pipe_idx); - pipes[pipe_idx].pipe.dest.vready_offset = get_vready_offset(&context->bw_ctx.dml, pipes, pipe_cnt, - pipe_idx); - if (context->res_ctx.pipe_ctx[i].stream->mall_stream_config.type == SUBVP_PHANTOM) { - // Phantom pipe requires that DET_SIZE = 0 and no unbounded requests - context->res_ctx.pipe_ctx[i].det_buffer_size_kb = 0; - context->res_ctx.pipe_ctx[i].unbounded_req = false; - } else { - context->res_ctx.pipe_ctx[i].det_buffer_size_kb = - context->bw_ctx.dml.ip.det_buffer_size_kbytes; - context->res_ctx.pipe_ctx[i].unbounded_req = pipes[pipe_idx].pipe.src.unbounded_req_mode; - } - if (context->bw_ctx.bw.dcn.clk.dppclk_khz < pipes[pipe_idx].clks_cfg.dppclk_mhz * 1000) - context->bw_ctx.bw.dcn.clk.dppclk_khz = pipes[pipe_idx].clks_cfg.dppclk_mhz * 1000; - context->res_ctx.pipe_ctx[i].plane_res.bw.dppclk_khz = pipes[pipe_idx].clks_cfg.dppclk_mhz * 1000; - context->res_ctx.pipe_ctx[i].pipe_dlg_param = pipes[pipe_idx].pipe.dest; - pipe_idx++; - } - /*save a original dppclock copy*/ - context->bw_ctx.bw.dcn.clk.bw_dppclk_khz = context->bw_ctx.bw.dcn.clk.dppclk_khz; - context->bw_ctx.bw.dcn.clk.bw_dispclk_khz = context->bw_ctx.bw.dcn.clk.dispclk_khz; - context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz = context->bw_ctx.dml.soc.clock_limits[vlevel].dppclk_mhz - * 1000; - context->bw_ctx.bw.dcn.clk.max_supported_dispclk_khz = context->bw_ctx.dml.soc.clock_limits[vlevel].dispclk_mhz - * 1000; - - context->bw_ctx.bw.dcn.compbuf_size_kb = context->bw_ctx.dml.ip.config_return_buffer_size_in_kbytes - - context->bw_ctx.dml.ip.det_buffer_size_kbytes * pipe_idx; - - for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) { - - if (!context->res_ctx.pipe_ctx[i].stream) - continue; - - context->bw_ctx.dml.funcs.rq_dlg_get_dlg_reg_v2(&context->bw_ctx.dml, - &context->res_ctx.pipe_ctx[i].dlg_regs, &context->res_ctx.pipe_ctx[i].ttu_regs, pipes, - pipe_cnt, pipe_idx); - - context->bw_ctx.dml.funcs.rq_dlg_get_rq_reg_v2(&context->res_ctx.pipe_ctx[i].rq_regs, - &context->bw_ctx.dml, pipes, pipe_cnt, pipe_idx); - - pipe_idx++; - } -} - -/* dcn32_update_bw_bounding_box - * This would override some dcn3_2 ip_or_soc initial parameters hardcoded from spreadsheet - * with actual values as per dGPU SKU: - * -with passed few options from dc->config - * -with dentist_vco_frequency from Clk Mgr (currently hardcoded, but might need to get it from PM FW) - * -with passed latency values (passed in ns units) in dc-> bb override for debugging purposes - * -with passed latencies from VBIOS (in 100_ns units) if available for certain dGPU SKU - * -with number of DRAM channels from VBIOS (which differ for certain dGPU SKU of the same ASIC) - * -clocks levels with passed clk_table entries from Clk Mgr as reported by PM FW for different - * clocks (which might differ for certain dGPU SKU of the same ASIC) - */ static void dcn32_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params) { - if (!IS_FPGA_MAXIMUS_DC(dc->ctx->dce_environment)) { - - /* Overrides from dc->config options */ - dcn3_2_ip.clamp_min_dcfclk = dc->config.clamp_min_dcfclk; - - /* Override from passed dc->bb_overrides if available*/ - if ((int)(dcn3_2_soc.sr_exit_time_us * 1000) != dc->bb_overrides.sr_exit_time_ns - && dc->bb_overrides.sr_exit_time_ns) { - dcn3_2_soc.sr_exit_time_us = dc->bb_overrides.sr_exit_time_ns / 1000.0; - } - - if ((int)(dcn3_2_soc.sr_enter_plus_exit_time_us * 1000) - != dc->bb_overrides.sr_enter_plus_exit_time_ns - && dc->bb_overrides.sr_enter_plus_exit_time_ns) { - dcn3_2_soc.sr_enter_plus_exit_time_us = - dc->bb_overrides.sr_enter_plus_exit_time_ns / 1000.0; - } - - if ((int)(dcn3_2_soc.urgent_latency_us * 1000) != dc->bb_overrides.urgent_latency_ns - && dc->bb_overrides.urgent_latency_ns) { - dcn3_2_soc.urgent_latency_us = dc->bb_overrides.urgent_latency_ns / 1000.0; - } - - if ((int)(dcn3_2_soc.dram_clock_change_latency_us * 1000) - != dc->bb_overrides.dram_clock_change_latency_ns - && dc->bb_overrides.dram_clock_change_latency_ns) { - dcn3_2_soc.dram_clock_change_latency_us = - dc->bb_overrides.dram_clock_change_latency_ns / 1000.0; - } - - if ((int)(dcn3_2_soc.dummy_pstate_latency_us * 1000) - != dc->bb_overrides.dummy_clock_change_latency_ns - && dc->bb_overrides.dummy_clock_change_latency_ns) { - dcn3_2_soc.dummy_pstate_latency_us = - dc->bb_overrides.dummy_clock_change_latency_ns / 1000.0; - } - - /* Override from VBIOS if VBIOS bb_info available */ - if (dc->ctx->dc_bios->funcs->get_soc_bb_info) { - struct bp_soc_bb_info bb_info = {0}; - - if (dc->ctx->dc_bios->funcs->get_soc_bb_info(dc->ctx->dc_bios, &bb_info) == BP_RESULT_OK) { - if (bb_info.dram_clock_change_latency_100ns > 0) - dcn3_2_soc.dram_clock_change_latency_us = bb_info.dram_clock_change_latency_100ns * 10; - - if (bb_info.dram_sr_enter_exit_latency_100ns > 0) - dcn3_2_soc.sr_enter_plus_exit_time_us = bb_info.dram_sr_enter_exit_latency_100ns * 10; - - if (bb_info.dram_sr_exit_latency_100ns > 0) - dcn3_2_soc.sr_exit_time_us = bb_info.dram_sr_exit_latency_100ns * 10; - } - } - - /* Override from VBIOS for num_chan */ - if (dc->ctx->dc_bios->vram_info.num_chans) - dcn3_2_soc.num_chans = dc->ctx->dc_bios->vram_info.num_chans; - - if (dc->ctx->dc_bios->vram_info.dram_channel_width_bytes) - dcn3_2_soc.dram_channel_width_bytes = dc->ctx->dc_bios->vram_info.dram_channel_width_bytes; - - } - - /* Override dispclk_dppclk_vco_speed_mhz from Clk Mgr */ - dcn3_2_soc.dispclk_dppclk_vco_speed_mhz = dc->clk_mgr->dentist_vco_freq_khz / 1000.0; - dc->dml.soc.dispclk_dppclk_vco_speed_mhz = dc->clk_mgr->dentist_vco_freq_khz / 1000.0; - - /* Overrides Clock levelsfrom CLK Mgr table entries as reported by PM FW */ - if ((!IS_FPGA_MAXIMUS_DC(dc->ctx->dce_environment)) && (bw_params->clk_table.entries[0].memclk_mhz)) { - unsigned int i = 0, j = 0, num_states = 0; - - unsigned int dcfclk_mhz[DC__VOLTAGE_STATES] = {0}; - unsigned int dram_speed_mts[DC__VOLTAGE_STATES] = {0}; - unsigned int optimal_uclk_for_dcfclk_sta_targets[DC__VOLTAGE_STATES] = {0}; - unsigned int optimal_dcfclk_for_uclk[DC__VOLTAGE_STATES] = {0}; - - unsigned int dcfclk_sta_targets[DC__VOLTAGE_STATES] = {615, 906, 1324, 1564}; - unsigned int num_dcfclk_sta_targets = 4, num_uclk_states = 0; - unsigned int max_dcfclk_mhz = 0, max_dispclk_mhz = 0, max_dppclk_mhz = 0, max_phyclk_mhz = 0; - - for (i = 0; i < MAX_NUM_DPM_LVL; i++) { - if (bw_params->clk_table.entries[i].dcfclk_mhz > max_dcfclk_mhz) - max_dcfclk_mhz = bw_params->clk_table.entries[i].dcfclk_mhz; - if (bw_params->clk_table.entries[i].dispclk_mhz > max_dispclk_mhz) - max_dispclk_mhz = bw_params->clk_table.entries[i].dispclk_mhz; - if (bw_params->clk_table.entries[i].dppclk_mhz > max_dppclk_mhz) - max_dppclk_mhz = bw_params->clk_table.entries[i].dppclk_mhz; - if (bw_params->clk_table.entries[i].phyclk_mhz > max_phyclk_mhz) - max_phyclk_mhz = bw_params->clk_table.entries[i].phyclk_mhz; - } - if (!max_dcfclk_mhz) - max_dcfclk_mhz = dcn3_2_soc.clock_limits[0].dcfclk_mhz; - if (!max_dispclk_mhz) - max_dispclk_mhz = dcn3_2_soc.clock_limits[0].dispclk_mhz; - if (!max_dppclk_mhz) - max_dppclk_mhz = dcn3_2_soc.clock_limits[0].dppclk_mhz; - if (!max_phyclk_mhz) - max_phyclk_mhz = dcn3_2_soc.clock_limits[0].phyclk_mhz; - - if (max_dcfclk_mhz > dcfclk_sta_targets[num_dcfclk_sta_targets-1]) { - // If max DCFCLK is greater than the max DCFCLK STA target, insert into the DCFCLK STA target array - dcfclk_sta_targets[num_dcfclk_sta_targets] = max_dcfclk_mhz; - num_dcfclk_sta_targets++; - } else if (max_dcfclk_mhz < dcfclk_sta_targets[num_dcfclk_sta_targets-1]) { - // If max DCFCLK is less than the max DCFCLK STA target, cap values and remove duplicates - for (i = 0; i < num_dcfclk_sta_targets; i++) { - if (dcfclk_sta_targets[i] > max_dcfclk_mhz) { - dcfclk_sta_targets[i] = max_dcfclk_mhz; - break; - } - } - // Update size of array since we "removed" duplicates - num_dcfclk_sta_targets = i + 1; - } - - num_uclk_states = bw_params->clk_table.num_entries; - - // Calculate optimal dcfclk for each uclk - for (i = 0; i < num_uclk_states; i++) { - dcn32_get_optimal_dcfclk_fclk_for_uclk(bw_params->clk_table.entries[i].memclk_mhz * 16, - &optimal_dcfclk_for_uclk[i], NULL); - if (optimal_dcfclk_for_uclk[i] < bw_params->clk_table.entries[0].dcfclk_mhz) { - optimal_dcfclk_for_uclk[i] = bw_params->clk_table.entries[0].dcfclk_mhz; - } - } - - // Calculate optimal uclk for each dcfclk sta target - for (i = 0; i < num_dcfclk_sta_targets; i++) { - for (j = 0; j < num_uclk_states; j++) { - if (dcfclk_sta_targets[i] < optimal_dcfclk_for_uclk[j]) { - optimal_uclk_for_dcfclk_sta_targets[i] = - bw_params->clk_table.entries[j].memclk_mhz * 16; - break; - } - } - } - - i = 0; - j = 0; - // create the final dcfclk and uclk table - while (i < num_dcfclk_sta_targets && j < num_uclk_states && num_states < DC__VOLTAGE_STATES) { - if (dcfclk_sta_targets[i] < optimal_dcfclk_for_uclk[j] && i < num_dcfclk_sta_targets) { - dcfclk_mhz[num_states] = dcfclk_sta_targets[i]; - dram_speed_mts[num_states++] = optimal_uclk_for_dcfclk_sta_targets[i++]; - } else { - if (j < num_uclk_states && optimal_dcfclk_for_uclk[j] <= max_dcfclk_mhz) { - dcfclk_mhz[num_states] = optimal_dcfclk_for_uclk[j]; - dram_speed_mts[num_states++] = bw_params->clk_table.entries[j++].memclk_mhz * 16; - } else { - j = num_uclk_states; - } - } - } - - while (i < num_dcfclk_sta_targets && num_states < DC__VOLTAGE_STATES) { - dcfclk_mhz[num_states] = dcfclk_sta_targets[i]; - dram_speed_mts[num_states++] = optimal_uclk_for_dcfclk_sta_targets[i++]; - } - - while (j < num_uclk_states && num_states < DC__VOLTAGE_STATES && - optimal_dcfclk_for_uclk[j] <= max_dcfclk_mhz) { - dcfclk_mhz[num_states] = optimal_dcfclk_for_uclk[j]; - dram_speed_mts[num_states++] = bw_params->clk_table.entries[j++].memclk_mhz * 16; - } - - dcn3_2_soc.num_states = num_states; - for (i = 0; i < dcn3_2_soc.num_states; i++) { - dcn3_2_soc.clock_limits[i].state = i; - dcn3_2_soc.clock_limits[i].dcfclk_mhz = dcfclk_mhz[i]; - dcn3_2_soc.clock_limits[i].fabricclk_mhz = dcfclk_mhz[i]; - - /* Fill all states with max values of all these clocks */ - dcn3_2_soc.clock_limits[i].dispclk_mhz = max_dispclk_mhz; - dcn3_2_soc.clock_limits[i].dppclk_mhz = max_dppclk_mhz; - dcn3_2_soc.clock_limits[i].phyclk_mhz = max_phyclk_mhz; - dcn3_2_soc.clock_limits[i].dscclk_mhz = max_dispclk_mhz / 3; - - /* Populate from bw_params for DTBCLK, SOCCLK */ - if (i > 0) { - if (!bw_params->clk_table.entries[i].dtbclk_mhz) { - dcn3_2_soc.clock_limits[i].dtbclk_mhz = dcn3_2_soc.clock_limits[i-1].dtbclk_mhz; - } else { - dcn3_2_soc.clock_limits[i].dtbclk_mhz = bw_params->clk_table.entries[i].dtbclk_mhz; - } - } else if (bw_params->clk_table.entries[i].dtbclk_mhz) { - dcn3_2_soc.clock_limits[i].dtbclk_mhz = bw_params->clk_table.entries[i].dtbclk_mhz; - } - - if (!bw_params->clk_table.entries[i].socclk_mhz && i > 0) - dcn3_2_soc.clock_limits[i].socclk_mhz = dcn3_2_soc.clock_limits[i-1].socclk_mhz; - else - dcn3_2_soc.clock_limits[i].socclk_mhz = bw_params->clk_table.entries[i].socclk_mhz; - - if (!dram_speed_mts[i] && i > 0) - dcn3_2_soc.clock_limits[i].dram_speed_mts = dcn3_2_soc.clock_limits[i-1].dram_speed_mts; - else - dcn3_2_soc.clock_limits[i].dram_speed_mts = dram_speed_mts[i]; - - /* These clocks cannot come from bw_params, always fill from dcn3_2_soc[0] */ - /* PHYCLK_D18, PHYCLK_D32 */ - dcn3_2_soc.clock_limits[i].phyclk_d18_mhz = dcn3_2_soc.clock_limits[0].phyclk_d18_mhz; - dcn3_2_soc.clock_limits[i].phyclk_d32_mhz = dcn3_2_soc.clock_limits[0].phyclk_d32_mhz; - } - - /* Re-init DML with updated bb */ - dml_init_instance(&dc->dml, &dcn3_2_soc, &dcn3_2_ip, DML_PROJECT_DCN32); - if (dc->current_state) - dml_init_instance(&dc->current_state->bw_ctx.dml, &dcn3_2_soc, &dcn3_2_ip, DML_PROJECT_DCN32); - } + DC_FP_START(); + dcn32_update_bw_bounding_box_fpu(dc, bw_params); + DC_FP_END(); } static struct resource_funcs dcn32_res_pool_funcs = { @@ -3602,7 +1973,7 @@ static struct resource_funcs dcn32_res_pool_funcs = { .validate_bandwidth = dcn32_validate_bandwidth, .calculate_wm_and_dlg = dcn32_calculate_wm_and_dlg, .populate_dml_pipes = dcn32_populate_dml_pipes_from_context, - .acquire_idle_pipe_for_layer = dcn20_acquire_idle_pipe_for_layer, + .acquire_idle_pipe_for_head_pipe_in_layer = dcn32_acquire_idle_pipe_for_head_pipe_in_layer, .add_stream_to_ctx = dcn30_add_stream_to_ctx, .add_dsc_to_stream_resource = dcn20_add_dsc_to_stream_resource, .remove_stream_from_ctx = dcn20_remove_stream_from_ctx, @@ -3631,7 +2002,7 @@ static bool dcn32_resource_construct( uint32_t pipe_fuses = 0; uint32_t num_pipes = 4; - DC_FP_START(); + DC_FP_START(); ctx->dc_bios->regs = &bios_regs; @@ -3689,6 +2060,7 @@ static bool dcn32_resource_construct( dc->caps.post_blend_color_processing = true; dc->caps.force_dp_tps4_for_cp2520 = true; dc->caps.dp_hpo = true; + dc->caps.dp_hdmi21_pcon_support = true; dc->caps.edp_dsc_support = true; dc->caps.extended_aux_timeout_support = true; dc->caps.dmcub_support = true; @@ -3708,7 +2080,7 @@ static bool dcn32_resource_construct( dc->caps.color.dpp.dgam_rom_for_yuv = 0; dc->caps.color.dpp.hw_3d_lut = 1; - dc->caps.color.dpp.ogam_ram = 0; //Blnd Gam also removed + dc->caps.color.dpp.ogam_ram = 0; // no OGAM in DPP since DCN1 // no OGAM ROM on DCN2 and later ASICs dc->caps.color.dpp.ogam_rom_caps.srgb = 0; dc->caps.color.dpp.ogam_rom_caps.bt2020 = 0; @@ -3970,13 +2342,13 @@ static bool dcn32_resource_construct( pool->base.oem_device = NULL; } - DC_FP_END(); + DC_FP_END(); return true; create_fail: - DC_FP_END(); + DC_FP_END(); dcn32_resource_destruct(pool); @@ -4000,3 +2372,108 @@ struct resource_pool *dcn32_create_resource_pool( kfree(pool); return NULL; } + +static struct pipe_ctx *find_idle_secondary_pipe_check_mpo( + struct resource_context *res_ctx, + const struct resource_pool *pool, + const struct pipe_ctx *primary_pipe) +{ + int i; + struct pipe_ctx *secondary_pipe = NULL; + struct pipe_ctx *next_odm_mpo_pipe = NULL; + int primary_index, preferred_pipe_idx; + struct pipe_ctx *old_primary_pipe = NULL; + + /* + * Modified from find_idle_secondary_pipe + * With windowed MPO and ODM, we want to avoid the case where we want a + * free pipe for the left side but the free pipe is being used on the + * right side. + * Add check on current_state if the primary_pipe is the left side, + * to check the right side ( primary_pipe->next_odm_pipe ) to see if + * it is using a pipe for MPO ( primary_pipe->next_odm_pipe->bottom_pipe ) + * - If so, then don't use this pipe + * EXCEPTION - 3 plane ( 2 MPO plane ) case + * - in this case, the primary pipe has already gotten a free pipe for the + * MPO window in the left + * - when it tries to get a free pipe for the MPO window on the right, + * it will see that it is already assigned to the right side + * ( primary_pipe->next_odm_pipe ). But in this case, we want this + * free pipe, since it will be for the right side. So add an + * additional condition, that skipping the free pipe on the right only + * applies if the primary pipe has no bottom pipe currently assigned + */ + if (primary_pipe) { + primary_index = primary_pipe->pipe_idx; + old_primary_pipe = &primary_pipe->stream->ctx->dc->current_state->res_ctx.pipe_ctx[primary_index]; + if ((old_primary_pipe->next_odm_pipe) && (old_primary_pipe->next_odm_pipe->bottom_pipe) + && (!primary_pipe->bottom_pipe)) + next_odm_mpo_pipe = old_primary_pipe->next_odm_pipe->bottom_pipe; + + preferred_pipe_idx = (pool->pipe_count - 1) - primary_pipe->pipe_idx; + if ((res_ctx->pipe_ctx[preferred_pipe_idx].stream == NULL) && + !(next_odm_mpo_pipe && next_odm_mpo_pipe->pipe_idx == preferred_pipe_idx)) { + secondary_pipe = &res_ctx->pipe_ctx[preferred_pipe_idx]; + secondary_pipe->pipe_idx = preferred_pipe_idx; + } + } + + /* + * search backwards for the second pipe to keep pipe + * assignment more consistent + */ + if (!secondary_pipe) + for (i = pool->pipe_count - 1; i >= 0; i--) { + if ((res_ctx->pipe_ctx[i].stream == NULL) && + !(next_odm_mpo_pipe && next_odm_mpo_pipe->pipe_idx == i)) { + secondary_pipe = &res_ctx->pipe_ctx[i]; + secondary_pipe->pipe_idx = i; + break; + } + } + + return secondary_pipe; +} + +struct pipe_ctx *dcn32_acquire_idle_pipe_for_head_pipe_in_layer( + struct dc_state *state, + const struct resource_pool *pool, + struct dc_stream_state *stream, + struct pipe_ctx *head_pipe) +{ + struct resource_context *res_ctx = &state->res_ctx; + struct pipe_ctx *idle_pipe, *pipe; + struct resource_context *old_ctx = &stream->ctx->dc->current_state->res_ctx; + int head_index; + + if (!head_pipe) + ASSERT(0); + + /* + * Modified from dcn20_acquire_idle_pipe_for_layer + * Check if head_pipe in old_context already has bottom_pipe allocated. + * - If so, check if that pipe is available in the current context. + * -- If so, reuse pipe from old_context + */ + head_index = head_pipe->pipe_idx; + pipe = &old_ctx->pipe_ctx[head_index]; + if (pipe->bottom_pipe && res_ctx->pipe_ctx[pipe->bottom_pipe->pipe_idx].stream == NULL) { + idle_pipe = &res_ctx->pipe_ctx[pipe->bottom_pipe->pipe_idx]; + idle_pipe->pipe_idx = pipe->bottom_pipe->pipe_idx; + } else { + idle_pipe = find_idle_secondary_pipe_check_mpo(res_ctx, pool, head_pipe); + if (!idle_pipe) + return NULL; + } + + idle_pipe->stream = head_pipe->stream; + idle_pipe->stream_res.tg = head_pipe->stream_res.tg; + idle_pipe->stream_res.opp = head_pipe->stream_res.opp; + + idle_pipe->plane_res.hubp = pool->hubps[idle_pipe->pipe_idx]; + idle_pipe->plane_res.ipp = pool->ipps[idle_pipe->pipe_idx]; + idle_pipe->plane_res.dpp = pool->dpps[idle_pipe->pipe_idx]; + idle_pipe->plane_res.mpcc_inst = pool->dpps[idle_pipe->pipe_idx]->inst; + + return idle_pipe; +} |