diff options
Diffstat (limited to 'drivers/gpu/drm/amd/display/dc/dml2')
28 files changed, 3148 insertions, 2950 deletions
diff --git a/drivers/gpu/drm/amd/display/dc/dml2/Makefile b/drivers/gpu/drm/amd/display/dc/dml2/Makefile index 21fd466dba26e..4c21ce42054c5 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/Makefile +++ b/drivers/gpu/drm/amd/display/dc/dml2/Makefile @@ -81,10 +81,11 @@ AMD_DAL_DML2 = $(addprefix $(AMDDALPATH)/dc/dml2/,$(DML2)) AMD_DISPLAY_FILES += $(AMD_DAL_DML2) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.o := $(dml2_ccflags) $(frame_warn_flag) +CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_utils.o := $(dml2_ccflags) $(frame_warn_flag) +CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_factory.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.o := $(dml2_ccflags) @@ -94,18 +95,16 @@ CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.o := $(dml2_ccflag CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.o := $(dml2_ccflags) - - CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml21_wrapper.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/dml21_translation_helper.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/dml21_utils.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/inc/dml2_debug.o := $(dml2_ccflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.o := $(dml2_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.o := $(dml2_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_factory.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_utils.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.o := $(dml2_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.o := $(dml2_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.o := $(dml2_rcflags) @@ -117,12 +116,11 @@ CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_standalone_libraries/lib_floa CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml21_wrapper.o := $(dml2_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/dml21_translation_helper.o := $(dml2_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/dml21_utils.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/inc/dml2_debug.o := $(dml2_rcflags) DML21 := src/dml2_top/dml2_top_interfaces.o DML21 += src/dml2_top/dml2_top_soc15.o -DML21 += src/inc/dml2_debug.o DML21 += src/dml2_core/dml2_core_dcn4.o +DML21 += src/dml2_core/dml2_core_utils.o DML21 += src/dml2_core/dml2_core_factory.o DML21 += src/dml2_core/dml2_core_dcn4_calcs.o DML21 += src/dml2_dpmm/dml2_dpmm_dcn4.o diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c index 7ae9c0ba0c9e6..715f9019a33e2 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c @@ -10189,7 +10189,7 @@ dml_uint_t dml_mode_support_ex(struct dml_mode_support_ex_params_st *in_out_para result = mode_support_pwr_states(&in_out_params->out_lowest_state_idx, in_out_params->mode_lib, in_out_params->in_display_cfg, - 0, + in_out_params->in_start_state_idx, in_out_params->mode_lib->states.num_states - 1); if (result) diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h index 0670e4dc4fd91..dbeb084660922 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h @@ -1917,6 +1917,7 @@ struct display_mode_lib_st { struct dml_mode_support_ex_params_st { struct display_mode_lib_st *mode_lib; const struct dml_display_cfg_st *in_display_cfg; + dml_uint_t in_start_state_idx; dml_uint_t out_lowest_state_idx; struct dml_mode_support_info_st *out_evaluation_info; }; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c index 731fbd4bc600b..a06217a9eef60 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c @@ -2,7 +2,6 @@ // // Copyright 2024 Advanced Micro Devices, Inc. - #include "dml21_wrapper.h" #include "dml2_core_dcn4_calcs.h" #include "dml2_internal_shared_types.h" @@ -11,277 +10,263 @@ #include "dml21_translation_helper.h" #include "bounding_boxes/dcn4_soc_bb.h" -static void dml21_init_socbb_params(struct dml2_initialize_instance_in_out *dml_init, - const struct dml2_configuration_options *config, - const struct dc *in_dc) -{ - const struct dml2_soc_bb *soc_bb; - const struct dml2_soc_qos_parameters *qos_params; - - switch (in_dc->ctx->dce_version) { - case DCN_VERSION_4_01: - default: - if (config->bb_from_dmub) - soc_bb = config->bb_from_dmub; - else - soc_bb = &dml2_socbb_dcn401; - - qos_params = &dml_dcn4_variant_a_soc_qos_params; - } - - /* patch soc bb */ - memcpy(&dml_init->soc_bb, soc_bb, sizeof(struct dml2_soc_bb)); - - /* patch qos params */ - memcpy(&dml_init->soc_bb.qos_parameters, qos_params, sizeof(struct dml2_soc_qos_parameters)); -} - -static void dml21_external_socbb_params(struct dml2_initialize_instance_in_out *dml_init, - const struct dml2_configuration_options *config) -{ - memcpy(&dml_init->soc_bb, &config->external_socbb_ip_params->soc_bb, sizeof(struct dml2_soc_bb)); -} - -static void dml21_external_ip_params(struct dml2_initialize_instance_in_out *dml_init, +static void dml21_populate_pmo_options(struct dml2_pmo_options *pmo_options, + const struct dc *in_dc, const struct dml2_configuration_options *config) { - memcpy(&dml_init->ip_caps, &config->external_socbb_ip_params->ip_params, sizeof(struct dml2_ip_capabilities)); + bool disable_fams2 = !in_dc->debug.fams2_config.bits.enable; + + /* ODM options */ + pmo_options->disable_dyn_odm = !config->minimize_dispclk_using_odm; + pmo_options->disable_dyn_odm_for_multi_stream = true; + pmo_options->disable_dyn_odm_for_stream_with_svp = true; + + pmo_options->disable_vblank = ((in_dc->debug.dml21_disable_pstate_method_mask >> 1) & 1); + + /* NOTE: DRR and SubVP Require FAMS2 */ + pmo_options->disable_svp = ((in_dc->debug.dml21_disable_pstate_method_mask >> 2) & 1) || + in_dc->debug.force_disable_subvp || + disable_fams2; + pmo_options->disable_drr_clamped = ((in_dc->debug.dml21_disable_pstate_method_mask >> 3) & 1) || + disable_fams2; + pmo_options->disable_drr_var = ((in_dc->debug.dml21_disable_pstate_method_mask >> 4) & 1) || + disable_fams2; + pmo_options->disable_fams2 = disable_fams2; + + pmo_options->disable_drr_var_when_var_active = in_dc->debug.disable_fams_gaming == INGAME_FAMS_DISABLE || + in_dc->debug.disable_fams_gaming == INGAME_FAMS_MULTI_DISP_CLAMPED_ONLY; + pmo_options->disable_drr_clamped_when_var_active = in_dc->debug.disable_fams_gaming == INGAME_FAMS_DISABLE; } -static void dml21_init_ip_params(struct dml2_initialize_instance_in_out *dml_init, +/* + * Populate dml_init based on default static values in soc bb. The default + * values are for reference and support at least minimal operation of current + * SoC and DCN hardware. The values could be modifed by subsequent override + * functions to reflect our true hardware capability. + */ +static void populate_default_dml_init_params(struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc) { - const struct dml2_ip_capabilities *ip_caps; - switch (in_dc->ctx->dce_version) { case DCN_VERSION_4_01: + dml_init->options.project_id = dml2_project_dcn4x_stage2_auto_drr_svp; + dml21_populate_pmo_options(&dml_init->options.pmo_options, in_dc, config); + dml_init->soc_bb = dml2_socbb_dcn401; + dml_init->soc_bb.qos_parameters = dml_dcn4_variant_a_soc_qos_params; + dml_init->ip_caps = dml2_dcn401_max_ip_caps; + break; default: - ip_caps = &dml2_dcn401_max_ip_caps; + memset(dml_init, 0, sizeof(*dml_init)); + DC_ERR("unsupported dcn version for DML21!"); + return; } - - memcpy(&dml_init->ip_caps, ip_caps, sizeof(struct dml2_ip_capabilities)); } -void dml21_initialize_soc_bb_params(struct dml2_initialize_instance_in_out *dml_init, +static void override_dml_init_with_values_from_hardware_default(struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc) { - if (config->use_native_soc_bb_construction) - dml21_init_socbb_params(dml_init, config, in_dc); - else - dml21_external_socbb_params(dml_init, config); + dml_init->soc_bb.dchub_refclk_mhz = in_dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000; + dml_init->soc_bb.dprefclk_mhz = in_dc->clk_mgr->dprefclk_khz / 1000; + dml_init->soc_bb.dispclk_dppclk_vco_speed_mhz = in_dc->clk_mgr->dentist_vco_freq_khz / 1000.0; } -void dml21_initialize_ip_params(struct dml2_initialize_instance_in_out *dml_init, +/* + * SMU stands for System Management Unit. It is a power management processor. + * It owns the initialization of dc's clock table and programming of clock values + * based on dc's requests. + * Our clock values in base soc bb is a dummy placeholder. The real clock values + * are retrieved from SMU firmware to dc clock table at runtime. + * This function overrides our dummy placeholder values with real values in dc + * clock table. + */ +static void override_dml_init_with_values_from_smu( + struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc) { - if (config->use_native_soc_bb_construction) - dml21_init_ip_params(dml_init, config, in_dc); - else - dml21_external_ip_params(dml_init, config); -} - -void dml21_apply_soc_bb_overrides(struct dml2_initialize_instance_in_out *dml_init, - const struct dml2_configuration_options *config, const struct dc *in_dc) -{ int i; - const struct clk_bw_params *dc_bw_params = in_dc->clk_mgr->bw_params; const struct clk_limit_table *dc_clk_table = &dc_bw_params->clk_table; - struct dml2_soc_bb *dml_soc_bb = &dml_init->soc_bb; - struct dml2_soc_state_table *dml_clk_table = &dml_soc_bb->clk_table; - - /* override clocks if smu is present */ - if (in_dc->clk_mgr->funcs->is_smu_present && in_dc->clk_mgr->funcs->is_smu_present(in_dc->clk_mgr)) { - /* dcfclk */ - if (dc_clk_table->num_entries_per_clk.num_dcfclk_levels) { - dml_clk_table->dcfclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dcfclk_levels; - for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { - if (i < dml_clk_table->dcfclk.num_clk_values) { - if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dcfclk_mhz && - dc_clk_table->entries[i].dcfclk_mhz > dc_bw_params->dc_mode_limit.dcfclk_mhz) { - if (i == 0 || dc_clk_table->entries[i-1].dcfclk_mhz < dc_bw_params->dc_mode_limit.dcfclk_mhz) { - dml_clk_table->dcfclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dcfclk_mhz * 1000; - dml_clk_table->dcfclk.num_clk_values = i + 1; - } else { - dml_clk_table->dcfclk.clk_values_khz[i] = 0; - dml_clk_table->dcfclk.num_clk_values = i; - } + struct dml2_soc_state_table *dml_clk_table = &dml_init->soc_bb.clk_table; + + if (!in_dc->clk_mgr->funcs->is_smu_present || + !in_dc->clk_mgr->funcs->is_smu_present(in_dc->clk_mgr)) + /* skip if smu is not present */ + return; + + /* dcfclk */ + if (dc_clk_table->num_entries_per_clk.num_dcfclk_levels) { + dml_clk_table->dcfclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dcfclk_levels; + for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { + if (i < dml_clk_table->dcfclk.num_clk_values) { + if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dcfclk_mhz && + dc_clk_table->entries[i].dcfclk_mhz > dc_bw_params->dc_mode_limit.dcfclk_mhz) { + if (i == 0 || dc_clk_table->entries[i-1].dcfclk_mhz < dc_bw_params->dc_mode_limit.dcfclk_mhz) { + dml_clk_table->dcfclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dcfclk_mhz * 1000; + dml_clk_table->dcfclk.num_clk_values = i + 1; } else { - dml_clk_table->dcfclk.clk_values_khz[i] = dc_clk_table->entries[i].dcfclk_mhz * 1000; + dml_clk_table->dcfclk.clk_values_khz[i] = 0; + dml_clk_table->dcfclk.num_clk_values = i; } } else { - dml_clk_table->dcfclk.clk_values_khz[i] = 0; + dml_clk_table->dcfclk.clk_values_khz[i] = dc_clk_table->entries[i].dcfclk_mhz * 1000; } + } else { + dml_clk_table->dcfclk.clk_values_khz[i] = 0; } } + } - /* fclk */ - if (dc_clk_table->num_entries_per_clk.num_fclk_levels) { - dml_clk_table->fclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_fclk_levels; - for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { - if (i < dml_clk_table->fclk.num_clk_values) { - if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.fclk_mhz && - dc_clk_table->entries[i].fclk_mhz > dc_bw_params->dc_mode_limit.fclk_mhz) { - if (i == 0 || dc_clk_table->entries[i-1].fclk_mhz < dc_bw_params->dc_mode_limit.fclk_mhz) { - dml_clk_table->fclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.fclk_mhz * 1000; - dml_clk_table->fclk.num_clk_values = i + 1; - } else { - dml_clk_table->fclk.clk_values_khz[i] = 0; - dml_clk_table->fclk.num_clk_values = i; - } + /* fclk */ + if (dc_clk_table->num_entries_per_clk.num_fclk_levels) { + dml_clk_table->fclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_fclk_levels; + for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { + if (i < dml_clk_table->fclk.num_clk_values) { + if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.fclk_mhz && + dc_clk_table->entries[i].fclk_mhz > dc_bw_params->dc_mode_limit.fclk_mhz) { + if (i == 0 || dc_clk_table->entries[i-1].fclk_mhz < dc_bw_params->dc_mode_limit.fclk_mhz) { + dml_clk_table->fclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.fclk_mhz * 1000; + dml_clk_table->fclk.num_clk_values = i + 1; } else { - dml_clk_table->fclk.clk_values_khz[i] = dc_clk_table->entries[i].fclk_mhz * 1000; + dml_clk_table->fclk.clk_values_khz[i] = 0; + dml_clk_table->fclk.num_clk_values = i; } } else { - dml_clk_table->fclk.clk_values_khz[i] = 0; + dml_clk_table->fclk.clk_values_khz[i] = dc_clk_table->entries[i].fclk_mhz * 1000; } + } else { + dml_clk_table->fclk.clk_values_khz[i] = 0; } } + } - /* uclk */ - if (dc_clk_table->num_entries_per_clk.num_memclk_levels) { - dml_clk_table->uclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_memclk_levels; - for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { - if (i < dml_clk_table->uclk.num_clk_values) { - if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.memclk_mhz && - dc_clk_table->entries[i].memclk_mhz > dc_bw_params->dc_mode_limit.memclk_mhz) { - if (i == 0 || dc_clk_table->entries[i-1].memclk_mhz < dc_bw_params->dc_mode_limit.memclk_mhz) { - dml_clk_table->uclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.memclk_mhz * 1000; - dml_clk_table->uclk.num_clk_values = i + 1; - } else { - dml_clk_table->uclk.clk_values_khz[i] = 0; - dml_clk_table->uclk.num_clk_values = i; - } + /* uclk */ + if (dc_clk_table->num_entries_per_clk.num_memclk_levels) { + dml_clk_table->uclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_memclk_levels; + for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { + if (i < dml_clk_table->uclk.num_clk_values) { + if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.memclk_mhz && + dc_clk_table->entries[i].memclk_mhz > dc_bw_params->dc_mode_limit.memclk_mhz) { + if (i == 0 || dc_clk_table->entries[i-1].memclk_mhz < dc_bw_params->dc_mode_limit.memclk_mhz) { + dml_clk_table->uclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.memclk_mhz * 1000; + dml_clk_table->uclk.num_clk_values = i + 1; } else { - dml_clk_table->uclk.clk_values_khz[i] = dc_clk_table->entries[i].memclk_mhz * 1000; + dml_clk_table->uclk.clk_values_khz[i] = 0; + dml_clk_table->uclk.num_clk_values = i; } } else { - dml_clk_table->uclk.clk_values_khz[i] = 0; + dml_clk_table->uclk.clk_values_khz[i] = dc_clk_table->entries[i].memclk_mhz * 1000; } + } else { + dml_clk_table->uclk.clk_values_khz[i] = 0; } } + } - /* dispclk */ - if (dc_clk_table->num_entries_per_clk.num_dispclk_levels) { - dml_clk_table->dispclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dispclk_levels; - for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { - if (i < dml_clk_table->dispclk.num_clk_values) { - if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dispclk_mhz && - dc_clk_table->entries[i].dispclk_mhz > dc_bw_params->dc_mode_limit.dispclk_mhz) { - if (i == 0 || dc_clk_table->entries[i-1].dispclk_mhz < dc_bw_params->dc_mode_limit.dispclk_mhz) { - dml_clk_table->dispclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dispclk_mhz * 1000; - dml_clk_table->dispclk.num_clk_values = i + 1; - } else { - dml_clk_table->dispclk.clk_values_khz[i] = 0; - dml_clk_table->dispclk.num_clk_values = i; - } + /* dispclk */ + if (dc_clk_table->num_entries_per_clk.num_dispclk_levels) { + dml_clk_table->dispclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dispclk_levels; + for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { + if (i < dml_clk_table->dispclk.num_clk_values) { + if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dispclk_mhz && + dc_clk_table->entries[i].dispclk_mhz > dc_bw_params->dc_mode_limit.dispclk_mhz) { + if (i == 0 || dc_clk_table->entries[i-1].dispclk_mhz < dc_bw_params->dc_mode_limit.dispclk_mhz) { + dml_clk_table->dispclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dispclk_mhz * 1000; + dml_clk_table->dispclk.num_clk_values = i + 1; } else { - dml_clk_table->dispclk.clk_values_khz[i] = dc_clk_table->entries[i].dispclk_mhz * 1000; + dml_clk_table->dispclk.clk_values_khz[i] = 0; + dml_clk_table->dispclk.num_clk_values = i; } } else { - dml_clk_table->dispclk.clk_values_khz[i] = 0; + dml_clk_table->dispclk.clk_values_khz[i] = dc_clk_table->entries[i].dispclk_mhz * 1000; } + } else { + dml_clk_table->dispclk.clk_values_khz[i] = 0; } } + } - /* dppclk */ - if (dc_clk_table->num_entries_per_clk.num_dppclk_levels) { - dml_clk_table->dppclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dppclk_levels; - for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { - if (i < dml_clk_table->dppclk.num_clk_values) { - if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dppclk_mhz && - dc_clk_table->entries[i].dppclk_mhz > dc_bw_params->dc_mode_limit.dppclk_mhz) { - if (i == 0 || dc_clk_table->entries[i-1].dppclk_mhz < dc_bw_params->dc_mode_limit.dppclk_mhz) { - dml_clk_table->dppclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dppclk_mhz * 1000; - dml_clk_table->dppclk.num_clk_values = i + 1; - } else { - dml_clk_table->dppclk.clk_values_khz[i] = 0; - dml_clk_table->dppclk.num_clk_values = i; - } + /* dppclk */ + if (dc_clk_table->num_entries_per_clk.num_dppclk_levels) { + dml_clk_table->dppclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dppclk_levels; + for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { + if (i < dml_clk_table->dppclk.num_clk_values) { + if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dppclk_mhz && + dc_clk_table->entries[i].dppclk_mhz > dc_bw_params->dc_mode_limit.dppclk_mhz) { + if (i == 0 || dc_clk_table->entries[i-1].dppclk_mhz < dc_bw_params->dc_mode_limit.dppclk_mhz) { + dml_clk_table->dppclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dppclk_mhz * 1000; + dml_clk_table->dppclk.num_clk_values = i + 1; } else { - dml_clk_table->dppclk.clk_values_khz[i] = dc_clk_table->entries[i].dppclk_mhz * 1000; + dml_clk_table->dppclk.clk_values_khz[i] = 0; + dml_clk_table->dppclk.num_clk_values = i; } } else { - dml_clk_table->dppclk.clk_values_khz[i] = 0; + dml_clk_table->dppclk.clk_values_khz[i] = dc_clk_table->entries[i].dppclk_mhz * 1000; } + } else { + dml_clk_table->dppclk.clk_values_khz[i] = 0; } } + } - /* dtbclk */ - if (dc_clk_table->num_entries_per_clk.num_dtbclk_levels) { - dml_clk_table->dtbclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dtbclk_levels; - for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { - if (i < dml_clk_table->dtbclk.num_clk_values) { - if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dtbclk_mhz && - dc_clk_table->entries[i].dtbclk_mhz > dc_bw_params->dc_mode_limit.dtbclk_mhz) { - if (i == 0 || dc_clk_table->entries[i-1].dtbclk_mhz < dc_bw_params->dc_mode_limit.dtbclk_mhz) { - dml_clk_table->dtbclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dtbclk_mhz * 1000; - dml_clk_table->dtbclk.num_clk_values = i + 1; - } else { - dml_clk_table->dtbclk.clk_values_khz[i] = 0; - dml_clk_table->dtbclk.num_clk_values = i; - } + /* dtbclk */ + if (dc_clk_table->num_entries_per_clk.num_dtbclk_levels) { + dml_clk_table->dtbclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dtbclk_levels; + for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { + if (i < dml_clk_table->dtbclk.num_clk_values) { + if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dtbclk_mhz && + dc_clk_table->entries[i].dtbclk_mhz > dc_bw_params->dc_mode_limit.dtbclk_mhz) { + if (i == 0 || dc_clk_table->entries[i-1].dtbclk_mhz < dc_bw_params->dc_mode_limit.dtbclk_mhz) { + dml_clk_table->dtbclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dtbclk_mhz * 1000; + dml_clk_table->dtbclk.num_clk_values = i + 1; } else { - dml_clk_table->dtbclk.clk_values_khz[i] = dc_clk_table->entries[i].dtbclk_mhz * 1000; + dml_clk_table->dtbclk.clk_values_khz[i] = 0; + dml_clk_table->dtbclk.num_clk_values = i; } } else { - dml_clk_table->dtbclk.clk_values_khz[i] = 0; + dml_clk_table->dtbclk.clk_values_khz[i] = dc_clk_table->entries[i].dtbclk_mhz * 1000; } + } else { + dml_clk_table->dtbclk.clk_values_khz[i] = 0; } } + } - /* socclk */ - if (dc_clk_table->num_entries_per_clk.num_socclk_levels) { - dml_clk_table->socclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_socclk_levels; - for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { - if (i < dml_clk_table->socclk.num_clk_values) { - if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.socclk_mhz && - dc_clk_table->entries[i].socclk_mhz > dc_bw_params->dc_mode_limit.socclk_mhz) { - if (i == 0 || dc_clk_table->entries[i-1].socclk_mhz < dc_bw_params->dc_mode_limit.socclk_mhz) { - dml_clk_table->socclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.socclk_mhz * 1000; - dml_clk_table->socclk.num_clk_values = i + 1; - } else { - dml_clk_table->socclk.clk_values_khz[i] = 0; - dml_clk_table->socclk.num_clk_values = i; - } + /* socclk */ + if (dc_clk_table->num_entries_per_clk.num_socclk_levels) { + dml_clk_table->socclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_socclk_levels; + for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { + if (i < dml_clk_table->socclk.num_clk_values) { + if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.socclk_mhz && + dc_clk_table->entries[i].socclk_mhz > dc_bw_params->dc_mode_limit.socclk_mhz) { + if (i == 0 || dc_clk_table->entries[i-1].socclk_mhz < dc_bw_params->dc_mode_limit.socclk_mhz) { + dml_clk_table->socclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.socclk_mhz * 1000; + dml_clk_table->socclk.num_clk_values = i + 1; } else { - dml_clk_table->socclk.clk_values_khz[i] = dc_clk_table->entries[i].socclk_mhz * 1000; + dml_clk_table->socclk.clk_values_khz[i] = 0; + dml_clk_table->socclk.num_clk_values = i; } } else { - dml_clk_table->socclk.clk_values_khz[i] = 0; + dml_clk_table->socclk.clk_values_khz[i] = dc_clk_table->entries[i].socclk_mhz * 1000; } + } else { + dml_clk_table->socclk.clk_values_khz[i] = 0; } } - - /* do not override phyclks for now */ - /* phyclk */ - // dml_clk_table->phyclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_phyclk_levels; - // for (i = 0; i < DML_MAX_CLK_TABLE_SIZE; i++) { - // dml_clk_table->phyclk.clk_values_khz[i] = dc_clk_table->entries[i].phyclk_mhz * 1000; - // } - - /* phyclk_d18 */ - // dml_clk_table->phyclk_d18.num_clk_values = dc_clk_table->num_entries_per_clk.num_phyclk_d18_levels; - // for (i = 0; i < DML_MAX_CLK_TABLE_SIZE; i++) { - // dml_clk_table->phyclk_d18.clk_values_khz[i] = dc_clk_table->entries[i].phyclk_d18_mhz * 1000; - // } - - /* phyclk_d32 */ - // dml_clk_table->phyclk_d32.num_clk_values = dc_clk_table->num_entries_per_clk.num_phyclk_d32_levels; - // for (i = 0; i < DML_MAX_CLK_TABLE_SIZE; i++) { - // dml_clk_table->phyclk_d32.clk_values_khz[i] = dc_clk_table->entries[i].phyclk_d32_mhz * 1000; - // } } +} - dml_soc_bb->dchub_refclk_mhz = in_dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000; - dml_soc_bb->dprefclk_mhz = in_dc->clk_mgr->dprefclk_khz / 1000; - dml_soc_bb->xtalclk_mhz = in_dc->ctx->dc_bios->fw_info.pll_info.crystal_frequency / 1000; - dml_soc_bb->dispclk_dppclk_vco_speed_mhz = in_dc->clk_mgr->dentist_vco_freq_khz / 1000.0; +static void override_dml_init_with_values_from_vbios( + struct dml2_initialize_instance_in_out *dml_init, + const struct dml2_configuration_options *config, + const struct dc *in_dc) +{ + const struct clk_bw_params *dc_bw_params = in_dc->clk_mgr->bw_params; + struct dml2_soc_bb *dml_soc_bb = &dml_init->soc_bb; + struct dml2_soc_state_table *dml_clk_table = &dml_init->soc_bb.clk_table; - /* override bounding box paramters from VBIOS */ if (in_dc->ctx->dc_bios->bb_info.dram_clock_change_latency_100ns > 0) dml_soc_bb->power_management_parameters.dram_clk_change_blackout_us = (in_dc->ctx->dc_bios->bb_info.dram_clock_change_latency_100ns + 9) / 10; @@ -308,32 +293,120 @@ void dml21_apply_soc_bb_overrides(struct dml2_initialize_instance_in_out *dml_in dml_clk_table->dram_config.channel_width_bytes = in_dc->ctx->dc_bios->vram_info.dram_channel_width_bytes; } - /* override bounding box paramters from DC config */ - if (in_dc->bb_overrides.sr_exit_time_ns) { - dml_soc_bb->power_management_parameters.stutter_exit_latency_us = - in_dc->bb_overrides.sr_exit_time_ns / 1000.0; + dml_init->soc_bb.xtalclk_mhz = in_dc->ctx->dc_bios->fw_info.pll_info.crystal_frequency / 1000; +} + + +static void override_dml_init_with_values_from_dmub(struct dml2_initialize_instance_in_out *dml_init, + const struct dml2_configuration_options *config, + const struct dc *in_dc) +{ + /* + * TODO - There seems to be overlaps between the values overriden from + * dmub and vbios. Investigate and identify the values that DMUB needs + * to own. + */ +// const struct dmub_soc_bb_params *dmub_bb_params = +// (const struct dmub_soc_bb_params *)config->bb_from_dmub; + +// if (dmub_bb_params == NULL) +// return; + +// if (dmub_bb_params->dram_clk_change_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.dram_clk_change_blackout_us = +// (double) dmub_bb_params->dram_clk_change_blackout_ns / 1000.0; +// if (dmub_bb_params->dram_clk_change_read_only_ns > 0) +// dml_init->soc_bb.power_management_parameters.dram_clk_change_read_only_us = +// (double) dmub_bb_params->dram_clk_change_read_only_ns / 1000.0; +// if (dmub_bb_params->dram_clk_change_write_only_ns > 0) +// dml_init->soc_bb.power_management_parameters.dram_clk_change_write_only_us = +// (double) dmub_bb_params->dram_clk_change_write_only_ns / 1000.0; +// if (dmub_bb_params->fclk_change_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.fclk_change_blackout_us = +// (double) dmub_bb_params->fclk_change_blackout_ns / 1000.0; +// if (dmub_bb_params->g7_ppt_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.g7_ppt_blackout_us = +// (double) dmub_bb_params->g7_ppt_blackout_ns / 1000.0; +// if (dmub_bb_params->stutter_enter_plus_exit_latency_ns > 0) +// dml_init->soc_bb.power_management_parameters.stutter_enter_plus_exit_latency_us = +// (double) dmub_bb_params->stutter_enter_plus_exit_latency_ns / 1000.0; +// if (dmub_bb_params->stutter_exit_latency_ns > 0) +// dml_init->soc_bb.power_management_parameters.stutter_exit_latency_us = +// (double) dmub_bb_params->stutter_exit_latency_ns / 1000.0; +// if (dmub_bb_params->z8_stutter_enter_plus_exit_latency_ns > 0) +// dml_init->soc_bb.power_management_parameters.z8_stutter_enter_plus_exit_latency_us = +// (double) dmub_bb_params->z8_stutter_enter_plus_exit_latency_ns / 1000.0; +// if (dmub_bb_params->z8_stutter_exit_latency_ns > 0) +// dml_init->soc_bb.power_management_parameters.z8_stutter_exit_latency_us = +// (double) dmub_bb_params->z8_stutter_exit_latency_ns / 1000.0; +// if (dmub_bb_params->z8_min_idle_time_ns > 0) +// dml_init->soc_bb.power_management_parameters.z8_min_idle_time = +// (double) dmub_bb_params->z8_min_idle_time_ns / 1000.0; +// #ifndef TRIM_DML2_DCN6B_IP_SENSITIVE +// if (dmub_bb_params->type_b_dram_clk_change_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.lpddr5_dram_clk_change_blackout_us = +// (double) dmub_bb_params->type_b_dram_clk_change_blackout_ns / 1000.0; +// if (dmub_bb_params->type_b_ppt_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.lpddr5_ppt_blackout_us = +// (double) dmub_bb_params->type_b_ppt_blackout_ns / 1000.0; +// #else +// if (dmub_bb_params->type_b_dram_clk_change_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.type_b_dram_clk_change_blackout_us = +// (double) dmub_bb_params->type_b_dram_clk_change_blackout_ns / 1000.0; +// if (dmub_bb_params->type_b_ppt_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.type_b_ppt_blackout_us = +// (double) dmub_bb_params->type_b_ppt_blackout_ns / 1000.0; +// #endif +// if (dmub_bb_params->vmin_limit_dispclk_khz > 0) +// dml_init->soc_bb.vmin_limit.dispclk_khz = dmub_bb_params->vmin_limit_dispclk_khz; +// if (dmub_bb_params->vmin_limit_dcfclk_khz > 0) +// dml_init->soc_bb.vmin_limit.dcfclk_khz = dmub_bb_params->vmin_limit_dcfclk_khz; +// if (dmub_bb_params->g7_temperature_read_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.g7_temperature_read_blackout_us = +// (double) dmub_bb_params->g7_temperature_read_blackout_ns / 1000.0; +} + +static void override_dml_init_with_values_from_software_policy(struct dml2_initialize_instance_in_out *dml_init, + const struct dml2_configuration_options *config, + const struct dc *in_dc) +{ + if (!config->use_native_soc_bb_construction) { + dml_init->soc_bb = config->external_socbb_ip_params->soc_bb; + dml_init->ip_caps = config->external_socbb_ip_params->ip_params; } - if (in_dc->bb_overrides.sr_enter_plus_exit_time_ns) { - dml_soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us = + if (in_dc->bb_overrides.sr_exit_time_ns) + dml_init->soc_bb.power_management_parameters.stutter_exit_latency_us = + in_dc->bb_overrides.sr_exit_time_ns / 1000.0; + + if (in_dc->bb_overrides.sr_enter_plus_exit_time_ns) + dml_init->soc_bb.power_management_parameters.stutter_enter_plus_exit_latency_us = in_dc->bb_overrides.sr_enter_plus_exit_time_ns / 1000.0; - } - if (in_dc->bb_overrides.dram_clock_change_latency_ns) { - dml_soc_bb->power_management_parameters.dram_clk_change_blackout_us = + if (in_dc->bb_overrides.dram_clock_change_latency_ns) + dml_init->soc_bb.power_management_parameters.dram_clk_change_blackout_us = in_dc->bb_overrides.dram_clock_change_latency_ns / 1000.0; - } - if (in_dc->bb_overrides.fclk_clock_change_latency_ns) { - dml_soc_bb->power_management_parameters.fclk_change_blackout_us = + if (in_dc->bb_overrides.fclk_clock_change_latency_ns) + dml_init->soc_bb.power_management_parameters.fclk_change_blackout_us = in_dc->bb_overrides.fclk_clock_change_latency_ns / 1000.0; - } +} - //TODO - // if (in_dc->bb_overrides.dummy_clock_change_latency_ns) { - // dml_soc_bb->power_management_parameters.dram_clk_change_blackout_us = - // in_dc->bb_overrides.dram_clock_change_latency_ns / 1000.0; - // } +void dml21_populate_dml_init_params(struct dml2_initialize_instance_in_out *dml_init, + const struct dml2_configuration_options *config, + const struct dc *in_dc) +{ + populate_default_dml_init_params(dml_init, config, in_dc); + + override_dml_init_with_values_from_hardware_default(dml_init, config, in_dc); + + override_dml_init_with_values_from_smu(dml_init, config, in_dc); + + override_dml_init_with_values_from_vbios(dml_init, config, in_dc); + + override_dml_init_with_values_from_dmub(dml_init, config, in_dc); + + override_dml_init_with_values_from_software_policy(dml_init, config, in_dc); } static unsigned int calc_max_hardware_v_total(const struct dc_stream_state *stream) @@ -526,7 +599,8 @@ static void populate_dml21_output_config_from_stream_state(struct dml2_link_outp static void populate_dml21_stream_overrides_from_stream_state( struct dml2_stream_parameters *stream_desc, - struct dc_stream_state *stream) + struct dc_stream_state *stream, + struct dc_stream_status *stream_status) { switch (stream->debug.force_odm_combine_segments) { case 0: @@ -551,7 +625,9 @@ static void populate_dml21_stream_overrides_from_stream_state( if (!stream->ctx->dc->debug.enable_single_display_2to1_odm_policy || stream->debug.force_odm_combine_segments > 0) stream_desc->overrides.disable_dynamic_odm = true; - stream_desc->overrides.disable_subvp = stream->ctx->dc->debug.force_disable_subvp || stream->hw_cursor_req; + stream_desc->overrides.disable_subvp = stream->ctx->dc->debug.force_disable_subvp || + stream->hw_cursor_req || + stream_status->mall_stream_config.cursor_size_limit_subvp; } static enum dml2_swizzle_mode gfx_addr3_to_dml2_swizzle_mode(enum swizzle_mode_addr3_values addr3_mode) @@ -723,7 +799,6 @@ static void populate_dml21_surface_config_from_plane_state( switch (plane_state->tiling_info.gfxversion) { case DcGfxVersion7: case DcGfxVersion8: - // Placeholder for programming the array_mode break; case DcGfxVersion9: case DcGfxVersion10: @@ -785,6 +860,7 @@ static void populate_dml21_plane_config_from_plane_state(struct dml2_context *dm plane->pixel_format = dml2_420_10; break; case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616: + case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616: case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F: case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F: plane->pixel_format = dml2_444_64; @@ -886,6 +962,7 @@ static void populate_dml21_plane_config_from_plane_state(struct dml2_context *dm plane->tdlut.tdlut_width_mode = dml2_tdlut_width_17_cube; break; case DC_CM2_GPU_MEM_SIZE_TRANSFORMED: + default: //plane->tdlut.tdlut_width_mode = dml2_tdlut_width_flatten; // dml2_tdlut_width_flatten undefined break; } @@ -946,7 +1023,7 @@ static unsigned int map_stream_to_dml21_display_cfg(const struct dml2_context *d return location; } -static unsigned int map_plane_to_dml21_display_cfg(const struct dml2_context *dml_ctx, unsigned int stream_id, +unsigned int map_plane_to_dml21_display_cfg(const struct dml2_context *dml_ctx, unsigned int stream_id, const struct dc_plane_state *plane, const struct dc_state *context) { unsigned int plane_id; @@ -1023,7 +1100,7 @@ bool dml21_map_dc_state_into_dml_display_cfg(const struct dc *in_dc, struct dc_s populate_dml21_timing_config_from_stream_state(&dml_dispcfg->stream_descriptors[disp_cfg_stream_location].timing, context->streams[stream_index], dml_ctx); adjust_dml21_hblank_timing_config_from_pipe_ctx(&dml_dispcfg->stream_descriptors[disp_cfg_stream_location].timing, &context->res_ctx.pipe_ctx[stream_index]); populate_dml21_output_config_from_stream_state(&dml_dispcfg->stream_descriptors[disp_cfg_stream_location].output, context->streams[stream_index], &context->res_ctx.pipe_ctx[stream_index]); - populate_dml21_stream_overrides_from_stream_state(&dml_dispcfg->stream_descriptors[disp_cfg_stream_location], context->streams[stream_index]); + populate_dml21_stream_overrides_from_stream_state(&dml_dispcfg->stream_descriptors[disp_cfg_stream_location], context->streams[stream_index], &context->stream_status[stream_index]); dml_dispcfg->stream_descriptors[disp_cfg_stream_location].overrides.hw.twait_budgeting.fclk_pstate = dml2_twait_budgeting_setting_if_needed; dml_dispcfg->stream_descriptors[disp_cfg_stream_location].overrides.hw.twait_budgeting.uclk_pstate = dml2_twait_budgeting_setting_if_needed; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.h index 069b939c672a9..9880d3e0398e2 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.h @@ -11,18 +11,18 @@ struct dc_state; struct dcn_watermarks; union dcn_watermark_set; struct pipe_ctx; +struct dc_plane_state; struct dml2_context; struct dml2_configuration_options; struct dml2_initialize_instance_in_out; -void dml21_apply_soc_bb_overrides(struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc); -void dml21_initialize_soc_bb_params(struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc); -void dml21_initialize_ip_params(struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc); +void dml21_populate_dml_init_params(struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc); bool dml21_map_dc_state_into_dml_display_cfg(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx); void dml21_copy_clocks_to_dc_state(struct dml2_context *in_ctx, struct dc_state *context); void dml21_extract_watermark_sets(const struct dc *in_dc, union dcn_watermark_set *watermarks, struct dml2_context *in_ctx); void dml21_map_hw_resources(struct dml2_context *dml_ctx); void dml21_get_pipe_mcache_config(struct dc_state *context, struct pipe_ctx *pipe_ctx, struct dml2_per_plane_programming *pln_prog, struct dml2_pipe_configuration_descriptor *mcache_pipe_config); void dml21_set_dc_p_state_type(struct pipe_ctx *pipe_ctx, struct dml2_per_stream_programming *stream_programming, bool sub_vp_enabled); +unsigned int map_plane_to_dml21_display_cfg(const struct dml2_context *dml_ctx, unsigned int stream_id, const struct dc_plane_state *plane, const struct dc_state *context); #endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c index 930e86cdb88a2..ee721606b8839 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c @@ -384,6 +384,7 @@ void dml21_build_fams2_programming(const struct dc *dc, /* reset fams2 data */ memset(&context->bw_ctx.bw.dcn.fams2_stream_base_params, 0, sizeof(union dmub_cmd_fams2_config) * DML2_MAX_PLANES); memset(&context->bw_ctx.bw.dcn.fams2_stream_sub_params, 0, sizeof(union dmub_cmd_fams2_config) * DML2_MAX_PLANES); + memset(&context->bw_ctx.bw.dcn.fams2_stream_sub_params_v2, 0, sizeof(union dmub_fams2_stream_static_sub_state_v2) * DML2_MAX_PLANES); memset(&context->bw_ctx.bw.dcn.fams2_global_config, 0, sizeof(struct dmub_cmd_fams2_global_config)); if (dml_ctx->v21.mode_programming.programming->fams2_required) { @@ -414,9 +415,16 @@ void dml21_build_fams2_programming(const struct dc *dc, memcpy(static_base_state, &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_idx].fams2_base_params, sizeof(union dmub_cmd_fams2_config)); - memcpy(static_sub_state, - &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_idx].fams2_sub_params, - sizeof(union dmub_cmd_fams2_config)); + + if (dc->debug.fams_version.major == 3) { + memcpy(&context->bw_ctx.bw.dcn.fams2_stream_sub_params_v2[num_fams2_streams], + &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_idx].fams2_sub_params_v2, + sizeof(union dmub_fams2_stream_static_sub_state_v2)); + } else { + memcpy(static_sub_state, + &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_idx].fams2_sub_params, + sizeof(union dmub_cmd_fams2_config)); + } switch (dc->debug.fams_version.minor) { case 1: diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c index ed6584535e898..03de3cf06ae59 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c @@ -2,8 +2,6 @@ // // Copyright 2024 Advanced Micro Devices, Inc. -#include <linux/vmalloc.h> - #include "dml2_internal_types.h" #include "dml_top.h" #include "dml2_core_dcn4_calcs.h" @@ -12,6 +10,8 @@ #include "dml21_translation_helper.h" #include "dml2_dc_resource_mgmt.h" +#define INVALID -1 + static bool dml21_allocate_memory(struct dml2_context **dml_ctx) { *dml_ctx = vzalloc(sizeof(struct dml2_context)); @@ -35,15 +35,11 @@ static bool dml21_allocate_memory(struct dml2_context **dml_ctx) return true; } -static void dml21_apply_debug_options(const struct dc *in_dc, struct dml2_context *dml_ctx, const struct dml2_configuration_options *config) +static void dml21_populate_configuration_options(const struct dc *in_dc, + struct dml2_context *dml_ctx, + const struct dml2_configuration_options *config) { - bool disable_fams2; - struct dml2_pmo_options *pmo_options = &dml_ctx->v21.dml_init.options.pmo_options; - - /* ODM options */ - pmo_options->disable_dyn_odm = !config->minimize_dispclk_using_odm; - pmo_options->disable_dyn_odm_for_multi_stream = true; - pmo_options->disable_dyn_odm_for_stream_with_svp = true; + dml_ctx->config = *config; /* UCLK P-State options */ if (in_dc->debug.dml21_force_pstate_method) { @@ -53,52 +49,20 @@ static void dml21_apply_debug_options(const struct dc *in_dc, struct dml2_contex } else { dml_ctx->config.pmo.force_pstate_method_enable = false; } - - pmo_options->disable_vblank = ((in_dc->debug.dml21_disable_pstate_method_mask >> 1) & 1); - - /* NOTE: DRR and SubVP Require FAMS2 */ - disable_fams2 = !in_dc->debug.fams2_config.bits.enable; - pmo_options->disable_svp = ((in_dc->debug.dml21_disable_pstate_method_mask >> 2) & 1) || - in_dc->debug.force_disable_subvp || - disable_fams2; - pmo_options->disable_drr_clamped = ((in_dc->debug.dml21_disable_pstate_method_mask >> 3) & 1) || - disable_fams2; - pmo_options->disable_drr_var = ((in_dc->debug.dml21_disable_pstate_method_mask >> 4) & 1) || - disable_fams2; - pmo_options->disable_fams2 = disable_fams2; - - pmo_options->disable_drr_var_when_var_active = in_dc->debug.disable_fams_gaming == INGAME_FAMS_DISABLE || - in_dc->debug.disable_fams_gaming == INGAME_FAMS_MULTI_DISP_CLAMPED_ONLY; - pmo_options->disable_drr_clamped_when_var_active = in_dc->debug.disable_fams_gaming == INGAME_FAMS_DISABLE; } -static void dml21_init(const struct dc *in_dc, struct dml2_context **dml_ctx, const struct dml2_configuration_options *config) +static void dml21_init(const struct dc *in_dc, struct dml2_context *dml_ctx, const struct dml2_configuration_options *config) { - switch (in_dc->ctx->dce_version) { - case DCN_VERSION_4_01: - (*dml_ctx)->v21.dml_init.options.project_id = dml2_project_dcn4x_stage2_auto_drr_svp; - break; - default: - (*dml_ctx)->v21.dml_init.options.project_id = dml2_project_invalid; - } - (*dml_ctx)->architecture = dml2_architecture_21; + dml_ctx->architecture = dml2_architecture_21; - /* Store configuration options */ - (*dml_ctx)->config = *config; + dml21_populate_configuration_options(in_dc, dml_ctx, config); DC_FP_START(); - /*Initialize SOCBB and DCNIP params */ - dml21_initialize_soc_bb_params(&(*dml_ctx)->v21.dml_init, config, in_dc); - dml21_initialize_ip_params(&(*dml_ctx)->v21.dml_init, config, in_dc); - dml21_apply_soc_bb_overrides(&(*dml_ctx)->v21.dml_init, config, in_dc); + dml21_populate_dml_init_params(&dml_ctx->v21.dml_init, config, in_dc); - /* apply debug overrides */ - dml21_apply_debug_options(in_dc, *dml_ctx, config); - - /*Initialize DML21 instance */ - dml2_initialize_instance(&(*dml_ctx)->v21.dml_init); + dml2_initialize_instance(&dml_ctx->v21.dml_init); DC_FP_END(); } @@ -109,7 +73,7 @@ bool dml21_create(const struct dc *in_dc, struct dml2_context **dml_ctx, const s if (!dml21_allocate_memory(dml_ctx)) return false; - dml21_init(in_dc, dml_ctx, config); + dml21_init(in_dc, *dml_ctx, config); return true; } @@ -208,10 +172,40 @@ static void dml21_calculate_rq_and_dlg_params(const struct dc *dc, struct dc_sta } } +static void dml21_prepare_mcache_params(struct dml2_context *dml_ctx, struct dc_state *context, struct dc_mcache_params *mcache_params) +{ + int dc_plane_idx = 0; + int dml_prog_idx, stream_idx, plane_idx; + struct dml2_per_plane_programming *pln_prog = NULL; + + for (stream_idx = 0; stream_idx < context->stream_count; stream_idx++) { + for (plane_idx = 0; plane_idx < context->stream_status[stream_idx].plane_count; plane_idx++) { + dml_prog_idx = map_plane_to_dml21_display_cfg(dml_ctx, context->streams[stream_idx]->stream_id, context->stream_status[stream_idx].plane_states[plane_idx], context); + if (dml_prog_idx == INVALID) { + continue; + } + pln_prog = &dml_ctx->v21.mode_programming.programming->plane_programming[dml_prog_idx]; + mcache_params[dc_plane_idx].valid = pln_prog->mcache_allocation.valid; + mcache_params[dc_plane_idx].num_mcaches_plane0 = pln_prog->mcache_allocation.num_mcaches_plane0; + mcache_params[dc_plane_idx].num_mcaches_plane1 = pln_prog->mcache_allocation.num_mcaches_plane1; + mcache_params[dc_plane_idx].requires_dedicated_mall_mcache = pln_prog->mcache_allocation.requires_dedicated_mall_mcache; + mcache_params[dc_plane_idx].last_slice_sharing.plane0_plane1 = pln_prog->mcache_allocation.last_slice_sharing.plane0_plane1; + memcpy(mcache_params[dc_plane_idx].mcache_x_offsets_plane0, + pln_prog->mcache_allocation.mcache_x_offsets_plane0, + sizeof(int) * (DML2_MAX_MCACHES + 1)); + memcpy(mcache_params[dc_plane_idx].mcache_x_offsets_plane1, + pln_prog->mcache_allocation.mcache_x_offsets_plane1, + sizeof(int) * (DML2_MAX_MCACHES + 1)); + dc_plane_idx++; + } + } +} + static bool dml21_mode_check_and_programming(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx) { bool result = false; struct dml2_build_mode_programming_in_out *mode_programming = &dml_ctx->v21.mode_programming; + struct dc_mcache_params mcache_params[MAX_PLANES] = {0}; memset(&dml_ctx->v21.display_config, 0, sizeof(struct dml2_display_cfg)); memset(&dml_ctx->v21.dml_to_dc_pipe_mapping, 0, sizeof(struct dml2_dml_to_dc_pipe_mapping)); @@ -246,6 +240,14 @@ static bool dml21_mode_check_and_programming(const struct dc *in_dc, struct dc_s dml2_map_dc_pipes(dml_ctx, context, NULL, &dml_ctx->v21.dml_to_dc_pipe_mapping, in_dc->current_state); /* if subvp phantoms are present, expand them into dc context */ dml21_handle_phantom_streams_planes(in_dc, context, dml_ctx); + + if (in_dc->res_pool->funcs->program_mcache_pipe_config) { + //Prepare mcache params for each plane based on mcache output from DML + dml21_prepare_mcache_params(dml_ctx, context, mcache_params); + + //populate mcache regs to each pipe + dml_ctx->config.callbacks.allocate_mcache(context, mcache_params); + } } /* Copy DML CLK, WM and REG outputs to bandwidth context */ @@ -288,12 +290,13 @@ static bool dml21_check_mode_support(const struct dc *in_dc, struct dc_state *co return true; } -bool dml21_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx, bool fast_validate) +bool dml21_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx, + enum dc_validate_mode validate_mode) { bool out = false; - /* Use dml_validate_only for fast_validate path */ - if (fast_validate) + /* Use dml21_check_mode_support for DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX path */ + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) out = dml21_check_mode_support(in_dc, context, dml_ctx); else out = dml21_mode_check_and_programming(in_dc, context, dml_ctx); @@ -456,7 +459,7 @@ bool dml21_create_copy(struct dml2_context **dst_dml_ctx, return true; } -void dml21_reinit(const struct dc *in_dc, struct dml2_context **dml_ctx, const struct dml2_configuration_options *config) +void dml21_reinit(const struct dc *in_dc, struct dml2_context *dml_ctx, const struct dml2_configuration_options *config) { dml21_init(in_dc, dml_ctx, config); } diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.h index b2075b8c363b5..15f92029d2e59 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.h @@ -8,11 +8,13 @@ #include "os_types.h" #include "dml_top_soc_parameter_types.h" +#include "dml_top_display_cfg_types.h" struct dc; struct dc_state; struct dml2_configuration_options; struct dml2_context; +enum dc_validate_mode; /** * dml2_create - Creates dml21_context. @@ -32,22 +34,23 @@ void dml21_copy(struct dml2_context *dst_dml_ctx, struct dml2_context *src_dml_ctx); bool dml21_create_copy(struct dml2_context **dst_dml_ctx, struct dml2_context *src_dml_ctx); -void dml21_reinit(const struct dc *in_dc, struct dml2_context **dml_ctx, const struct dml2_configuration_options *config); +void dml21_reinit(const struct dc *in_dc, struct dml2_context *dml_ctx, const struct dml2_configuration_options *config); /** * dml21_validate - Determines if a display configuration is supported or not. * @in_dc: dc. * @context: dc_state to be validated. - * @fast_validate: Fast validate will not populate context.res_ctx. + * @validate_mode: DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX + * will not populate context.res_ctx. * * Based on fast_validate option internally would call: * - * -dml21_mode_check_and_programming - for non fast_validate option + * -dml21_mode_check_and_programming - for DC_VALIDATE_MODE_AND_PROGRAMMING option * Calculates if dc_state can be supported on the input display * configuration. If supported, generates the necessary HW * programming for the new dc_state. * - * -dml21_check_mode_support - for fast_validate option + * -dml21_check_mode_support - for DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX option * Calculates if dc_state can be supported for the input display * config. @@ -55,7 +58,8 @@ void dml21_reinit(const struct dc *in_dc, struct dml2_context **dml_ctx, const s * separate dc_states for validation. * Return: True if mode is supported, false otherwise. */ -bool dml21_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx, bool fast_validate); +bool dml21_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx, + enum dc_validate_mode validate_mode); /* Prepare hubp mcache_regs for hubp mcache ID and split coordinate programming */ void dml21_prepare_mcache_programming(struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx); @@ -65,4 +69,67 @@ struct socbb_ip_params_external { struct dml2_ip_capabilities ip_params; struct dml2_soc_bb soc_bb; }; + +/*mcache parameters decided by dml*/ +struct dc_mcache_params { + bool valid; + /* + * For iMALL, dedicated mall mcaches are required (sharing of last + * slice possible), for legacy phantom or phantom without return + * the only mall mcaches need to be valid. + */ + bool requires_dedicated_mall_mcache; + unsigned int num_mcaches_plane0; + unsigned int num_mcaches_plane1; + /* + * Generally, plane0/1 slices must use a disjoint set of caches + * but in some cases the final segement of the two planes can + * use the same cache. If plane0_plane1 is set, then this is + * allowed. + * + * Similarly, the caches allocated to MALL prefetcher are generally + * disjoint, but if mall_prefetch is set, then the final segment + * between the main and the mall pixel requestor can use the same + * cache. + * + * Note that both bits may be set at the same time. + */ + struct { + bool mall_comb_mcache_p0; + bool mall_comb_mcache_p1; + bool plane0_plane1; + } last_slice_sharing; + /* + * A plane is divided into vertical slices of mcaches, + * which wrap on the surface width. + * + * For example, if the surface width is 7680, and split into + * three slices of equal width, the boundary array would contain + * [2560, 5120, 7680] + * + * The assignments are + * 0 = [0 .. 2559] + * 1 = [2560 .. 5119] + * 2 = [5120 .. 7679] + * 0 = [7680 .. INF] + * The final element implicitly is the same as the first, and + * at first seems invalid since it is never referenced (since) + * it is outside the surface. However, its useful when shifting + * (see below). + * + * For any given valid mcache assignment, a shifted version, wrapped + * on the surface width boundary is also assumed to be valid. + * + * For example, shifting [2560, 5120, 7680] by -50 results in + * [2510, 5170, 7630]. + * + * The assignments are now: + * 0 = [0 .. 2509] + * 1 = [2510 .. 5169] + * 2 = [5170 .. 7629] + * 0 = [7630 .. INF] + */ + int mcache_x_offsets_plane0[DML2_MAX_MCACHES + 1]; + int mcache_x_offsets_plane1[DML2_MAX_MCACHES + 1]; +}; #endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_dchub_registers.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_dchub_registers.h index 25b607e7b726e..b05030926ce85 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_dchub_registers.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_dchub_registers.h @@ -46,6 +46,7 @@ struct dml2_display_dlg_regs { uint32_t dst_y_delta_drq_limit; uint32_t refcyc_per_vm_dmdata; uint32_t dmdata_dl_delta; + uint32_t dst_y_svp_drq_limit; // MRQ uint32_t refcyc_per_meta_chunk_vblank_l; @@ -156,6 +157,8 @@ struct dml2_dchub_watermark_regs { uint32_t urgent; uint32_t sr_enter; uint32_t sr_exit; + uint32_t sr_enter_z8; + uint32_t sr_exit_z8; uint32_t uclk_pstate; uint32_t fclk_pstate; uint32_t temp_read_or_ppt; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_display_cfg_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_display_cfg_types.h index 5e1ab6d976404..e8dc6471c0bec 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_display_cfg_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_display_cfg_types.h @@ -166,7 +166,7 @@ struct dml2_surface_cfg { enum dml2_swizzle_mode tiling; struct { - unsigned long pitch; + unsigned long pitch; // In elements, two pixels per element in 422 packed format unsigned long width; unsigned long height; } plane0; @@ -222,6 +222,7 @@ struct dml2_composition_cfg { struct { bool enabled; + bool upsp_enabled; struct { double h_ratio; double v_ratio; @@ -385,6 +386,7 @@ struct dml2_plane_parameters { long reserved_vblank_time_ns; unsigned int max_vactive_det_fill_delay_us; // 0 = no reserved time, +ve = explicit max delay unsigned int gpuvm_min_page_size_kbytes; + unsigned int hostvm_min_page_size_kbytes; enum dml2_svp_mode_override legacy_svp_config; //TODO remove in favor of svp_config @@ -425,6 +427,7 @@ struct dml2_stream_parameters { struct dml2_display_cfg { bool gpuvm_enable; + bool ffbm_enable; bool hostvm_enable; // Allocate DET proportionally between streams based on pixel rate diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_soc_parameter_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_soc_parameter_types.h index 5f0bc42d1d2f7..8c9f414aa6bf9 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_soc_parameter_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_soc_parameter_types.h @@ -93,12 +93,15 @@ struct dml2_soc_power_management_parameters { double dram_clk_change_write_only_us; double fclk_change_blackout_us; double g7_ppt_blackout_us; + double g7_temperature_read_blackout_us; double stutter_enter_plus_exit_latency_us; double stutter_exit_latency_us; double z8_stutter_enter_plus_exit_latency_us; double z8_stutter_exit_latency_us; double z8_min_idle_time; double g6_temp_read_blackout_us[DML_MAX_CLK_TABLE_SIZE]; + double type_b_dram_clk_change_blackout_us; + double type_b_ppt_blackout_us; }; struct dml2_clk_table { @@ -130,6 +133,7 @@ struct dml2_soc_state_table { struct dml2_soc_vmin_clock_limits { unsigned long dispclk_khz; + unsigned long dcfclk_khz; }; struct dml2_soc_bb { @@ -138,6 +142,7 @@ struct dml2_soc_bb { struct dml2_soc_power_management_parameters power_management_parameters; struct dml2_soc_vmin_clock_limits vmin_limit; + double lower_bound_bandwidth_dchub; unsigned int dprefclk_mhz; unsigned int xtalclk_mhz; unsigned int pcie_refclk_mhz; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h index 0dbf886d89261..98c0234e2f474 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h @@ -53,7 +53,9 @@ enum dml2_output_type_and_rate__rate { dml2_output_rate_hdmi_rate_6x4 = 9, dml2_output_rate_hdmi_rate_8x4 = 10, dml2_output_rate_hdmi_rate_10x4 = 11, - dml2_output_rate_hdmi_rate_12x4 = 12 + dml2_output_rate_hdmi_rate_12x4 = 12, + dml2_output_rate_hdmi_rate_16x4 = 13, + dml2_output_rate_hdmi_rate_20x4 = 14 }; struct dml2_pmo_options { @@ -279,7 +281,10 @@ struct dml2_per_stream_programming { } phantom_stream; union dmub_cmd_fams2_config fams2_base_params; - union dmub_cmd_fams2_config fams2_sub_params; + union { + union dmub_cmd_fams2_config fams2_sub_params; + union dmub_fams2_stream_static_sub_state_v2 fams2_sub_params_v2; + }; }; //----------------- @@ -674,9 +679,14 @@ struct dml2_display_cfg_programming { // unlimited # of mcache struct dml2_mcache_surface_allocation non_optimized_mcache_allocation[DML2_MAX_PLANES]; + bool failed_prefetch; + bool failed_uclk_pstate; bool failed_mcache_validation; bool failed_dpmm; bool failed_mode_programming; + bool failed_mode_programming_dcfclk; + bool failed_mode_programming_prefetch; + bool failed_mode_programming_flip; bool failed_map_watermarks; } informative; }; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c index bb863c8c6b39c..6ee37386f6728 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c @@ -456,10 +456,10 @@ bool core_dcn4_mode_support(struct dml2_core_mode_support_in_out *in_out) in_out->mode_support_result.global.active.urgent_bw_dram_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->urg_bandwidth_required_flip[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram] * 1000), 1.0); in_out->mode_support_result.global.svp_prefetch.average_bw_dram_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] * 1000), 1.0); in_out->mode_support_result.global.svp_prefetch.urgent_bw_dram_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->urg_bandwidth_required_flip[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] * 1000), 1.0); - dml2_printf("DML::%s: in_out->mode_support_result.global.active.urgent_bw_sdp_kbps = %ld\n", __func__, in_out->mode_support_result.global.active.urgent_bw_sdp_kbps); - dml2_printf("DML::%s: in_out->mode_support_result.global.svp_prefetch.urgent_bw_sdp_kbps = %ld\n", __func__, in_out->mode_support_result.global.svp_prefetch.urgent_bw_sdp_kbps); - dml2_printf("DML::%s: in_out->mode_support_result.global.active.urgent_bw_dram_kbps = %ld\n", __func__, in_out->mode_support_result.global.active.urgent_bw_dram_kbps); - dml2_printf("DML::%s: in_out->mode_support_result.global.svp_prefetch.urgent_bw_dram_kbps = %ld\n", __func__, in_out->mode_support_result.global.svp_prefetch.urgent_bw_dram_kbps); + DML_LOG_VERBOSE("DML::%s: in_out->mode_support_result.global.active.urgent_bw_sdp_kbps = %ld\n", __func__, in_out->mode_support_result.global.active.urgent_bw_sdp_kbps); + DML_LOG_VERBOSE("DML::%s: in_out->mode_support_result.global.svp_prefetch.urgent_bw_sdp_kbps = %ld\n", __func__, in_out->mode_support_result.global.svp_prefetch.urgent_bw_sdp_kbps); + DML_LOG_VERBOSE("DML::%s: in_out->mode_support_result.global.active.urgent_bw_dram_kbps = %ld\n", __func__, in_out->mode_support_result.global.active.urgent_bw_dram_kbps); + DML_LOG_VERBOSE("DML::%s: in_out->mode_support_result.global.svp_prefetch.urgent_bw_dram_kbps = %ld\n", __func__, in_out->mode_support_result.global.svp_prefetch.urgent_bw_dram_kbps); for (i = 0; i < l->svp_expanded_display_cfg.num_planes; i++) { in_out->mode_support_result.per_plane[i].dppclk_khz = (unsigned int)(core->clean_me_up.mode_lib.ms.RequiredDPPCLK[i] * 1000); @@ -509,7 +509,7 @@ bool core_dcn4_mode_support(struct dml2_core_mode_support_in_out *in_out) stream_index = l->svp_expanded_display_cfg.plane_descriptors[i].stream_index; in_out->mode_support_result.per_stream[stream_index].dscclk_khz = (unsigned int)core->clean_me_up.mode_lib.ms.required_dscclk_freq_mhz[i] * 1000; - dml2_printf("CORE_DCN4::%s: i=%d stream_index=%d, in_out->mode_support_result.per_stream[stream_index].dscclk_khz = %u\n", __func__, i, stream_index, in_out->mode_support_result.per_stream[stream_index].dscclk_khz); + DML_LOG_VERBOSE("CORE_DCN4::%s: i=%d stream_index=%d, in_out->mode_support_result.per_stream[stream_index].dscclk_khz = %u\n", __func__, i, stream_index, in_out->mode_support_result.per_stream[stream_index].dscclk_khz); if (!((stream_bitmask >> stream_index) & 0x1)) { in_out->mode_support_result.cfg_support_info.stream_support_info[stream_index].odms_used = odm_count; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c index 4c504cb0e1c53..b9cff21985110 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c @@ -54,104 +54,104 @@ static double dml2_core_div_rem(double dividend, unsigned int divisor, unsigned static void dml2_print_mode_support_info(const struct dml2_core_internal_mode_support_info *support, bool fail_only) { - dml2_printf("DML: ===================================== \n"); - dml2_printf("DML: DML_MODE_SUPPORT_INFO_ST\n"); + DML_LOG_VERBOSE("DML: ===================================== \n"); + DML_LOG_VERBOSE("DML: DML_MODE_SUPPORT_INFO_ST\n"); if (!fail_only || support->ScaleRatioAndTapsSupport == 0) - dml2_printf("DML: support: ScaleRatioAndTapsSupport = %d\n", support->ScaleRatioAndTapsSupport); + DML_LOG_VERBOSE("DML: support: ScaleRatioAndTapsSupport = %d\n", support->ScaleRatioAndTapsSupport); if (!fail_only || support->SourceFormatPixelAndScanSupport == 0) - dml2_printf("DML: support: SourceFormatPixelAndScanSupport = %d\n", support->SourceFormatPixelAndScanSupport); + DML_LOG_VERBOSE("DML: support: SourceFormatPixelAndScanSupport = %d\n", support->SourceFormatPixelAndScanSupport); if (!fail_only || support->ViewportSizeSupport == 0) - dml2_printf("DML: support: ViewportSizeSupport = %d\n", support->ViewportSizeSupport); + DML_LOG_VERBOSE("DML: support: ViewportSizeSupport = %d\n", support->ViewportSizeSupport); if (!fail_only || support->LinkRateDoesNotMatchDPVersion == 1) - dml2_printf("DML: support: LinkRateDoesNotMatchDPVersion = %d\n", support->LinkRateDoesNotMatchDPVersion); + DML_LOG_VERBOSE("DML: support: LinkRateDoesNotMatchDPVersion = %d\n", support->LinkRateDoesNotMatchDPVersion); if (!fail_only || support->LinkRateForMultistreamNotIndicated == 1) - dml2_printf("DML: support: LinkRateForMultistreamNotIndicated = %d\n", support->LinkRateForMultistreamNotIndicated); + DML_LOG_VERBOSE("DML: support: LinkRateForMultistreamNotIndicated = %d\n", support->LinkRateForMultistreamNotIndicated); if (!fail_only || support->BPPForMultistreamNotIndicated == 1) - dml2_printf("DML: support: BPPForMultistreamNotIndicated = %d\n", support->BPPForMultistreamNotIndicated); + DML_LOG_VERBOSE("DML: support: BPPForMultistreamNotIndicated = %d\n", support->BPPForMultistreamNotIndicated); if (!fail_only || support->MultistreamWithHDMIOreDP == 1) - dml2_printf("DML: support: MultistreamWithHDMIOreDP = %d\n", support->MultistreamWithHDMIOreDP); + DML_LOG_VERBOSE("DML: support: MultistreamWithHDMIOreDP = %d\n", support->MultistreamWithHDMIOreDP); if (!fail_only || support->ExceededMultistreamSlots == 1) - dml2_printf("DML: support: ExceededMultistreamSlots = %d\n", support->ExceededMultistreamSlots); + DML_LOG_VERBOSE("DML: support: ExceededMultistreamSlots = %d\n", support->ExceededMultistreamSlots); if (!fail_only || support->MSOOrODMSplitWithNonDPLink == 1) - dml2_printf("DML: support: MSOOrODMSplitWithNonDPLink = %d\n", support->MSOOrODMSplitWithNonDPLink); + DML_LOG_VERBOSE("DML: support: MSOOrODMSplitWithNonDPLink = %d\n", support->MSOOrODMSplitWithNonDPLink); if (!fail_only || support->NotEnoughLanesForMSO == 1) - dml2_printf("DML: support: NotEnoughLanesForMSO = %d\n", support->NotEnoughLanesForMSO); + DML_LOG_VERBOSE("DML: support: NotEnoughLanesForMSO = %d\n", support->NotEnoughLanesForMSO); if (!fail_only || support->P2IWith420 == 1) - dml2_printf("DML: support: P2IWith420 = %d\n", support->P2IWith420); + DML_LOG_VERBOSE("DML: support: P2IWith420 = %d\n", support->P2IWith420); if (!fail_only || support->DSC422NativeNotSupported == 1) - dml2_printf("DML: support: DSC422NativeNotSupported = %d\n", support->DSC422NativeNotSupported); + DML_LOG_VERBOSE("DML: support: DSC422NativeNotSupported = %d\n", support->DSC422NativeNotSupported); if (!fail_only || support->DSCSlicesODMModeSupported == 0) - dml2_printf("DML: support: DSCSlicesODMModeSupported = %d\n", support->DSCSlicesODMModeSupported); + DML_LOG_VERBOSE("DML: support: DSCSlicesODMModeSupported = %d\n", support->DSCSlicesODMModeSupported); if (!fail_only || support->NotEnoughDSCUnits == 1) - dml2_printf("DML: support: NotEnoughDSCUnits = %d\n", support->NotEnoughDSCUnits); + DML_LOG_VERBOSE("DML: support: NotEnoughDSCUnits = %d\n", support->NotEnoughDSCUnits); if (!fail_only || support->NotEnoughDSCSlices == 1) - dml2_printf("DML: support: NotEnoughDSCSlices = %d\n", support->NotEnoughDSCSlices); + DML_LOG_VERBOSE("DML: support: NotEnoughDSCSlices = %d\n", support->NotEnoughDSCSlices); if (!fail_only || support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe == 1) - dml2_printf("DML: support: ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = %d\n", support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe); + DML_LOG_VERBOSE("DML: support: ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = %d\n", support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe); if (!fail_only || support->InvalidCombinationOfMALLUseForPStateAndStaticScreen == 1) - dml2_printf("DML: support: InvalidCombinationOfMALLUseForPStateAndStaticScreen = %d\n", support->InvalidCombinationOfMALLUseForPStateAndStaticScreen); + DML_LOG_VERBOSE("DML: support: InvalidCombinationOfMALLUseForPStateAndStaticScreen = %d\n", support->InvalidCombinationOfMALLUseForPStateAndStaticScreen); if (!fail_only || support->DSCCLKRequiredMoreThanSupported == 1) - dml2_printf("DML: support: DSCCLKRequiredMoreThanSupported = %d\n", support->DSCCLKRequiredMoreThanSupported); + DML_LOG_VERBOSE("DML: support: DSCCLKRequiredMoreThanSupported = %d\n", support->DSCCLKRequiredMoreThanSupported); if (!fail_only || support->PixelsPerLinePerDSCUnitSupport == 0) - dml2_printf("DML: support: PixelsPerLinePerDSCUnitSupport = %d\n", support->PixelsPerLinePerDSCUnitSupport); + DML_LOG_VERBOSE("DML: support: PixelsPerLinePerDSCUnitSupport = %d\n", support->PixelsPerLinePerDSCUnitSupport); if (!fail_only || support->DTBCLKRequiredMoreThanSupported == 1) - dml2_printf("DML: support: DTBCLKRequiredMoreThanSupported = %d\n", support->DTBCLKRequiredMoreThanSupported); + DML_LOG_VERBOSE("DML: support: DTBCLKRequiredMoreThanSupported = %d\n", support->DTBCLKRequiredMoreThanSupported); if (!fail_only || support->InvalidCombinationOfMALLUseForPState == 1) - dml2_printf("DML: support: InvalidCombinationOfMALLUseForPState = %d\n", support->InvalidCombinationOfMALLUseForPState); + DML_LOG_VERBOSE("DML: support: InvalidCombinationOfMALLUseForPState = %d\n", support->InvalidCombinationOfMALLUseForPState); if (!fail_only || support->ROBSupport == 0) - dml2_printf("DML: support: ROBSupport = %d\n", support->ROBSupport); + DML_LOG_VERBOSE("DML: support: ROBSupport = %d\n", support->ROBSupport); if (!fail_only || support->OutstandingRequestsSupport == 0) - dml2_printf("DML: support: OutstandingRequestsSupport = %d\n", support->OutstandingRequestsSupport); + DML_LOG_VERBOSE("DML: support: OutstandingRequestsSupport = %d\n", support->OutstandingRequestsSupport); if (!fail_only || support->OutstandingRequestsUrgencyAvoidance == 0) - dml2_printf("DML: support: OutstandingRequestsUrgencyAvoidance = %d\n", support->OutstandingRequestsUrgencyAvoidance); + DML_LOG_VERBOSE("DML: support: OutstandingRequestsUrgencyAvoidance = %d\n", support->OutstandingRequestsUrgencyAvoidance); if (!fail_only || support->DISPCLK_DPPCLK_Support == 0) - dml2_printf("DML: support: DISPCLK_DPPCLK_Support = %d\n", support->DISPCLK_DPPCLK_Support); + DML_LOG_VERBOSE("DML: support: DISPCLK_DPPCLK_Support = %d\n", support->DISPCLK_DPPCLK_Support); if (!fail_only || support->TotalAvailablePipesSupport == 0) - dml2_printf("DML: support: TotalAvailablePipesSupport = %d\n", support->TotalAvailablePipesSupport); + DML_LOG_VERBOSE("DML: support: TotalAvailablePipesSupport = %d\n", support->TotalAvailablePipesSupport); if (!fail_only || support->NumberOfOTGSupport == 0) - dml2_printf("DML: support: NumberOfOTGSupport = %d\n", support->NumberOfOTGSupport); + DML_LOG_VERBOSE("DML: support: NumberOfOTGSupport = %d\n", support->NumberOfOTGSupport); if (!fail_only || support->NumberOfHDMIFRLSupport == 0) - dml2_printf("DML: support: NumberOfHDMIFRLSupport = %d\n", support->NumberOfHDMIFRLSupport); + DML_LOG_VERBOSE("DML: support: NumberOfHDMIFRLSupport = %d\n", support->NumberOfHDMIFRLSupport); if (!fail_only || support->NumberOfDP2p0Support == 0) - dml2_printf("DML: support: NumberOfDP2p0Support = %d\n", support->NumberOfDP2p0Support); + DML_LOG_VERBOSE("DML: support: NumberOfDP2p0Support = %d\n", support->NumberOfDP2p0Support); if (!fail_only || support->EnoughWritebackUnits == 0) - dml2_printf("DML: support: EnoughWritebackUnits = %d\n", support->EnoughWritebackUnits); + DML_LOG_VERBOSE("DML: support: EnoughWritebackUnits = %d\n", support->EnoughWritebackUnits); if (!fail_only || support->WritebackScaleRatioAndTapsSupport == 0) - dml2_printf("DML: support: WritebackScaleRatioAndTapsSupport = %d\n", support->WritebackScaleRatioAndTapsSupport); + DML_LOG_VERBOSE("DML: support: WritebackScaleRatioAndTapsSupport = %d\n", support->WritebackScaleRatioAndTapsSupport); if (!fail_only || support->WritebackLatencySupport == 0) - dml2_printf("DML: support: WritebackLatencySupport = %d\n", support->WritebackLatencySupport); + DML_LOG_VERBOSE("DML: support: WritebackLatencySupport = %d\n", support->WritebackLatencySupport); if (!fail_only || support->CursorSupport == 0) - dml2_printf("DML: support: CursorSupport = %d\n", support->CursorSupport); + DML_LOG_VERBOSE("DML: support: CursorSupport = %d\n", support->CursorSupport); if (!fail_only || support->PitchSupport == 0) - dml2_printf("DML: support: PitchSupport = %d\n", support->PitchSupport); + DML_LOG_VERBOSE("DML: support: PitchSupport = %d\n", support->PitchSupport); if (!fail_only || support->ViewportExceedsSurface == 1) - dml2_printf("DML: support: ViewportExceedsSurface = %d\n", support->ViewportExceedsSurface); + DML_LOG_VERBOSE("DML: support: ViewportExceedsSurface = %d\n", support->ViewportExceedsSurface); if (!fail_only || support->PrefetchSupported == 0) - dml2_printf("DML: support: PrefetchSupported = %d\n", support->PrefetchSupported); + DML_LOG_VERBOSE("DML: support: PrefetchSupported = %d\n", support->PrefetchSupported); if (!fail_only || support->EnoughUrgentLatencyHidingSupport == 0) - dml2_printf("DML: support: EnoughUrgentLatencyHidingSupport = %d\n", support->EnoughUrgentLatencyHidingSupport); + DML_LOG_VERBOSE("DML: support: EnoughUrgentLatencyHidingSupport = %d\n", support->EnoughUrgentLatencyHidingSupport); if (!fail_only || support->AvgBandwidthSupport == 0) - dml2_printf("DML: support: AvgBandwidthSupport = %d\n", support->AvgBandwidthSupport); + DML_LOG_VERBOSE("DML: support: AvgBandwidthSupport = %d\n", support->AvgBandwidthSupport); if (!fail_only || support->DynamicMetadataSupported == 0) - dml2_printf("DML: support: DynamicMetadataSupported = %d\n", support->DynamicMetadataSupported); + DML_LOG_VERBOSE("DML: support: DynamicMetadataSupported = %d\n", support->DynamicMetadataSupported); if (!fail_only || support->VRatioInPrefetchSupported == 0) - dml2_printf("DML: support: VRatioInPrefetchSupported = %d\n", support->VRatioInPrefetchSupported); + DML_LOG_VERBOSE("DML: support: VRatioInPrefetchSupported = %d\n", support->VRatioInPrefetchSupported); if (!fail_only || support->PTEBufferSizeNotExceeded == 0) - dml2_printf("DML: support: PTEBufferSizeNotExceeded = %d\n", support->PTEBufferSizeNotExceeded); + DML_LOG_VERBOSE("DML: support: PTEBufferSizeNotExceeded = %d\n", support->PTEBufferSizeNotExceeded); if (!fail_only || support->DCCMetaBufferSizeNotExceeded == 0) - dml2_printf("DML: support: DCCMetaBufferSizeNotExceeded = %d\n", support->DCCMetaBufferSizeNotExceeded); + DML_LOG_VERBOSE("DML: support: DCCMetaBufferSizeNotExceeded = %d\n", support->DCCMetaBufferSizeNotExceeded); if (!fail_only || support->ExceededMALLSize == 1) - dml2_printf("DML: support: ExceededMALLSize = %d\n", support->ExceededMALLSize); + DML_LOG_VERBOSE("DML: support: ExceededMALLSize = %d\n", support->ExceededMALLSize); if (!fail_only || support->g6_temp_read_support == 0) - dml2_printf("DML: support: g6_temp_read_support = %d\n", support->g6_temp_read_support); + DML_LOG_VERBOSE("DML: support: g6_temp_read_support = %d\n", support->g6_temp_read_support); if (!fail_only || support->ImmediateFlipSupport == 0) - dml2_printf("DML: support: ImmediateFlipSupport = %d\n", support->ImmediateFlipSupport); + DML_LOG_VERBOSE("DML: support: ImmediateFlipSupport = %d\n", support->ImmediateFlipSupport); if (!fail_only || support->LinkCapacitySupport == 0) - dml2_printf("DML: support: LinkCapacitySupport = %d\n", support->LinkCapacitySupport); + DML_LOG_VERBOSE("DML: support: LinkCapacitySupport = %d\n", support->LinkCapacitySupport); if (!fail_only || support->ModeSupport == 0) - dml2_printf("DML: support: ModeSupport = %d\n", support->ModeSupport); - dml2_printf("DML: ===================================== \n"); + DML_LOG_VERBOSE("DML: support: ModeSupport = %d\n", support->ModeSupport); + DML_LOG_VERBOSE("DML: ===================================== \n"); } static void get_stream_output_bpp(double *out_bpp, const struct dml2_display_cfg *display_cfg) @@ -179,11 +179,9 @@ static void get_stream_output_bpp(double *out_bpp, const struct dml2_display_cfg } else { out_bpp[k] = 0; } -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%d bpc=%f\n", __func__, k, bpc); - dml2_printf("DML::%s: k=%d dsc.enable=%d\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable); - dml2_printf("DML::%s: k=%d out_bpp=%f\n", __func__, k, out_bpp[k]); -#endif + DML_LOG_VERBOSE("DML::%s: k=%d bpc=%f\n", __func__, k, bpc); + DML_LOG_VERBOSE("DML::%s: k=%d dsc.enable=%d\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable); + DML_LOG_VERBOSE("DML::%s: k=%d out_bpp=%f\n", __func__, k, out_bpp[k]); } } @@ -212,9 +210,7 @@ static unsigned int dml_get_num_active_pipes(int unsigned num_planes, const stru num_active_pipes = num_active_pipes + (unsigned int)cfg_support_info->plane_support_info[k].dpps_used; } -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: num_active_pipes = %d\n", __func__, num_active_pipes); -#endif + DML_LOG_VERBOSE("DML::%s: num_active_pipes = %d\n", __func__, num_active_pipes); return num_active_pipes; } @@ -251,7 +247,7 @@ static bool dml_get_is_phantom_pipe(const struct dml2_display_cfg *display_cfg, unsigned int plane_idx = mode_lib->mp.pipe_plane[pipe_idx]; bool is_phantom = dml_is_phantom_pipe(&display_cfg->plane_descriptors[plane_idx]); - dml2_printf("DML::%s: pipe_idx=%d legacy_svp_config=%0d is_phantom=%d\n", __func__, pipe_idx, display_cfg->plane_descriptors[plane_idx].overrides.legacy_svp_config, is_phantom); + DML_LOG_VERBOSE("DML::%s: pipe_idx=%d legacy_svp_config=%0d is_phantom=%d\n", __func__, pipe_idx, display_cfg->plane_descriptors[plane_idx].overrides.legacy_svp_config, is_phantom); return is_phantom; } @@ -415,19 +411,17 @@ static void CalculateMaxDETAndMinCompressedBufferSize( *nomDETInKByte = (unsigned int)(math_floor2((double)*MaxTotalDETInKByte / (double)MaxNumDPP, ConfigReturnBufferSegmentSizeInKByte)); *MinCompressedBufferSizeInKByte = ConfigReturnBufferSizeInKByte - *MaxTotalDETInKByte; -#if defined(__DML_VBA_DEBUG__) - dml2_printf("DML::%s: is_mrq_present = %u\n", __func__, is_mrq_present); - dml2_printf("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, ConfigReturnBufferSizeInKByte); - dml2_printf("DML::%s: ROBBufferSizeInKByte = %u\n", __func__, ROBBufferSizeInKByte); - dml2_printf("DML::%s: MaxNumDPP = %u\n", __func__, MaxNumDPP); - dml2_printf("DML::%s: MaxTotalDETInKByte = %u\n", __func__, *MaxTotalDETInKByte); - dml2_printf("DML::%s: nomDETInKByte = %u\n", __func__, *nomDETInKByte); - dml2_printf("DML::%s: MinCompressedBufferSizeInKByte = %u\n", __func__, *MinCompressedBufferSizeInKByte); -#endif + DML_LOG_VERBOSE("DML::%s: is_mrq_present = %u\n", __func__, is_mrq_present); + DML_LOG_VERBOSE("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, ConfigReturnBufferSizeInKByte); + DML_LOG_VERBOSE("DML::%s: ROBBufferSizeInKByte = %u\n", __func__, ROBBufferSizeInKByte); + DML_LOG_VERBOSE("DML::%s: MaxNumDPP = %u\n", __func__, MaxNumDPP); + DML_LOG_VERBOSE("DML::%s: MaxTotalDETInKByte = %u\n", __func__, *MaxTotalDETInKByte); + DML_LOG_VERBOSE("DML::%s: nomDETInKByte = %u\n", __func__, *nomDETInKByte); + DML_LOG_VERBOSE("DML::%s: MinCompressedBufferSizeInKByte = %u\n", __func__, *MinCompressedBufferSizeInKByte); if (nomDETInKByteOverrideEnable) { *nomDETInKByte = nomDETInKByteOverrideValue; - dml2_printf("DML::%s: nomDETInKByte = %u (overrided)\n", __func__, *nomDETInKByte); + DML_LOG_VERBOSE("DML::%s: nomDETInKByte = %u (overrided)\n", __func__, *nomDETInKByte); } } @@ -502,7 +496,7 @@ static bool dml_is_420(enum dml2_source_format_class source_format) val = 0; break; default: - DML2_ASSERT(0); + DML_ASSERT(0); break; } return val; @@ -535,7 +529,7 @@ static unsigned int dml_get_tile_block_size_bytes(enum dml2_swizzle_mode sw_mode else if (sw_mode == dml2_gfx11_sw_256kb_r_x) return 262144; else { - DML2_ASSERT(0); + DML_ASSERT(0); return 256; } } @@ -570,8 +564,8 @@ static int unsigned dml_get_gfx_version(enum dml2_swizzle_mode sw_mode) sw_mode == dml2_gfx11_sw_256kb_r_x) { version = 11; } else { - dml2_printf("ERROR: Invalid sw_mode setting! val=%u\n", sw_mode); - DML2_ASSERT(0); + DML_LOG_VERBOSE("ERROR: Invalid sw_mode setting! val=%u\n", sw_mode); + DML_ASSERT(0); } return version; @@ -645,21 +639,19 @@ static void CalculateBytePerPixelAndBlockSizes( *BytePerPixelY = 2; *BytePerPixelC = 4; } else { - dml2_printf("ERROR: DML::%s: SourcePixelFormat = %u not supported!\n", __func__, SourcePixelFormat); - DML2_ASSERT(0); + DML_LOG_VERBOSE("ERROR: DML::%s: SourcePixelFormat = %u not supported!\n", __func__, SourcePixelFormat); + DML_ASSERT(0); } -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: SourcePixelFormat = %u\n", __func__, SourcePixelFormat); - dml2_printf("DML::%s: BytePerPixelDETY = %f\n", __func__, *BytePerPixelDETY); - dml2_printf("DML::%s: BytePerPixelDETC = %f\n", __func__, *BytePerPixelDETC); - dml2_printf("DML::%s: BytePerPixelY = %u\n", __func__, *BytePerPixelY); - dml2_printf("DML::%s: BytePerPixelC = %u\n", __func__, *BytePerPixelC); - dml2_printf("DML::%s: pitch_y = %u\n", __func__, pitch_y); - dml2_printf("DML::%s: pitch_c = %u\n", __func__, pitch_c); - dml2_printf("DML::%s: surf_linear128_l = %u\n", __func__, *surf_linear128_l); - dml2_printf("DML::%s: surf_linear128_c = %u\n", __func__, *surf_linear128_c); -#endif + DML_LOG_VERBOSE("DML::%s: SourcePixelFormat = %u\n", __func__, SourcePixelFormat); + DML_LOG_VERBOSE("DML::%s: BytePerPixelDETY = %f\n", __func__, *BytePerPixelDETY); + DML_LOG_VERBOSE("DML::%s: BytePerPixelDETC = %f\n", __func__, *BytePerPixelDETC); + DML_LOG_VERBOSE("DML::%s: BytePerPixelY = %u\n", __func__, *BytePerPixelY); + DML_LOG_VERBOSE("DML::%s: BytePerPixelC = %u\n", __func__, *BytePerPixelC); + DML_LOG_VERBOSE("DML::%s: pitch_y = %u\n", __func__, pitch_y); + DML_LOG_VERBOSE("DML::%s: pitch_c = %u\n", __func__, pitch_c); + DML_LOG_VERBOSE("DML::%s: surf_linear128_l = %u\n", __func__, *surf_linear128_l); + DML_LOG_VERBOSE("DML::%s: surf_linear128_c = %u\n", __func__, *surf_linear128_c); if (dml_get_gfx_version(SurfaceTiling) == 11) { *surf_linear128_l = 0; @@ -703,12 +695,10 @@ static void CalculateBytePerPixelAndBlockSizes( *BlockWidth256BytesY = 256U / *BytePerPixelY / *BlockHeight256BytesY; *BlockWidth256BytesC = 256U / *BytePerPixelC / *BlockHeight256BytesC; } -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: BlockWidth256BytesY = %u\n", __func__, *BlockWidth256BytesY); - dml2_printf("DML::%s: BlockHeight256BytesY = %u\n", __func__, *BlockHeight256BytesY); - dml2_printf("DML::%s: BlockWidth256BytesC = %u\n", __func__, *BlockWidth256BytesC); - dml2_printf("DML::%s: BlockHeight256BytesC = %u\n", __func__, *BlockHeight256BytesC); -#endif + DML_LOG_VERBOSE("DML::%s: BlockWidth256BytesY = %u\n", __func__, *BlockWidth256BytesY); + DML_LOG_VERBOSE("DML::%s: BlockHeight256BytesY = %u\n", __func__, *BlockHeight256BytesY); + DML_LOG_VERBOSE("DML::%s: BlockWidth256BytesC = %u\n", __func__, *BlockWidth256BytesC); + DML_LOG_VERBOSE("DML::%s: BlockHeight256BytesC = %u\n", __func__, *BlockHeight256BytesC); if (dml_get_gfx_version(SurfaceTiling) == 11) { if (SurfaceTiling == dml2_gfx11_sw_linear) { @@ -752,8 +742,8 @@ static void CalculateBytePerPixelAndBlockSizes( } else if (SurfaceTiling == dml2_sw_256kb_2d) { macro_tile_scale = 32; } else { - dml2_printf("ERROR: Invalid SurfaceTiling setting! val=%u\n", SurfaceTiling); - DML2_ASSERT(0); + DML_LOG_VERBOSE("ERROR: Invalid SurfaceTiling setting! val=%u\n", SurfaceTiling); + DML_ASSERT(0); } *MacroTileHeightY = macro_tile_scale * *BlockHeight256BytesY; @@ -766,12 +756,10 @@ static void CalculateBytePerPixelAndBlockSizes( } } -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: MacroTileWidthY = %u\n", __func__, *MacroTileWidthY); - dml2_printf("DML::%s: MacroTileHeightY = %u\n", __func__, *MacroTileHeightY); - dml2_printf("DML::%s: MacroTileWidthC = %u\n", __func__, *MacroTileWidthC); - dml2_printf("DML::%s: MacroTileHeightC = %u\n", __func__, *MacroTileHeightC); -#endif + DML_LOG_VERBOSE("DML::%s: MacroTileWidthY = %u\n", __func__, *MacroTileWidthY); + DML_LOG_VERBOSE("DML::%s: MacroTileHeightY = %u\n", __func__, *MacroTileHeightY); + DML_LOG_VERBOSE("DML::%s: MacroTileWidthC = %u\n", __func__, *MacroTileWidthC); + DML_LOG_VERBOSE("DML::%s: MacroTileHeightC = %u\n", __func__, *MacroTileHeightC); } static void CalculateSinglePipeDPPCLKAndSCLThroughput( @@ -860,10 +848,8 @@ static void CalculateSwathWidth( unsigned int surface_width_ub_c; unsigned int surface_height_ub_c; -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: ForceSingleDPP = %u\n", __func__, ForceSingleDPP); - dml2_printf("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); -#endif + DML_LOG_VERBOSE("DML::%s: ForceSingleDPP = %u\n", __func__, ForceSingleDPP); + DML_LOG_VERBOSE("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { if (!dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle)) { @@ -872,11 +858,9 @@ static void CalculateSwathWidth( SwathWidthSingleDPPY[k] = (unsigned int)display_cfg->plane_descriptors[k].composition.viewport.plane0.height; } -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u ViewportWidth=%u\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.width); - dml2_printf("DML::%s: k=%u ViewportHeight=%u\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.height); - dml2_printf("DML::%s: k=%u DPPPerSurface=%u\n", __func__, k, DPPPerSurface[k]); -#endif + DML_LOG_VERBOSE("DML::%s: k=%u ViewportWidth=%lu\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.width); + DML_LOG_VERBOSE("DML::%s: k=%u ViewportHeight=%lu\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.height); + DML_LOG_VERBOSE("DML::%s: k=%u DPPPerSurface=%u\n", __func__, k, DPPPerSurface[k]); MainSurfaceODMMode = ODMMode[k]; @@ -899,13 +883,11 @@ static void CalculateSwathWidth( } } -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u HActive=%u\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active); - dml2_printf("DML::%s: k=%u HRatio=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio); - dml2_printf("DML::%s: k=%u MainSurfaceODMMode=%u\n", __func__, k, MainSurfaceODMMode); - dml2_printf("DML::%s: k=%u SwathWidthSingleDPPY=%u\n", __func__, k, SwathWidthSingleDPPY[k]); - dml2_printf("DML::%s: k=%u SwathWidthY=%u\n", __func__, k, SwathWidthY[k]); -#endif + DML_LOG_VERBOSE("DML::%s: k=%u HActive=%lu\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active); + DML_LOG_VERBOSE("DML::%s: k=%u HRatio=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u MainSurfaceODMMode=%u\n", __func__, k, MainSurfaceODMMode); + DML_LOG_VERBOSE("DML::%s: k=%u SwathWidthSingleDPPY=%u\n", __func__, k, SwathWidthSingleDPPY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u SwathWidthY=%u\n", __func__, k, SwathWidthY[k]); if (dml_is_420(display_cfg->plane_descriptors[k].pixel_format)) { SwathWidthC[k] = SwathWidthY[k] / 2; @@ -934,22 +916,20 @@ static void CalculateSwathWidth( surface_width_ub_c = (unsigned int)math_ceil2((double)display_cfg->plane_descriptors[k].surface.plane1.width, req_width_horz_c); surface_height_ub_c = (unsigned int)math_ceil2((double)display_cfg->plane_descriptors[k].surface.plane1.height, Read256BytesBlockHeightC[k]); -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u surface_width_ub_l=%u\n", __func__, k, surface_width_ub_l); - dml2_printf("DML::%s: k=%u surface_height_ub_l=%u\n", __func__, k, surface_height_ub_l); - dml2_printf("DML::%s: k=%u surface_width_ub_c=%u\n", __func__, k, surface_width_ub_c); - dml2_printf("DML::%s: k=%u surface_height_ub_c=%u\n", __func__, k, surface_height_ub_c); - dml2_printf("DML::%s: k=%u req_width_horz_y=%u\n", __func__, k, req_width_horz_y); - dml2_printf("DML::%s: k=%u req_width_horz_c=%u\n", __func__, k, req_width_horz_c); - dml2_printf("DML::%s: k=%u Read256BytesBlockWidthY=%u\n", __func__, k, Read256BytesBlockWidthY[k]); - dml2_printf("DML::%s: k=%u Read256BytesBlockHeightY=%u\n", __func__, k, Read256BytesBlockHeightY[k]); - dml2_printf("DML::%s: k=%u Read256BytesBlockWidthC=%u\n", __func__, k, Read256BytesBlockWidthC[k]); - dml2_printf("DML::%s: k=%u Read256BytesBlockHeightC=%u\n", __func__, k, Read256BytesBlockHeightC[k]); - dml2_printf("DML::%s: k=%u req_width_horz_y=%u\n", __func__, k, req_width_horz_y); - dml2_printf("DML::%s: k=%u req_width_horz_c=%u\n", __func__, k, req_width_horz_c); - dml2_printf("DML::%s: k=%u ViewportStationary=%u\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.stationary); - dml2_printf("DML::%s: k=%u DPPPerSurface=%u\n", __func__, k, DPPPerSurface[k]); -#endif + DML_LOG_VERBOSE("DML::%s: k=%u surface_width_ub_l=%u\n", __func__, k, surface_width_ub_l); + DML_LOG_VERBOSE("DML::%s: k=%u surface_height_ub_l=%u\n", __func__, k, surface_height_ub_l); + DML_LOG_VERBOSE("DML::%s: k=%u surface_width_ub_c=%u\n", __func__, k, surface_width_ub_c); + DML_LOG_VERBOSE("DML::%s: k=%u surface_height_ub_c=%u\n", __func__, k, surface_height_ub_c); + DML_LOG_VERBOSE("DML::%s: k=%u req_width_horz_y=%u\n", __func__, k, req_width_horz_y); + DML_LOG_VERBOSE("DML::%s: k=%u req_width_horz_c=%u\n", __func__, k, req_width_horz_c); + DML_LOG_VERBOSE("DML::%s: k=%u Read256BytesBlockWidthY=%u\n", __func__, k, Read256BytesBlockWidthY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u Read256BytesBlockHeightY=%u\n", __func__, k, Read256BytesBlockHeightY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u Read256BytesBlockWidthC=%u\n", __func__, k, Read256BytesBlockWidthC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u Read256BytesBlockHeightC=%u\n", __func__, k, Read256BytesBlockHeightC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u req_width_horz_y=%u\n", __func__, k, req_width_horz_y); + DML_LOG_VERBOSE("DML::%s: k=%u req_width_horz_c=%u\n", __func__, k, req_width_horz_c); + DML_LOG_VERBOSE("DML::%s: k=%u ViewportStationary=%u\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.stationary); + DML_LOG_VERBOSE("DML::%s: k=%u DPPPerSurface=%u\n", __func__, k, DPPPerSurface[k]); req_per_swath_ub_l[k] = 0; req_per_swath_ub_c[k] = 0; @@ -995,15 +975,12 @@ static void CalculateSwathWidth( } } -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u swath_width_luma_ub=%u\n", __func__, k, swath_width_luma_ub[k]); - dml2_printf("DML::%s: k=%u swath_width_chroma_ub=%u\n", __func__, k, swath_width_chroma_ub[k]); - dml2_printf("DML::%s: k=%u MaximumSwathHeightY=%u\n", __func__, k, MaximumSwathHeightY[k]); - dml2_printf("DML::%s: k=%u MaximumSwathHeightC=%u\n", __func__, k, MaximumSwathHeightC[k]); - dml2_printf("DML::%s: k=%u req_per_swath_ub_l=%u\n", __func__, k, req_per_swath_ub_l[k]); - dml2_printf("DML::%s: k=%u req_per_swath_ub_c=%u\n", __func__, k, req_per_swath_ub_c[k]); -#endif - + DML_LOG_VERBOSE("DML::%s: k=%u swath_width_luma_ub=%u\n", __func__, k, swath_width_luma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u swath_width_chroma_ub=%u\n", __func__, k, swath_width_chroma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathHeightY=%u\n", __func__, k, MaximumSwathHeightY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathHeightC=%u\n", __func__, k, MaximumSwathHeightC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u req_per_swath_ub_l=%u\n", __func__, k, req_per_swath_ub_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u req_per_swath_ub_c=%u\n", __func__, k, req_per_swath_ub_c[k]); } } @@ -1018,13 +995,11 @@ static bool UnboundedRequest(bool unb_req_force_en, bool unb_req_force_val, unsi if (unb_req_force_en) { unb_req_en = unb_req_force_val && unb_req_ok; } -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: unb_req_force_en = %u\n", __func__, unb_req_force_en); - dml2_printf("DML::%s: unb_req_force_val = %u\n", __func__, unb_req_force_val); - dml2_printf("DML::%s: unb_req_ok = %u\n", __func__, unb_req_ok); - dml2_printf("DML::%s: unb_req_en = %u\n", __func__, unb_req_en); -#endif - return (unb_req_en); + DML_LOG_VERBOSE("DML::%s: unb_req_force_en = %u\n", __func__, unb_req_force_en); + DML_LOG_VERBOSE("DML::%s: unb_req_force_val = %u\n", __func__, unb_req_force_val); + DML_LOG_VERBOSE("DML::%s: unb_req_ok = %u\n", __func__, unb_req_ok); + DML_LOG_VERBOSE("DML::%s: unb_req_en = %u\n", __func__, unb_req_en); + return unb_req_en; } static void CalculateDETBufferSize( @@ -1054,16 +1029,14 @@ static void CalculateDETBufferSize( bool NextPotentialSurfaceToAssignDETPieceFound; bool MinimizeReallocationSuccess = false; -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: ForceSingleDPP = %u\n", __func__, ForceSingleDPP); - dml2_printf("DML::%s: nomDETInKByte = %u\n", __func__, nomDETInKByte); - dml2_printf("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); - dml2_printf("DML::%s: UnboundedRequestEnabled = %u\n", __func__, UnboundedRequestEnabled); - dml2_printf("DML::%s: MaxTotalDETInKByte = %u\n", __func__, MaxTotalDETInKByte); - dml2_printf("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, ConfigReturnBufferSizeInKByte); - dml2_printf("DML::%s: MinCompressedBufferSizeInKByte = %u\n", __func__, MinCompressedBufferSizeInKByte); - dml2_printf("DML::%s: CompressedBufferSegmentSizeInkByte = %u\n", __func__, CompressedBufferSegmentSizeInkByte); -#endif + DML_LOG_VERBOSE("DML::%s: ForceSingleDPP = %u\n", __func__, ForceSingleDPP); + DML_LOG_VERBOSE("DML::%s: nomDETInKByte = %u\n", __func__, nomDETInKByte); + DML_LOG_VERBOSE("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); + DML_LOG_VERBOSE("DML::%s: UnboundedRequestEnabled = %u\n", __func__, UnboundedRequestEnabled); + DML_LOG_VERBOSE("DML::%s: MaxTotalDETInKByte = %u\n", __func__, MaxTotalDETInKByte); + DML_LOG_VERBOSE("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, ConfigReturnBufferSizeInKByte); + DML_LOG_VERBOSE("DML::%s: MinCompressedBufferSizeInKByte = %u\n", __func__, MinCompressedBufferSizeInKByte); + DML_LOG_VERBOSE("DML::%s: CompressedBufferSegmentSizeInkByte = %u\n", __func__, CompressedBufferSegmentSizeInkByte); // Note: Will use default det size if that fits 2 swaths if (UnboundedRequestEnabled) { @@ -1092,19 +1065,15 @@ static void CalculateDETBufferSize( l->minDET = l->minDET + ConfigReturnBufferSegmentSizeInkByte; } -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u minDET = %u\n", __func__, k, l->minDET); - dml2_printf("DML::%s: k=%u max_minDET = %u\n", __func__, k, l->max_minDET); - dml2_printf("DML::%s: k=%u minDET_pipe = %u\n", __func__, k, l->minDET_pipe); - dml2_printf("DML::%s: k=%u full_swath_bytes_l = %u\n", __func__, k, full_swath_bytes_l[k]); - dml2_printf("DML::%s: k=%u full_swath_bytes_c = %u\n", __func__, k, full_swath_bytes_c[k]); -#endif + DML_LOG_VERBOSE("DML::%s: k=%u minDET = %u\n", __func__, k, l->minDET); + DML_LOG_VERBOSE("DML::%s: k=%u max_minDET = %u\n", __func__, k, l->max_minDET); + DML_LOG_VERBOSE("DML::%s: k=%u minDET_pipe = %u\n", __func__, k, l->minDET_pipe); + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_l = %u\n", __func__, k, full_swath_bytes_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_c = %u\n", __func__, k, full_swath_bytes_c[k]); if (l->minDET_pipe == 0) { l->minDET_pipe = (unsigned int)(math_max2(128, math_ceil2(((double)full_swath_bytes_l[k] + (double)full_swath_bytes_c[k]) / 1024.0, ConfigReturnBufferSegmentSizeInkByte))); -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u minDET_pipe = %u (assume each plane take half DET)\n", __func__, k, l->minDET_pipe); -#endif + DML_LOG_VERBOSE("DML::%s: k=%u minDET_pipe = %u (assume each plane take half DET)\n", __func__, k, l->minDET_pipe); } if (dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { @@ -1117,12 +1086,10 @@ static void CalculateDETBufferSize( l->DETBufferSizePoolInKByte = l->DETBufferSizePoolInKByte - (ForceSingleDPP ? 1 : DPPPerSurface[k]) * l->minDET_pipe; } -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u DPPPerSurface = %u\n", __func__, k, DPPPerSurface[k]); - dml2_printf("DML::%s: k=%u DETSizeOverride = %u\n", __func__, k, display_cfg->plane_descriptors[k].overrides.det_size_override_kb); - dml2_printf("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, DETBufferSizeInKByte[k]); - dml2_printf("DML::%s: DETBufferSizePoolInKByte = %u\n", __func__, l->DETBufferSizePoolInKByte); -#endif + DML_LOG_VERBOSE("DML::%s: k=%u DPPPerSurface = %u\n", __func__, k, DPPPerSurface[k]); + DML_LOG_VERBOSE("DML::%s: k=%u DETSizeOverride = %u\n", __func__, k, display_cfg->plane_descriptors[k].overrides.det_size_override_kb); + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, DETBufferSizeInKByte[k]); + DML_LOG_VERBOSE("DML::%s: DETBufferSizePoolInKByte = %u\n", __func__, l->DETBufferSizePoolInKByte); } if (display_cfg->minimize_det_reallocation) { @@ -1194,14 +1161,12 @@ static void CalculateDETBufferSize( l->TotalBandwidth = l->TotalBandwidth + ReadBandwidthLuma[k] + ReadBandwidthChroma[k]; } } -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: --- Before bandwidth adjustment ---\n", __func__); + DML_LOG_VERBOSE("DML::%s: --- Before bandwidth adjustment ---\n", __func__); for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - dml2_printf("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, DETBufferSizeInKByte[k]); + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, DETBufferSizeInKByte[k]); } - dml2_printf("DML::%s: --- DET allocation with bandwidth ---\n", __func__); -#endif - dml2_printf("DML::%s: TotalBandwidth = %f\n", __func__, l->TotalBandwidth); + DML_LOG_VERBOSE("DML::%s: --- DET allocation with bandwidth ---\n", __func__); + DML_LOG_VERBOSE("DML::%s: TotalBandwidth = %f\n", __func__, l->TotalBandwidth); l->BandwidthOfSurfacesNotAssignedDETPiece = l->TotalBandwidth; for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { @@ -1213,10 +1178,8 @@ static void CalculateDETBufferSize( } else { DETPieceAssignedToThisSurfaceAlready[k] = false; } -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u DETPieceAssignedToThisSurfaceAlready = %u\n", __func__, k, DETPieceAssignedToThisSurfaceAlready[k]); - dml2_printf("DML::%s: k=%u BandwidthOfSurfacesNotAssignedDETPiece = %f\n", __func__, k, l->BandwidthOfSurfacesNotAssignedDETPiece); -#endif + DML_LOG_VERBOSE("DML::%s: k=%u DETPieceAssignedToThisSurfaceAlready = %u\n", __func__, k, DETPieceAssignedToThisSurfaceAlready[k]); + DML_LOG_VERBOSE("DML::%s: k=%u BandwidthOfSurfacesNotAssignedDETPiece = %f\n", __func__, k, l->BandwidthOfSurfacesNotAssignedDETPiece); } for (unsigned int j = 0; j < NumberOfActiveSurfaces; ++j) { @@ -1224,22 +1187,18 @@ static void CalculateDETBufferSize( l->NextSurfaceToAssignDETPiece = 0; for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: j=%u k=%u, ReadBandwidthLuma[k] = %f\n", __func__, j, k, ReadBandwidthLuma[k]); - dml2_printf("DML::%s: j=%u k=%u, ReadBandwidthChroma[k] = %f\n", __func__, j, k, ReadBandwidthChroma[k]); - dml2_printf("DML::%s: j=%u k=%u, ReadBandwidthLuma[Next] = %f\n", __func__, j, k, ReadBandwidthLuma[l->NextSurfaceToAssignDETPiece]); - dml2_printf("DML::%s: j=%u k=%u, ReadBandwidthChroma[Next] = %f\n", __func__, j, k, ReadBandwidthChroma[l->NextSurfaceToAssignDETPiece]); - dml2_printf("DML::%s: j=%u k=%u, NextSurfaceToAssignDETPiece = %u\n", __func__, j, k, l->NextSurfaceToAssignDETPiece); -#endif + DML_LOG_VERBOSE("DML::%s: j=%u k=%u, ReadBandwidthLuma[k] = %f\n", __func__, j, k, ReadBandwidthLuma[k]); + DML_LOG_VERBOSE("DML::%s: j=%u k=%u, ReadBandwidthChroma[k] = %f\n", __func__, j, k, ReadBandwidthChroma[k]); + DML_LOG_VERBOSE("DML::%s: j=%u k=%u, ReadBandwidthLuma[Next] = %f\n", __func__, j, k, ReadBandwidthLuma[l->NextSurfaceToAssignDETPiece]); + DML_LOG_VERBOSE("DML::%s: j=%u k=%u, ReadBandwidthChroma[Next] = %f\n", __func__, j, k, ReadBandwidthChroma[l->NextSurfaceToAssignDETPiece]); + DML_LOG_VERBOSE("DML::%s: j=%u k=%u, NextSurfaceToAssignDETPiece = %u\n", __func__, j, k, l->NextSurfaceToAssignDETPiece); if (!DETPieceAssignedToThisSurfaceAlready[k] && (!NextPotentialSurfaceToAssignDETPieceFound || ReadBandwidthLuma[k] + ReadBandwidthChroma[k] < ReadBandwidthLuma[l->NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[l->NextSurfaceToAssignDETPiece])) { l->NextSurfaceToAssignDETPiece = k; NextPotentialSurfaceToAssignDETPieceFound = true; } -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: j=%u k=%u, DETPieceAssignedToThisSurfaceAlready = %u\n", __func__, j, k, DETPieceAssignedToThisSurfaceAlready[k]); - dml2_printf("DML::%s: j=%u k=%u, NextPotentialSurfaceToAssignDETPieceFound = %u\n", __func__, j, k, NextPotentialSurfaceToAssignDETPieceFound); -#endif + DML_LOG_VERBOSE("DML::%s: j=%u k=%u, DETPieceAssignedToThisSurfaceAlready = %u\n", __func__, j, k, DETPieceAssignedToThisSurfaceAlready[k]); + DML_LOG_VERBOSE("DML::%s: j=%u k=%u, NextPotentialSurfaceToAssignDETPieceFound = %u\n", __func__, j, k, NextPotentialSurfaceToAssignDETPieceFound); } if (NextPotentialSurfaceToAssignDETPieceFound) { @@ -1249,20 +1208,16 @@ static void CalculateDETBufferSize( * (ForceSingleDPP ? 1 : DPPPerSurface[l->NextSurfaceToAssignDETPiece]) * ConfigReturnBufferSegmentSizeInkByte, math_floor2((double)l->DETBufferSizePoolInKByte, (ForceSingleDPP ? 1 : DPPPerSurface[l->NextSurfaceToAssignDETPiece]) * ConfigReturnBufferSegmentSizeInkByte))); -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: j=%u, DETBufferSizePoolInKByte = %u\n", __func__, j, l->DETBufferSizePoolInKByte); - dml2_printf("DML::%s: j=%u, NextSurfaceToAssignDETPiece = %u\n", __func__, j, l->NextSurfaceToAssignDETPiece); - dml2_printf("DML::%s: j=%u, ReadBandwidthLuma[%u] = %f\n", __func__, j, l->NextSurfaceToAssignDETPiece, ReadBandwidthLuma[l->NextSurfaceToAssignDETPiece]); - dml2_printf("DML::%s: j=%u, ReadBandwidthChroma[%u] = %f\n", __func__, j, l->NextSurfaceToAssignDETPiece, ReadBandwidthChroma[l->NextSurfaceToAssignDETPiece]); - dml2_printf("DML::%s: j=%u, BandwidthOfSurfacesNotAssignedDETPiece = %f\n", __func__, j, l->BandwidthOfSurfacesNotAssignedDETPiece); - dml2_printf("DML::%s: j=%u, NextDETBufferPieceInKByte = %u\n", __func__, j, l->NextDETBufferPieceInKByte); - dml2_printf("DML::%s: j=%u, DETBufferSizeInKByte[%u] increases from %u ", __func__, j, l->NextSurfaceToAssignDETPiece, DETBufferSizeInKByte[l->NextSurfaceToAssignDETPiece]); -#endif + DML_LOG_VERBOSE("DML::%s: j=%u, DETBufferSizePoolInKByte = %u\n", __func__, j, l->DETBufferSizePoolInKByte); + DML_LOG_VERBOSE("DML::%s: j=%u, NextSurfaceToAssignDETPiece = %u\n", __func__, j, l->NextSurfaceToAssignDETPiece); + DML_LOG_VERBOSE("DML::%s: j=%u, ReadBandwidthLuma[%u] = %f\n", __func__, j, l->NextSurfaceToAssignDETPiece, ReadBandwidthLuma[l->NextSurfaceToAssignDETPiece]); + DML_LOG_VERBOSE("DML::%s: j=%u, ReadBandwidthChroma[%u] = %f\n", __func__, j, l->NextSurfaceToAssignDETPiece, ReadBandwidthChroma[l->NextSurfaceToAssignDETPiece]); + DML_LOG_VERBOSE("DML::%s: j=%u, BandwidthOfSurfacesNotAssignedDETPiece = %f\n", __func__, j, l->BandwidthOfSurfacesNotAssignedDETPiece); + DML_LOG_VERBOSE("DML::%s: j=%u, NextDETBufferPieceInKByte = %u\n", __func__, j, l->NextDETBufferPieceInKByte); + DML_LOG_VERBOSE("DML::%s: j=%u, DETBufferSizeInKByte[%u] increases from %u ", __func__, j, l->NextSurfaceToAssignDETPiece, DETBufferSizeInKByte[l->NextSurfaceToAssignDETPiece]); DETBufferSizeInKByte[l->NextSurfaceToAssignDETPiece] = DETBufferSizeInKByte[l->NextSurfaceToAssignDETPiece] + l->NextDETBufferPieceInKByte / (ForceSingleDPP ? 1 : DPPPerSurface[l->NextSurfaceToAssignDETPiece]); -#ifdef __DML_VBA_DEBUG__ - dml2_printf("to %u\n", DETBufferSizeInKByte[l->NextSurfaceToAssignDETPiece]); -#endif + DML_LOG_VERBOSE("to %u\n", DETBufferSizeInKByte[l->NextSurfaceToAssignDETPiece]); l->DETBufferSizePoolInKByte = l->DETBufferSizePoolInKByte - l->NextDETBufferPieceInKByte; DETPieceAssignedToThisSurfaceAlready[l->NextSurfaceToAssignDETPiece] = true; @@ -1274,13 +1229,11 @@ static void CalculateDETBufferSize( } *CompressedBufferSizeInkByte = *CompressedBufferSizeInkByte * CompressedBufferSegmentSizeInkByte / ConfigReturnBufferSegmentSizeInkByte; -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: --- After bandwidth adjustment ---\n", __func__); - dml2_printf("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, *CompressedBufferSizeInkByte); + DML_LOG_VERBOSE("DML::%s: --- After bandwidth adjustment ---\n", __func__); + DML_LOG_VERBOSE("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, *CompressedBufferSizeInkByte); for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - dml2_printf("DML::%s: k=%u DETBufferSizeInKByte = %u (TotalReadBandWidth=%f)\n", __func__, k, DETBufferSizeInKByte[k], ReadBandwidthLuma[k] + ReadBandwidthChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByte = %u (TotalReadBandWidth=%f)\n", __func__, k, DETBufferSizeInKByte[k], ReadBandwidthLuma[k] + ReadBandwidthChroma[k]); } -#endif } static double CalculateRequiredDispclk( @@ -1510,15 +1463,13 @@ static unsigned int dscceComputeDelay( //pixel delay is group_delay (converted to pixels) + pipeline, however, first group is a special case since it is processed as soon as it arrives (i.e., in 3 cycles regardless of pixel format) pixels = (group_delay - 1) * cycles_per_group + 3 + pipeline_delay; -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: bpc: %u\n", __func__, bpc); - dml2_printf("DML::%s: BPP: %f\n", __func__, BPP); - dml2_printf("DML::%s: sliceWidth: %u\n", __func__, sliceWidth); - dml2_printf("DML::%s: numSlices: %u\n", __func__, numSlices); - dml2_printf("DML::%s: pixelFormat: %u\n", __func__, pixelFormat); - dml2_printf("DML::%s: Output: %u\n", __func__, Output); - dml2_printf("DML::%s: pixels: %u\n", __func__, pixels); -#endif + DML_LOG_VERBOSE("DML::%s: bpc: %u\n", __func__, bpc); + DML_LOG_VERBOSE("DML::%s: BPP: %f\n", __func__, BPP); + DML_LOG_VERBOSE("DML::%s: sliceWidth: %u\n", __func__, sliceWidth); + DML_LOG_VERBOSE("DML::%s: numSlices: %u\n", __func__, numSlices); + DML_LOG_VERBOSE("DML::%s: pixelFormat: %u\n", __func__, pixelFormat); + DML_LOG_VERBOSE("DML::%s: Output: %u\n", __func__, Output); + DML_LOG_VERBOSE("DML::%s: pixels: %u\n", __func__, pixels); return pixels; } @@ -1593,10 +1544,8 @@ static unsigned int dscComputeDelay(enum dml2_output_format_class pixelFormat, e // sft Delay = Delay + 1; -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: pixelFormat = %u\n", __func__, pixelFormat); - dml2_printf("DML::%s: Delay = %u\n", __func__, Delay); -#endif + DML_LOG_VERBOSE("DML::%s: pixelFormat = %u\n", __func__, pixelFormat); + DML_LOG_VERBOSE("DML::%s: Delay = %u\n", __func__, Delay); return Delay; } @@ -1667,10 +1616,8 @@ static unsigned int CalculateVMAndRowBytes(struct dml2_core_shared_calculate_vm_ } meta_surface_bytes = (unsigned int)(p->DCCMetaPitch * vp_height_meta_ub * p->BytePerPixel / 256.0); -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: DCCMetaPitch = %u\n", __func__, p->DCCMetaPitch); - dml2_printf("DML::%s: meta_surface_bytes = %u\n", __func__, meta_surface_bytes); -#endif + DML_LOG_VERBOSE("DML::%s: DCCMetaPitch = %u\n", __func__, p->DCCMetaPitch); + DML_LOG_VERBOSE("DML::%s: meta_surface_bytes = %u\n", __func__, meta_surface_bytes); if (p->GPUVMEnable == true) { double meta_vmpg_bytes = 4.0 * 1024.0; *p->meta_pte_bytes_per_frame_ub = (unsigned int)((math_ceil2((double) (meta_surface_bytes - meta_vmpg_bytes) / (8 * meta_vmpg_bytes), 1) + 1) * 64); @@ -1724,25 +1671,23 @@ static unsigned int CalculateVMAndRowBytes(struct dml2_core_shared_calculate_vm_ vm_bytes = *p->meta_pte_bytes_per_frame_ub + extra_mpde_bytes + *p->dpde0_bytes_per_frame_ub + extra_dpde_bytes; -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: DCCEnable = %u\n", __func__, p->DCCEnable); - dml2_printf("DML::%s: GPUVMEnable = %u\n", __func__, p->GPUVMEnable); - dml2_printf("DML::%s: SwModeLinear = %u\n", __func__, p->SurfaceTiling == dml2_sw_linear); - dml2_printf("DML::%s: BytePerPixel = %u\n", __func__, p->BytePerPixel); - dml2_printf("DML::%s: GPUVMMaxPageTableLevels = %u\n", __func__, p->GPUVMMaxPageTableLevels); - dml2_printf("DML::%s: BlockHeight256Bytes = %u\n", __func__, p->BlockHeight256Bytes); - dml2_printf("DML::%s: BlockWidth256Bytes = %u\n", __func__, p->BlockWidth256Bytes); - dml2_printf("DML::%s: MacroTileHeight = %u\n", __func__, p->MacroTileHeight); - dml2_printf("DML::%s: MacroTileWidth = %u\n", __func__, p->MacroTileWidth); - dml2_printf("DML::%s: meta_pte_bytes_per_frame_ub = %u\n", __func__, *p->meta_pte_bytes_per_frame_ub); - dml2_printf("DML::%s: dpde0_bytes_per_frame_ub = %u\n", __func__, *p->dpde0_bytes_per_frame_ub); - dml2_printf("DML::%s: extra_mpde_bytes = %u\n", __func__, extra_mpde_bytes); - dml2_printf("DML::%s: extra_dpde_bytes = %u\n", __func__, extra_dpde_bytes); - dml2_printf("DML::%s: vm_bytes = %u\n", __func__, vm_bytes); - dml2_printf("DML::%s: ViewportHeight = %u\n", __func__, p->ViewportHeight); - dml2_printf("DML::%s: SwathWidth = %u\n", __func__, p->SwathWidth); - dml2_printf("DML::%s: vp_height_dpte_ub = %u\n", __func__, vp_height_dpte_ub); -#endif + DML_LOG_VERBOSE("DML::%s: DCCEnable = %u\n", __func__, p->DCCEnable); + DML_LOG_VERBOSE("DML::%s: GPUVMEnable = %u\n", __func__, p->GPUVMEnable); + DML_LOG_VERBOSE("DML::%s: SwModeLinear = %u\n", __func__, p->SurfaceTiling == dml2_sw_linear); + DML_LOG_VERBOSE("DML::%s: BytePerPixel = %u\n", __func__, p->BytePerPixel); + DML_LOG_VERBOSE("DML::%s: GPUVMMaxPageTableLevels = %u\n", __func__, p->GPUVMMaxPageTableLevels); + DML_LOG_VERBOSE("DML::%s: BlockHeight256Bytes = %u\n", __func__, p->BlockHeight256Bytes); + DML_LOG_VERBOSE("DML::%s: BlockWidth256Bytes = %u\n", __func__, p->BlockWidth256Bytes); + DML_LOG_VERBOSE("DML::%s: MacroTileHeight = %u\n", __func__, p->MacroTileHeight); + DML_LOG_VERBOSE("DML::%s: MacroTileWidth = %u\n", __func__, p->MacroTileWidth); + DML_LOG_VERBOSE("DML::%s: meta_pte_bytes_per_frame_ub = %u\n", __func__, *p->meta_pte_bytes_per_frame_ub); + DML_LOG_VERBOSE("DML::%s: dpde0_bytes_per_frame_ub = %u\n", __func__, *p->dpde0_bytes_per_frame_ub); + DML_LOG_VERBOSE("DML::%s: extra_mpde_bytes = %u\n", __func__, extra_mpde_bytes); + DML_LOG_VERBOSE("DML::%s: extra_dpde_bytes = %u\n", __func__, extra_dpde_bytes); + DML_LOG_VERBOSE("DML::%s: vm_bytes = %u\n", __func__, vm_bytes); + DML_LOG_VERBOSE("DML::%s: ViewportHeight = %u\n", __func__, p->ViewportHeight); + DML_LOG_VERBOSE("DML::%s: SwathWidth = %u\n", __func__, p->SwathWidth); + DML_LOG_VERBOSE("DML::%s: vp_height_dpte_ub = %u\n", __func__, vp_height_dpte_ub); if (p->SurfaceTiling == dml2_sw_linear) { *p->PixelPTEReqHeight = 1; @@ -1778,22 +1723,20 @@ static unsigned int CalculateVMAndRowBytes(struct dml2_core_shared_calculate_vm_ *p->vmpg_width = 1024 * p->GPUVMMinPageSizeKBytes / (p->MacroTileHeight * p->BytePerPixel); if (p->GPUVMEnable == true) { - dml2_printf("DML::%s: GPUVMMinPageSizeKBytes=%u and sw_mode=%u (tile_size=%d) not supported!\n", + DML_LOG_VERBOSE("DML::%s: GPUVMMinPageSizeKBytes=%u and sw_mode=%u (tile_size=%d) not supported!\n", __func__, p->GPUVMMinPageSizeKBytes, p->SurfaceTiling, dml_get_tile_block_size_bytes(p->SurfaceTiling)); - DML2_ASSERT(0); + DML_ASSERT(0); } } -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: GPUVMMinPageSizeKBytes = %u\n", __func__, p->GPUVMMinPageSizeKBytes); - dml2_printf("DML::%s: PixelPTEReqHeight = %u\n", __func__, *p->PixelPTEReqHeight); - dml2_printf("DML::%s: PixelPTEReqWidth = %u\n", __func__, *p->PixelPTEReqWidth); - dml2_printf("DML::%s: PixelPTEReqWidth_linear = %u\n", __func__, PixelPTEReqWidth_linear); - dml2_printf("DML::%s: PTERequestSize = %u\n", __func__, *p->PTERequestSize); - dml2_printf("DML::%s: Pitch = %u\n", __func__, p->Pitch); - dml2_printf("DML::%s: vmpg_width = %u\n", __func__, *p->vmpg_width); - dml2_printf("DML::%s: vmpg_height = %u\n", __func__, *p->vmpg_height); -#endif + DML_LOG_VERBOSE("DML::%s: GPUVMMinPageSizeKBytes = %u\n", __func__, p->GPUVMMinPageSizeKBytes); + DML_LOG_VERBOSE("DML::%s: PixelPTEReqHeight = %u\n", __func__, *p->PixelPTEReqHeight); + DML_LOG_VERBOSE("DML::%s: PixelPTEReqWidth = %u\n", __func__, *p->PixelPTEReqWidth); + DML_LOG_VERBOSE("DML::%s: PixelPTEReqWidth_linear = %u\n", __func__, PixelPTEReqWidth_linear); + DML_LOG_VERBOSE("DML::%s: PTERequestSize = %u\n", __func__, *p->PTERequestSize); + DML_LOG_VERBOSE("DML::%s: Pitch = %u\n", __func__, p->Pitch); + DML_LOG_VERBOSE("DML::%s: vmpg_width = %u\n", __func__, *p->vmpg_width); + DML_LOG_VERBOSE("DML::%s: vmpg_height = %u\n", __func__, *p->vmpg_height); *p->dpte_row_height_one_row_per_frame = vp_height_dpte_ub; *p->dpte_row_width_ub_one_row_per_frame = (unsigned int)((math_ceil2(((double)p->Pitch * (double)*p->dpte_row_height_one_row_per_frame / (double)*p->PixelPTEReqHeight - 1) / (double)*p->PixelPTEReqWidth, 1) + 1) * (double)*p->PixelPTEReqWidth); @@ -1811,7 +1754,7 @@ static unsigned int CalculateVMAndRowBytes(struct dml2_core_shared_calculate_vm_ *p->dpte_row_height_linear = 128; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: dpte_row_width_ub = %u (linear)\n", __func__, *p->dpte_row_width_ub); + DML_LOG_VERBOSE("DML::%s: dpte_row_width_ub = %u (linear)\n", __func__, *p->dpte_row_width_ub); #endif } else if (!dml_is_vertical_rotation(p->RotationAngle)) { @@ -1825,7 +1768,7 @@ static unsigned int CalculateVMAndRowBytes(struct dml2_core_shared_calculate_vm_ *p->dpte_row_width_ub = (unsigned int)((math_ceil2((double)(p->SwathWidth - 1) / (double)*p->PixelPTEReqWidth, 1) + 1.0) * *p->PixelPTEReqWidth); } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: dpte_row_width_ub = %u (tiled horz)\n", __func__, *p->dpte_row_width_ub); + DML_LOG_VERBOSE("DML::%s: dpte_row_width_ub = %u (tiled horz)\n", __func__, *p->dpte_row_width_ub); #endif *p->PixelPTEBytesPerRow = *p->dpte_row_width_ub / *p->PixelPTEReqWidth * *p->PTERequestSize; @@ -1840,7 +1783,7 @@ static unsigned int CalculateVMAndRowBytes(struct dml2_core_shared_calculate_vm_ *p->PixelPTEBytesPerRow = (unsigned int)((double)*p->dpte_row_width_ub / (double)*p->PixelPTEReqHeight * *p->PTERequestSize); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: dpte_row_width_ub = %u (tiled vert)\n", __func__, *p->dpte_row_width_ub); + DML_LOG_VERBOSE("DML::%s: dpte_row_width_ub = %u (tiled vert)\n", __func__, *p->dpte_row_width_ub); #endif } @@ -1852,18 +1795,18 @@ static unsigned int CalculateVMAndRowBytes(struct dml2_core_shared_calculate_vm_ *p->PixelPTEBytesPerRowStorage = *p->PixelPTEBytesPerRow; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: GPUVMMinPageSizeKBytes = %u\n", __func__, p->GPUVMMinPageSizeKBytes); - dml2_printf("DML::%s: GPUVMEnable = %u\n", __func__, p->GPUVMEnable); - dml2_printf("DML::%s: meta_row_height = %u\n", __func__, *p->meta_row_height); - dml2_printf("DML::%s: dpte_row_height = %u\n", __func__, *p->dpte_row_height); - dml2_printf("DML::%s: dpte_row_height_linear = %u\n", __func__, *p->dpte_row_height_linear); - dml2_printf("DML::%s: dpte_row_width_ub = %u\n", __func__, *p->dpte_row_width_ub); - dml2_printf("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, *p->PixelPTEBytesPerRow); - dml2_printf("DML::%s: PixelPTEBytesPerRowStorage = %u\n", __func__, *p->PixelPTEBytesPerRowStorage); - dml2_printf("DML::%s: PTEBufferSizeInRequests = %u\n", __func__, p->PTEBufferSizeInRequests); - dml2_printf("DML::%s: dpte_row_height_one_row_per_frame = %u\n", __func__, *p->dpte_row_height_one_row_per_frame); - dml2_printf("DML::%s: dpte_row_width_ub_one_row_per_frame = %u\n", __func__, *p->dpte_row_width_ub_one_row_per_frame); - dml2_printf("DML::%s: PixelPTEBytesPerRow_one_row_per_frame = %u\n", __func__, *p->PixelPTEBytesPerRow_one_row_per_frame); + DML_LOG_VERBOSE("DML::%s: GPUVMMinPageSizeKBytes = %u\n", __func__, p->GPUVMMinPageSizeKBytes); + DML_LOG_VERBOSE("DML::%s: GPUVMEnable = %u\n", __func__, p->GPUVMEnable); + DML_LOG_VERBOSE("DML::%s: meta_row_height = %u\n", __func__, *p->meta_row_height); + DML_LOG_VERBOSE("DML::%s: dpte_row_height = %u\n", __func__, *p->dpte_row_height); + DML_LOG_VERBOSE("DML::%s: dpte_row_height_linear = %u\n", __func__, *p->dpte_row_height_linear); + DML_LOG_VERBOSE("DML::%s: dpte_row_width_ub = %u\n", __func__, *p->dpte_row_width_ub); + DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, *p->PixelPTEBytesPerRow); + DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRowStorage = %u\n", __func__, *p->PixelPTEBytesPerRowStorage); + DML_LOG_VERBOSE("DML::%s: PTEBufferSizeInRequests = %u\n", __func__, p->PTEBufferSizeInRequests); + DML_LOG_VERBOSE("DML::%s: dpte_row_height_one_row_per_frame = %u\n", __func__, *p->dpte_row_height_one_row_per_frame); + DML_LOG_VERBOSE("DML::%s: dpte_row_width_ub_one_row_per_frame = %u\n", __func__, *p->dpte_row_width_ub_one_row_per_frame); + DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRow_one_row_per_frame = %u\n", __func__, *p->PixelPTEBytesPerRow_one_row_per_frame); #endif return vm_bytes; @@ -1894,12 +1837,12 @@ static unsigned int CalculatePrefetchSourceLines( double numLines = 0; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: VRatio = %f\n", __func__, VRatio); - dml2_printf("DML::%s: VTaps = %u\n", __func__, VTaps); - dml2_printf("DML::%s: ViewportXStart = %u\n", __func__, ViewportXStart); - dml2_printf("DML::%s: ViewportYStart = %u\n", __func__, ViewportYStart); - dml2_printf("DML::%s: ViewportStationary = %u\n", __func__, ViewportStationary); - dml2_printf("DML::%s: SwathHeight = %u\n", __func__, SwathHeight); + DML_LOG_VERBOSE("DML::%s: VRatio = %f\n", __func__, VRatio); + DML_LOG_VERBOSE("DML::%s: VTaps = %u\n", __func__, VTaps); + DML_LOG_VERBOSE("DML::%s: ViewportXStart = %u\n", __func__, ViewportXStart); + DML_LOG_VERBOSE("DML::%s: ViewportYStart = %u\n", __func__, ViewportYStart); + DML_LOG_VERBOSE("DML::%s: ViewportStationary = %u\n", __func__, ViewportStationary); + DML_LOG_VERBOSE("DML::%s: SwathHeight = %u\n", __func__, SwathHeight); #endif if (ProgressiveToInterlaceUnitInOPP) *VInitPreFill = (unsigned int)(math_floor2((VRatio + (double)VTaps + 1) / 2.0, 1)); @@ -1934,11 +1877,11 @@ static unsigned int CalculatePrefetchSourceLines( numLines = *MaxNumSwath * SwathHeight + MaxPartialSwath; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: vp_start_rot = %u\n", __func__, vp_start_rot); - dml2_printf("DML::%s: VInitPreFill = %u\n", __func__, *VInitPreFill); - dml2_printf("DML::%s: MaxPartialSwath = %u\n", __func__, MaxPartialSwath); - dml2_printf("DML::%s: MaxNumSwath = %u\n", __func__, *MaxNumSwath); - dml2_printf("DML::%s: Prefetch source lines = %3.2f\n", __func__, numLines); + DML_LOG_VERBOSE("DML::%s: vp_start_rot = %u\n", __func__, vp_start_rot); + DML_LOG_VERBOSE("DML::%s: VInitPreFill = %u\n", __func__, *VInitPreFill); + DML_LOG_VERBOSE("DML::%s: MaxPartialSwath = %u\n", __func__, MaxPartialSwath); + DML_LOG_VERBOSE("DML::%s: MaxNumSwath = %u\n", __func__, *MaxNumSwath); + DML_LOG_VERBOSE("DML::%s: Prefetch source lines = %3.2f\n", __func__, numLines); #endif return (unsigned int)(numLines); @@ -2007,8 +1950,8 @@ static void CalculateMALLUseForStaticScreen( if (is_using_mall_for_ss[k]) TotalSurfaceSizeInMALL = TotalSurfaceSizeInMALL + SurfaceSizeInMALL[k]; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, is_using_mall_for_ss = %u\n", __func__, k, is_using_mall_for_ss[k]); - dml2_printf("DML::%s: k=%u, TotalSurfaceSizeInMALL = %u\n", __func__, k, TotalSurfaceSizeInMALL); + DML_LOG_VERBOSE("DML::%s: k=%u, is_using_mall_for_ss = %u\n", __func__, k, is_using_mall_for_ss[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, TotalSurfaceSizeInMALL = %u\n", __func__, k, TotalSurfaceSizeInMALL); #endif } @@ -2022,7 +1965,7 @@ static void CalculateMALLUseForStaticScreen( (!CanAddAnotherSurfaceToMALL || SurfaceSizeInMALL[k] < SurfaceSizeInMALL[SurfaceToAddToMALL])) { CanAddAnotherSurfaceToMALL = true; SurfaceToAddToMALL = k; - dml2_printf("DML::%s: k=%u, UseMALLForStaticScreen = %u (dis, en, optimize)\n", __func__, k, display_cfg->plane_descriptors[k].overrides.refresh_from_mall); + DML_LOG_VERBOSE("DML::%s: k=%u, UseMALLForStaticScreen = %u (dis, en, optimize)\n", __func__, k, display_cfg->plane_descriptors[k].overrides.refresh_from_mall); } } if (CanAddAnotherSurfaceToMALL) { @@ -2030,8 +1973,8 @@ static void CalculateMALLUseForStaticScreen( TotalSurfaceSizeInMALL = TotalSurfaceSizeInMALL + SurfaceSizeInMALL[SurfaceToAddToMALL]; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: SurfaceToAddToMALL = %u\n", __func__, SurfaceToAddToMALL); - dml2_printf("DML::%s: TotalSurfaceSizeInMALL = %u\n", __func__, TotalSurfaceSizeInMALL); + DML_LOG_VERBOSE("DML::%s: SurfaceToAddToMALL = %u\n", __func__, SurfaceToAddToMALL); + DML_LOG_VERBOSE("DML::%s: TotalSurfaceSizeInMALL = %u\n", __func__, TotalSurfaceSizeInMALL); #endif } } @@ -2203,15 +2146,15 @@ static void CalculateDCCConfiguration( segment_order_vert_contiguous_chroma = 0; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: DCCEnabled = %u\n", __func__, DCCEnabled); - dml2_printf("DML::%s: nomDETInKByte = %u\n", __func__, nomDETInKByte); - dml2_printf("DML::%s: DETBufferSizeForDCC = %u\n", __func__, DETBufferSizeForDCC); - dml2_printf("DML::%s: req128_horz_wc_l = %u\n", __func__, req128_horz_wc_l); - dml2_printf("DML::%s: req128_horz_wc_c = %u\n", __func__, req128_horz_wc_c); - dml2_printf("DML::%s: full_swath_bytes_horz_wc_l = %u\n", __func__, full_swath_bytes_horz_wc_l); - dml2_printf("DML::%s: full_swath_bytes_vert_wc_c = %u\n", __func__, full_swath_bytes_vert_wc_c); - dml2_printf("DML::%s: segment_order_horz_contiguous_luma = %u\n", __func__, segment_order_horz_contiguous_luma); - dml2_printf("DML::%s: segment_order_horz_contiguous_chroma = %u\n", __func__, segment_order_horz_contiguous_chroma); + DML_LOG_VERBOSE("DML::%s: DCCEnabled = %u\n", __func__, DCCEnabled); + DML_LOG_VERBOSE("DML::%s: nomDETInKByte = %u\n", __func__, nomDETInKByte); + DML_LOG_VERBOSE("DML::%s: DETBufferSizeForDCC = %u\n", __func__, DETBufferSizeForDCC); + DML_LOG_VERBOSE("DML::%s: req128_horz_wc_l = %u\n", __func__, req128_horz_wc_l); + DML_LOG_VERBOSE("DML::%s: req128_horz_wc_c = %u\n", __func__, req128_horz_wc_c); + DML_LOG_VERBOSE("DML::%s: full_swath_bytes_horz_wc_l = %u\n", __func__, full_swath_bytes_horz_wc_l); + DML_LOG_VERBOSE("DML::%s: full_swath_bytes_vert_wc_c = %u\n", __func__, full_swath_bytes_vert_wc_c); + DML_LOG_VERBOSE("DML::%s: segment_order_horz_contiguous_luma = %u\n", __func__, segment_order_horz_contiguous_luma); + DML_LOG_VERBOSE("DML::%s: segment_order_horz_contiguous_chroma = %u\n", __func__, segment_order_horz_contiguous_chroma); #endif if (DCCProgrammingAssumesScanDirectionUnknown == true) { if (req128_horz_wc_l == 0 && req128_vert_wc_l == 0) { @@ -2301,12 +2244,12 @@ static void CalculateDCCConfiguration( } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: MaxUncompressedBlockLuma = %u\n", __func__, *MaxUncompressedBlockLuma); - dml2_printf("DML::%s: MaxCompressedBlockLuma = %u\n", __func__, *MaxCompressedBlockLuma); - dml2_printf("DML::%s: IndependentBlockLuma = %u\n", __func__, *IndependentBlockLuma); - dml2_printf("DML::%s: MaxUncompressedBlockChroma = %u\n", __func__, *MaxUncompressedBlockChroma); - dml2_printf("DML::%s: MaxCompressedBlockChroma = %u\n", __func__, *MaxCompressedBlockChroma); - dml2_printf("DML::%s: IndependentBlockChroma = %u\n", __func__, *IndependentBlockChroma); + DML_LOG_VERBOSE("DML::%s: MaxUncompressedBlockLuma = %u\n", __func__, *MaxUncompressedBlockLuma); + DML_LOG_VERBOSE("DML::%s: MaxCompressedBlockLuma = %u\n", __func__, *MaxCompressedBlockLuma); + DML_LOG_VERBOSE("DML::%s: IndependentBlockLuma = %u\n", __func__, *IndependentBlockLuma); + DML_LOG_VERBOSE("DML::%s: MaxUncompressedBlockChroma = %u\n", __func__, *MaxUncompressedBlockChroma); + DML_LOG_VERBOSE("DML::%s: MaxCompressedBlockChroma = %u\n", __func__, *MaxCompressedBlockChroma); + DML_LOG_VERBOSE("DML::%s: IndependentBlockChroma = %u\n", __func__, *IndependentBlockChroma); #endif } @@ -2326,26 +2269,26 @@ static void calculate_mcache_row_bytes( unsigned int mvmpg_per_mcache; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: num_chans = %u\n", __func__, p->num_chans); - dml2_printf("DML::%s: mem_word_bytes = %u\n", __func__, p->mem_word_bytes); - dml2_printf("DML::%s: mcache_line_size_bytes = %u\n", __func__, p->mcache_line_size_bytes); - dml2_printf("DML::%s: mcache_size_bytes = %u\n", __func__, p->mcache_size_bytes); - dml2_printf("DML::%s: gpuvm_enable = %u\n", __func__, p->gpuvm_enable); - dml2_printf("DML::%s: gpuvm_page_size_kbytes = %u\n", __func__, p->gpuvm_page_size_kbytes); - dml2_printf("DML::%s: vp_stationary = %u\n", __func__, p->vp_stationary); - dml2_printf("DML::%s: tiling_mode = %u\n", __func__, p->tiling_mode); - dml2_printf("DML::%s: vp_start_x = %u\n", __func__, p->vp_start_x); - dml2_printf("DML::%s: vp_start_y = %u\n", __func__, p->vp_start_y); - dml2_printf("DML::%s: full_vp_width = %u\n", __func__, p->full_vp_width); - dml2_printf("DML::%s: full_vp_height = %u\n", __func__, p->full_vp_height); - dml2_printf("DML::%s: blk_width = %u\n", __func__, p->blk_width); - dml2_printf("DML::%s: blk_height = %u\n", __func__, p->blk_height); - dml2_printf("DML::%s: vmpg_width = %u\n", __func__, p->vmpg_width); - dml2_printf("DML::%s: vmpg_height = %u\n", __func__, p->vmpg_height); - dml2_printf("DML::%s: full_swath_bytes = %u\n", __func__, p->full_swath_bytes); -#endif - DML2_ASSERT(p->mcache_line_size_bytes != 0); - DML2_ASSERT(p->mcache_size_bytes != 0); + DML_LOG_VERBOSE("DML::%s: num_chans = %u\n", __func__, p->num_chans); + DML_LOG_VERBOSE("DML::%s: mem_word_bytes = %u\n", __func__, p->mem_word_bytes); + DML_LOG_VERBOSE("DML::%s: mcache_line_size_bytes = %u\n", __func__, p->mcache_line_size_bytes); + DML_LOG_VERBOSE("DML::%s: mcache_size_bytes = %u\n", __func__, p->mcache_size_bytes); + DML_LOG_VERBOSE("DML::%s: gpuvm_enable = %u\n", __func__, p->gpuvm_enable); + DML_LOG_VERBOSE("DML::%s: gpuvm_page_size_kbytes = %u\n", __func__, p->gpuvm_page_size_kbytes); + DML_LOG_VERBOSE("DML::%s: vp_stationary = %u\n", __func__, p->vp_stationary); + DML_LOG_VERBOSE("DML::%s: tiling_mode = %u\n", __func__, p->tiling_mode); + DML_LOG_VERBOSE("DML::%s: vp_start_x = %u\n", __func__, p->vp_start_x); + DML_LOG_VERBOSE("DML::%s: vp_start_y = %u\n", __func__, p->vp_start_y); + DML_LOG_VERBOSE("DML::%s: full_vp_width = %u\n", __func__, p->full_vp_width); + DML_LOG_VERBOSE("DML::%s: full_vp_height = %u\n", __func__, p->full_vp_height); + DML_LOG_VERBOSE("DML::%s: blk_width = %u\n", __func__, p->blk_width); + DML_LOG_VERBOSE("DML::%s: blk_height = %u\n", __func__, p->blk_height); + DML_LOG_VERBOSE("DML::%s: vmpg_width = %u\n", __func__, p->vmpg_width); + DML_LOG_VERBOSE("DML::%s: vmpg_height = %u\n", __func__, p->vmpg_height); + DML_LOG_VERBOSE("DML::%s: full_swath_bytes = %u\n", __func__, p->full_swath_bytes); +#endif + DML_ASSERT(p->mcache_line_size_bytes != 0); + DML_ASSERT(p->mcache_size_bytes != 0); *p->mvmpg_width = 0; *p->mvmpg_height = 0; @@ -2370,8 +2313,8 @@ static void calculate_mcache_row_bytes( *p->mvmpg_width = p->vmpg_width; *p->mvmpg_height = p->vmpg_height; } else if (!((blk_bytes == 65536) && (vmpg_bytes == 4096))) { - dml2_printf("ERROR: DML::%s: Tiling size and vm page size combination not supported\n", __func__); - DML2_ASSERT(0); + DML_LOG_VERBOSE("ERROR: DML::%s: Tiling size and vm page size combination not supported\n", __func__); + DML_ASSERT(0); } } @@ -2439,25 +2382,25 @@ static void calculate_mcache_row_bytes( *p->mvmpg_per_mcache_lb = (unsigned int)math_floor2(mvmpg_per_mcache, 1); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: gpuvm_enable = %u\n", __func__, p->gpuvm_enable); - dml2_printf("DML::%s: vmpg_bytes = %u\n", __func__, vmpg_bytes); - dml2_printf("DML::%s: blk_bytes = %u\n", __func__, blk_bytes); - dml2_printf("DML::%s: meta_per_mvmpg_per_channel = %f\n", __func__, meta_per_mvmpg_per_channel); - dml2_printf("DML::%s: mvmpg_per_row_ub = %u\n", __func__, mvmpg_per_row_ub); - dml2_printf("DML::%s: meta_row_width_ub = %u\n", __func__, *p->meta_row_width_ub); - dml2_printf("DML::%s: mvmpg_width = %u\n", __func__, *p->mvmpg_width); - dml2_printf("DML::%s: mvmpg_height = %u\n", __func__, *p->mvmpg_height); - dml2_printf("DML::%s: dcc_dram_bw_nom_overhead_factor = %f\n", __func__, *p->dcc_dram_bw_nom_overhead_factor); - dml2_printf("DML::%s: dcc_dram_bw_pref_overhead_factor = %f\n", __func__, *p->dcc_dram_bw_pref_overhead_factor); + DML_LOG_VERBOSE("DML::%s: gpuvm_enable = %u\n", __func__, p->gpuvm_enable); + DML_LOG_VERBOSE("DML::%s: vmpg_bytes = %u\n", __func__, vmpg_bytes); + DML_LOG_VERBOSE("DML::%s: blk_bytes = %u\n", __func__, blk_bytes); + DML_LOG_VERBOSE("DML::%s: meta_per_mvmpg_per_channel = %f\n", __func__, meta_per_mvmpg_per_channel); + DML_LOG_VERBOSE("DML::%s: mvmpg_per_row_ub = %u\n", __func__, mvmpg_per_row_ub); + DML_LOG_VERBOSE("DML::%s: meta_row_width_ub = %u\n", __func__, *p->meta_row_width_ub); + DML_LOG_VERBOSE("DML::%s: mvmpg_width = %u\n", __func__, *p->mvmpg_width); + DML_LOG_VERBOSE("DML::%s: mvmpg_height = %u\n", __func__, *p->mvmpg_height); + DML_LOG_VERBOSE("DML::%s: dcc_dram_bw_nom_overhead_factor = %f\n", __func__, *p->dcc_dram_bw_nom_overhead_factor); + DML_LOG_VERBOSE("DML::%s: dcc_dram_bw_pref_overhead_factor = %f\n", __func__, *p->dcc_dram_bw_pref_overhead_factor); #endif } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: mcache_row_bytes = %u\n", __func__, *p->mcache_row_bytes); - dml2_printf("DML::%s: mcache_row_bytes_per_channel = %u\n", __func__, *p->mcache_row_bytes_per_channel); - dml2_printf("DML::%s: num_mcaches = %u\n", __func__, *p->num_mcaches); + DML_LOG_VERBOSE("DML::%s: mcache_row_bytes = %u\n", __func__, *p->mcache_row_bytes); + DML_LOG_VERBOSE("DML::%s: mcache_row_bytes_per_channel = %u\n", __func__, *p->mcache_row_bytes_per_channel); + DML_LOG_VERBOSE("DML::%s: num_mcaches = %u\n", __func__, *p->num_mcaches); #endif - DML2_ASSERT(*p->num_mcaches > 0); + DML_ASSERT(*p->num_mcaches > 0); } static void calculate_mcache_setting( @@ -2523,7 +2466,7 @@ static void calculate_mcache_setting( l->l_p.mvmpg_per_mcache_lb = &l->mvmpg_per_mcache_lb_l; calculate_mcache_row_bytes(scratch, &l->l_p); - DML2_ASSERT(*p->num_mcaches_l > 0); + DML_ASSERT(*p->num_mcaches_l > 0); if (l->is_dual_plane) { l->c_p.num_chans = p->num_chans; @@ -2559,7 +2502,7 @@ static void calculate_mcache_setting( l->c_p.mvmpg_per_mcache_lb = &l->mvmpg_per_mcache_lb_c; calculate_mcache_row_bytes(scratch, &l->c_p); - DML2_ASSERT(*p->num_mcaches_c > 0); + DML_ASSERT(*p->num_mcaches_c > 0); } // Sharing for iMALL access @@ -2598,28 +2541,28 @@ static void calculate_mcache_setting( } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: imall_enable = %u\n", __func__, p->imall_enable); - dml2_printf("DML::%s: is_dual_plane = %u\n", __func__, l->is_dual_plane); - dml2_printf("DML::%s: surf_vert = %u\n", __func__, p->surf_vert); - dml2_printf("DML::%s: mvmpg_width_l = %u\n", __func__, l->mvmpg_width_l); - dml2_printf("DML::%s: mvmpg_height_l = %u\n", __func__, l->mvmpg_height_l); - dml2_printf("DML::%s: mcache_remainder_l = %f\n", __func__, l->mcache_remainder_l); - dml2_printf("DML::%s: num_mcaches_l = %u\n", __func__, *p->num_mcaches_l); - dml2_printf("DML::%s: avg_mcache_element_size_l = %u\n", __func__, l->avg_mcache_element_size_l); - dml2_printf("DML::%s: mvmpg_access_width_l = %u\n", __func__, l->mvmpg_access_width_l); - dml2_printf("DML::%s: mall_comb_mcache_l = %u\n", __func__, *p->mall_comb_mcache_l); + DML_LOG_VERBOSE("DML::%s: imall_enable = %u\n", __func__, p->imall_enable); + DML_LOG_VERBOSE("DML::%s: is_dual_plane = %u\n", __func__, l->is_dual_plane); + DML_LOG_VERBOSE("DML::%s: surf_vert = %u\n", __func__, p->surf_vert); + DML_LOG_VERBOSE("DML::%s: mvmpg_width_l = %u\n", __func__, l->mvmpg_width_l); + DML_LOG_VERBOSE("DML::%s: mvmpg_height_l = %u\n", __func__, l->mvmpg_height_l); + DML_LOG_VERBOSE("DML::%s: mcache_remainder_l = %f\n", __func__, l->mcache_remainder_l); + DML_LOG_VERBOSE("DML::%s: num_mcaches_l = %u\n", __func__, *p->num_mcaches_l); + DML_LOG_VERBOSE("DML::%s: avg_mcache_element_size_l = %u\n", __func__, l->avg_mcache_element_size_l); + DML_LOG_VERBOSE("DML::%s: mvmpg_access_width_l = %u\n", __func__, l->mvmpg_access_width_l); + DML_LOG_VERBOSE("DML::%s: mall_comb_mcache_l = %u\n", __func__, *p->mall_comb_mcache_l); if (l->is_dual_plane) { - dml2_printf("DML::%s: mvmpg_width_c = %u\n", __func__, l->mvmpg_width_c); - dml2_printf("DML::%s: mvmpg_height_c = %u\n", __func__, l->mvmpg_height_c); - dml2_printf("DML::%s: mcache_remainder_c = %f\n", __func__, l->mcache_remainder_c); - dml2_printf("DML::%s: luma_time_factor = %f\n", __func__, l->luma_time_factor); - dml2_printf("DML::%s: num_mcaches_c = %u\n", __func__, *p->num_mcaches_c); - dml2_printf("DML::%s: avg_mcache_element_size_c = %u\n", __func__, l->avg_mcache_element_size_c); - dml2_printf("DML::%s: mvmpg_access_width_c = %u\n", __func__, l->mvmpg_access_width_c); - dml2_printf("DML::%s: mall_comb_mcache_c = %u\n", __func__, *p->mall_comb_mcache_c); - dml2_printf("DML::%s: lc_comb_last_mcache_size = %u\n", __func__, l->lc_comb_last_mcache_size); - dml2_printf("DML::%s: lc_comb_mcache = %u\n", __func__, *p->lc_comb_mcache); + DML_LOG_VERBOSE("DML::%s: mvmpg_width_c = %u\n", __func__, l->mvmpg_width_c); + DML_LOG_VERBOSE("DML::%s: mvmpg_height_c = %u\n", __func__, l->mvmpg_height_c); + DML_LOG_VERBOSE("DML::%s: mcache_remainder_c = %f\n", __func__, l->mcache_remainder_c); + DML_LOG_VERBOSE("DML::%s: luma_time_factor = %f\n", __func__, l->luma_time_factor); + DML_LOG_VERBOSE("DML::%s: num_mcaches_c = %u\n", __func__, *p->num_mcaches_c); + DML_LOG_VERBOSE("DML::%s: avg_mcache_element_size_c = %u\n", __func__, l->avg_mcache_element_size_c); + DML_LOG_VERBOSE("DML::%s: mvmpg_access_width_c = %u\n", __func__, l->mvmpg_access_width_c); + DML_LOG_VERBOSE("DML::%s: mall_comb_mcache_c = %u\n", __func__, *p->mall_comb_mcache_c); + DML_LOG_VERBOSE("DML::%s: lc_comb_last_mcache_size = %u\n", __func__, l->lc_comb_last_mcache_size); + DML_LOG_VERBOSE("DML::%s: lc_comb_mcache = %u\n", __func__, *p->lc_comb_mcache); } #endif // calculate split_coordinate @@ -2639,11 +2582,11 @@ static void calculate_mcache_setting( } #ifdef __DML_VBA_DEBUG__ for (n = 0; n < *p->num_mcaches_l; n++) - dml2_printf("DML::%s: mcache_offsets_l[%u] = %u\n", __func__, n, p->mcache_offsets_l[n]); + DML_LOG_VERBOSE("DML::%s: mcache_offsets_l[%u] = %u\n", __func__, n, p->mcache_offsets_l[n]); if (l->is_dual_plane) { for (n = 0; n < *p->num_mcaches_c; n++) - dml2_printf("DML::%s: mcache_offsets_c[%u] = %u\n", __func__, n, p->mcache_offsets_c[n]); + DML_LOG_VERBOSE("DML::%s: mcache_offsets_c[%u] = %u\n", __func__, n, p->mcache_offsets_c[n]); } #endif @@ -2660,10 +2603,10 @@ static void calculate_mcache_setting( #ifdef __DML_VBA_DEBUG__ for (n = 0; n < *p->num_mcaches_l; n++) - dml2_printf("DML::%s: mcache_offsets_l[%u] = %u\n", __func__, n, p->mcache_offsets_l[n]); + DML_LOG_VERBOSE("DML::%s: mcache_offsets_l[%u] = %u\n", __func__, n, p->mcache_offsets_l[n]); for (n = 0; n < *p->num_mcaches_c; n++) - dml2_printf("DML::%s: mcache_offsets_c[%u] = %u\n", __func__, n, p->mcache_offsets_c[n]); + DML_LOG_VERBOSE("DML::%s: mcache_offsets_c[%u] = %u\n", __func__, n, p->mcache_offsets_c[n]); #endif } @@ -2694,8 +2637,8 @@ static void calculate_mall_bw_overhead_factor( mall_prefetch_dram_overhead_factor[k] = 2.0; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, mall_prefetch_sdp_overhead_factor = %f\n", __func__, k, mall_prefetch_sdp_overhead_factor[k]); - dml2_printf("DML::%s: k=%u, mall_prefetch_dram_overhead_factor = %f\n", __func__, k, mall_prefetch_dram_overhead_factor[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, mall_prefetch_sdp_overhead_factor = %f\n", __func__, k, mall_prefetch_sdp_overhead_factor[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, mall_prefetch_dram_overhead_factor = %f\n", __func__, k, mall_prefetch_dram_overhead_factor[k]); #endif } } @@ -2772,22 +2715,20 @@ static double dml_get_return_bandwidth_available( else // dml2_core_internal_bw_dram return_bw_mbps = derate_dram_bandwidth; -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: is_avg_bw = %u\n", __func__, is_avg_bw); - dml2_printf("DML::%s: is_hvm_en = %u\n", __func__, is_hvm_en); - dml2_printf("DML::%s: is_hvm_only = %u\n", __func__, is_hvm_only); - dml2_printf("DML::%s: state_type = %s\n", __func__, dml2_core_internal_soc_state_type_str(state_type)); - dml2_printf("DML::%s: bw_type = %s\n", __func__, dml2_core_internal_bw_type_str(bw_type)); - dml2_printf("DML::%s: dcfclk_mhz = %f\n", __func__, dcfclk_mhz); - dml2_printf("DML::%s: fclk_mhz = %f\n", __func__, fclk_mhz); - dml2_printf("DML::%s: ideal_sdp_bandwidth = %f\n", __func__, ideal_sdp_bandwidth); - dml2_printf("DML::%s: ideal_fabric_bandwidth = %f\n", __func__, ideal_fabric_bandwidth); - dml2_printf("DML::%s: ideal_dram_bandwidth = %f\n", __func__, ideal_dram_bandwidth); - dml2_printf("DML::%s: derate_sdp_bandwidth = %f (derate %f)\n", __func__, derate_sdp_bandwidth, derate_sdp_factor); - dml2_printf("DML::%s: derate_fabric_bandwidth = %f (derate %f)\n", __func__, derate_fabric_bandwidth, derate_fabric_factor); - dml2_printf("DML::%s: derate_dram_bandwidth = %f (derate %f)\n", __func__, derate_dram_bandwidth, derate_dram_factor); - dml2_printf("DML::%s: return_bw_mbps = %f\n", __func__, return_bw_mbps); -#endif + DML_LOG_VERBOSE("DML::%s: is_avg_bw = %u\n", __func__, is_avg_bw); + DML_LOG_VERBOSE("DML::%s: is_hvm_en = %u\n", __func__, is_hvm_en); + DML_LOG_VERBOSE("DML::%s: is_hvm_only = %u\n", __func__, is_hvm_only); + DML_LOG_VERBOSE("DML::%s: state_type = %s\n", __func__, dml2_core_internal_soc_state_type_str(state_type)); + DML_LOG_VERBOSE("DML::%s: bw_type = %s\n", __func__, dml2_core_internal_bw_type_str(bw_type)); + DML_LOG_VERBOSE("DML::%s: dcfclk_mhz = %f\n", __func__, dcfclk_mhz); + DML_LOG_VERBOSE("DML::%s: fclk_mhz = %f\n", __func__, fclk_mhz); + DML_LOG_VERBOSE("DML::%s: ideal_sdp_bandwidth = %f\n", __func__, ideal_sdp_bandwidth); + DML_LOG_VERBOSE("DML::%s: ideal_fabric_bandwidth = %f\n", __func__, ideal_fabric_bandwidth); + DML_LOG_VERBOSE("DML::%s: ideal_dram_bandwidth = %f\n", __func__, ideal_dram_bandwidth); + DML_LOG_VERBOSE("DML::%s: derate_sdp_bandwidth = %f (derate %f)\n", __func__, derate_sdp_bandwidth, derate_sdp_factor); + DML_LOG_VERBOSE("DML::%s: derate_fabric_bandwidth = %f (derate %f)\n", __func__, derate_fabric_bandwidth, derate_fabric_factor); + DML_LOG_VERBOSE("DML::%s: derate_dram_bandwidth = %f (derate %f)\n", __func__, derate_dram_bandwidth, derate_dram_factor); + DML_LOG_VERBOSE("DML::%s: return_bw_mbps = %f\n", __func__, return_bw_mbps); return return_bw_mbps; } @@ -2807,9 +2748,9 @@ static noinline_for_stack void calculate_bandwidth_available( { unsigned int n, m; - dml2_printf("DML::%s: dcfclk_mhz = %f\n", __func__, dcfclk_mhz); - dml2_printf("DML::%s: fclk_mhz = %f\n", __func__, fclk_mhz); - dml2_printf("DML::%s: dram_bw_mbps = %f\n", __func__, dram_bw_mbps); + DML_LOG_VERBOSE("DML::%s: dcfclk_mhz = %f\n", __func__, dcfclk_mhz); + DML_LOG_VERBOSE("DML::%s: fclk_mhz = %f\n", __func__, fclk_mhz); + DML_LOG_VERBOSE("DML::%s: dram_bw_mbps = %f\n", __func__, dram_bw_mbps); // Calculate all the bandwidth availabe for (m = 0; m < dml2_core_internal_soc_state_max; m++) { @@ -2828,8 +2769,8 @@ static noinline_for_stack void calculate_bandwidth_available( #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: avg_bandwidth_available[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), avg_bandwidth_available[m][n]); - dml2_printf("DML::%s: urg_bandwidth_available[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), urg_bandwidth_available[m][n]); + DML_LOG_VERBOSE("DML::%s: avg_bandwidth_available[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), avg_bandwidth_available[m][n]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), urg_bandwidth_available[m][n]); #endif // urg_bandwidth_available_vm_only is indexed by soc_state @@ -2843,9 +2784,9 @@ static noinline_for_stack void calculate_bandwidth_available( urg_bandwidth_available_min[m] = math_min2(urg_bandwidth_available[m][dml2_core_internal_bw_dram], urg_bandwidth_available[m][dml2_core_internal_bw_sdp]); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: avg_bandwidth_available_min[%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), avg_bandwidth_available_min[m]); - dml2_printf("DML::%s: urg_bandwidth_available_min[%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), urg_bandwidth_available_min[m]); - dml2_printf("DML::%s: urg_bandwidth_available_vm_only[%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), urg_bandwidth_available_vm_only[n]); + DML_LOG_VERBOSE("DML::%s: avg_bandwidth_available_min[%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), avg_bandwidth_available_min[m]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available_min[%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), urg_bandwidth_available_min[m]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available_vm_only[%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), urg_bandwidth_available_vm_only[n]); #endif } } @@ -2879,13 +2820,13 @@ static void calculate_avg_bandwidth_required( // SysActive and SVP Prefetch AVG bandwidth Check for (k = 0; k < num_active_planes; ++k) { #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: plane %0d\n", __func__, k); - dml2_printf("DML::%s: ReadBandwidthLuma=%f\n", __func__, ReadBandwidthLuma[k]); - dml2_printf("DML::%s: ReadBandwidthChroma=%f\n", __func__, ReadBandwidthChroma[k]); - dml2_printf("DML::%s: dcc_dram_bw_nom_overhead_factor_p0=%f\n", __func__, dcc_dram_bw_nom_overhead_factor_p0[k]); - dml2_printf("DML::%s: dcc_dram_bw_nom_overhead_factor_p1=%f\n", __func__, dcc_dram_bw_nom_overhead_factor_p1[k]); - dml2_printf("DML::%s: mall_prefetch_dram_overhead_factor=%f\n", __func__, mall_prefetch_dram_overhead_factor[k]); - dml2_printf("DML::%s: mall_prefetch_sdp_overhead_factor=%f\n", __func__, mall_prefetch_sdp_overhead_factor[k]); + DML_LOG_VERBOSE("DML::%s: plane %0d\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: ReadBandwidthLuma=%f\n", __func__, ReadBandwidthLuma[k]); + DML_LOG_VERBOSE("DML::%s: ReadBandwidthChroma=%f\n", __func__, ReadBandwidthChroma[k]); + DML_LOG_VERBOSE("DML::%s: dcc_dram_bw_nom_overhead_factor_p0=%f\n", __func__, dcc_dram_bw_nom_overhead_factor_p0[k]); + DML_LOG_VERBOSE("DML::%s: dcc_dram_bw_nom_overhead_factor_p1=%f\n", __func__, dcc_dram_bw_nom_overhead_factor_p1[k]); + DML_LOG_VERBOSE("DML::%s: mall_prefetch_dram_overhead_factor=%f\n", __func__, mall_prefetch_dram_overhead_factor[k]); + DML_LOG_VERBOSE("DML::%s: mall_prefetch_sdp_overhead_factor=%f\n", __func__, mall_prefetch_sdp_overhead_factor[k]); #endif sdp_overhead_factor = mall_prefetch_sdp_overhead_factor[k]; @@ -2902,10 +2843,10 @@ static void calculate_avg_bandwidth_required( avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] += dram_overhead_factor_p0 * ReadBandwidthLuma[k] + dram_overhead_factor_p1 * ReadBandwidthChroma[k] + cursor_bw[k]; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_sys_active), dml2_core_internal_bw_type_str(dml2_core_internal_bw_sdp), avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); - dml2_printf("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_sys_active), dml2_core_internal_bw_type_str(dml2_core_internal_bw_dram), avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); - dml2_printf("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_svp_prefetch), dml2_core_internal_bw_type_str(dml2_core_internal_bw_sdp), avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); - dml2_printf("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_svp_prefetch), dml2_core_internal_bw_type_str(dml2_core_internal_bw_dram), avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); + DML_LOG_VERBOSE("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_sys_active), dml2_core_internal_bw_type_str(dml2_core_internal_bw_sdp), avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); + DML_LOG_VERBOSE("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_sys_active), dml2_core_internal_bw_type_str(dml2_core_internal_bw_dram), avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); + DML_LOG_VERBOSE("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_svp_prefetch), dml2_core_internal_bw_type_str(dml2_core_internal_bw_sdp), avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); + DML_LOG_VERBOSE("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_svp_prefetch), dml2_core_internal_bw_type_str(dml2_core_internal_bw_dram), avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); #endif } } @@ -3080,10 +3021,10 @@ static void CalculateVMRowAndSwath(struct dml2_core_internal_scratch *scratch, &p->MaxNumSwathY[k]); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, vm_bytes_l = %u (before hvm level)\n", __func__, k, s->vm_bytes_l); - dml2_printf("DML::%s: k=%u, vm_bytes_c = %u (before hvm level)\n", __func__, k, s->vm_bytes_c); - dml2_printf("DML::%s: k=%u, meta_row_bytes_per_row_ub_l = %u\n", __func__, k, s->meta_row_bytes_per_row_ub_l[k]); - dml2_printf("DML::%s: k=%u, meta_row_bytes_per_row_ub_c = %u\n", __func__, k, s->meta_row_bytes_per_row_ub_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, vm_bytes_l = %u (before hvm level)\n", __func__, k, s->vm_bytes_l); + DML_LOG_VERBOSE("DML::%s: k=%u, vm_bytes_c = %u (before hvm level)\n", __func__, k, s->vm_bytes_c); + DML_LOG_VERBOSE("DML::%s: k=%u, meta_row_bytes_per_row_ub_l = %u\n", __func__, k, s->meta_row_bytes_per_row_ub_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, meta_row_bytes_per_row_ub_c = %u\n", __func__, k, s->meta_row_bytes_per_row_ub_c[k]); #endif p->vm_bytes[k] = (s->vm_bytes_l + s->vm_bytes_c) * (1 + 8 * s->HostVMDynamicLevels); p->meta_row_bytes[k] = s->meta_row_bytes_per_row_ub_l[k] + s->meta_row_bytes_per_row_ub_c[k]; @@ -3091,8 +3032,8 @@ static void CalculateVMRowAndSwath(struct dml2_core_internal_scratch *scratch, p->meta_row_bytes_per_row_ub_c[k] = s->meta_row_bytes_per_row_ub_c[k]; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, meta_row_bytes = %u\n", __func__, k, p->meta_row_bytes[k]); - dml2_printf("DML::%s: k=%u, vm_bytes = %u (after hvm level)\n", __func__, k, p->vm_bytes[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, meta_row_bytes = %u\n", __func__, k, p->meta_row_bytes[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, vm_bytes = %u (after hvm level)\n", __func__, k, p->vm_bytes[k]); #endif if (s->PixelPTEBytesPerRowStorageY[k] <= 64 * s->PTEBufferSizeInRequestsForLuma[k] && s->PixelPTEBytesPerRowStorageC[k] <= 64 * s->PTEBufferSizeInRequestsForChroma[k]) { p->PTEBufferSizeNotExceeded[k] = true; @@ -3104,18 +3045,18 @@ static void CalculateVMRowAndSwath(struct dml2_core_internal_scratch *scratch, s->PixelPTEBytesPerRowC_one_row_per_frame[k] <= 64 * 2 * s->PTEBufferSizeInRequestsForChroma[k]); #ifdef __DML_VBA_DEBUG__ if (p->PTEBufferSizeNotExceeded[k] == 0 || s->one_row_per_frame_fits_in_buffer[k] == 0) { - dml2_printf("DML::%s: k=%u, PixelPTEBytesPerRowY = %u (before hvm level)\n", __func__, k, s->PixelPTEBytesPerRowY[k]); - dml2_printf("DML::%s: k=%u, PixelPTEBytesPerRowC = %u (before hvm level)\n", __func__, k, s->PixelPTEBytesPerRowC[k]); - dml2_printf("DML::%s: k=%u, PixelPTEBytesPerRowStorageY = %u\n", __func__, k, s->PixelPTEBytesPerRowStorageY[k]); - dml2_printf("DML::%s: k=%u, PixelPTEBytesPerRowStorageC = %u\n", __func__, k, s->PixelPTEBytesPerRowStorageC[k]); - dml2_printf("DML::%s: k=%u, PTEBufferSizeInRequestsForLuma = %u\n", __func__, k, s->PTEBufferSizeInRequestsForLuma[k]); - dml2_printf("DML::%s: k=%u, PTEBufferSizeInRequestsForChroma = %u\n", __func__, k, s->PTEBufferSizeInRequestsForChroma[k]); - dml2_printf("DML::%s: k=%u, PTEBufferSizeNotExceeded (not one_row_per_frame) = %u\n", __func__, k, p->PTEBufferSizeNotExceeded[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowY = %u (before hvm level)\n", __func__, k, s->PixelPTEBytesPerRowY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowC = %u (before hvm level)\n", __func__, k, s->PixelPTEBytesPerRowC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowStorageY = %u\n", __func__, k, s->PixelPTEBytesPerRowStorageY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowStorageC = %u\n", __func__, k, s->PixelPTEBytesPerRowStorageC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PTEBufferSizeInRequestsForLuma = %u\n", __func__, k, s->PTEBufferSizeInRequestsForLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PTEBufferSizeInRequestsForChroma = %u\n", __func__, k, s->PTEBufferSizeInRequestsForChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PTEBufferSizeNotExceeded (not one_row_per_frame) = %u\n", __func__, k, p->PTEBufferSizeNotExceeded[k]); - dml2_printf("DML::%s: k=%u, HostVMDynamicLevels = %u\n", __func__, k, s->HostVMDynamicLevels); - dml2_printf("DML::%s: k=%u, PixelPTEBytesPerRowY_one_row_per_frame = %u\n", __func__, k, s->PixelPTEBytesPerRowY_one_row_per_frame[k]); - dml2_printf("DML::%s: k=%u, PixelPTEBytesPerRowC_one_row_per_frame = %u\n", __func__, k, s->PixelPTEBytesPerRowC_one_row_per_frame[k]); - dml2_printf("DML::%s: k=%u, one_row_per_frame_fits_in_buffer = %u\n", __func__, k, s->one_row_per_frame_fits_in_buffer[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, HostVMDynamicLevels = %u\n", __func__, k, s->HostVMDynamicLevels); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowY_one_row_per_frame = %u\n", __func__, k, s->PixelPTEBytesPerRowY_one_row_per_frame[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowC_one_row_per_frame = %u\n", __func__, k, s->PixelPTEBytesPerRowC_one_row_per_frame[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, one_row_per_frame_fits_in_buffer = %u\n", __func__, k, s->one_row_per_frame_fits_in_buffer[k]); } #endif } @@ -3146,8 +3087,8 @@ static void CalculateVMRowAndSwath(struct dml2_core_internal_scratch *scratch, for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { p->DCCMetaBufferSizeNotExceeded[k] = true; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, SurfaceSizeInMALL = %u\n", __func__, k, p->SurfaceSizeInMALL[k]); - dml2_printf("DML::%s: k=%u, is_using_mall_for_ss = %u\n", __func__, k, p->is_using_mall_for_ss[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, SurfaceSizeInMALL = %u\n", __func__, k, p->SurfaceSizeInMALL[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, is_using_mall_for_ss = %u\n", __func__, k, p->is_using_mall_for_ss[k]); #endif p->use_one_row_for_frame[k] = p->myPipe[k].FORCE_ONE_ROW_FOR_FRAME || p->is_using_mall_for_ss[k] || (p->display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_main_pipe) || (dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k])) || (p->display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes > 64 && dml_is_vertical_rotation(p->myPipe[k].RotationAngle)); @@ -3170,9 +3111,9 @@ static void CalculateVMRowAndSwath(struct dml2_core_internal_scratch *scratch, p->DCCMetaBufferSizeNotExceeded[k] = false; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%d, meta_row_bytes = %d\n", __func__, k, p->meta_row_bytes[k]); - dml2_printf("DML::%s: k=%d, DCCMetaBufferSizeBytes = %d\n", __func__, k, p->DCCMetaBufferSizeBytes); - dml2_printf("DML::%s: k=%d, DCCMetaBufferSizeNotExceeded = %d\n", __func__, k, p->DCCMetaBufferSizeNotExceeded[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, meta_row_bytes = %d\n", __func__, k, p->meta_row_bytes[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, DCCMetaBufferSizeBytes = %d\n", __func__, k, p->DCCMetaBufferSizeBytes); + DML_LOG_VERBOSE("DML::%s: k=%d, DCCMetaBufferSizeNotExceeded = %d\n", __func__, k, p->DCCMetaBufferSizeNotExceeded[k]); #endif } @@ -3209,20 +3150,20 @@ static void CalculateVMRowAndSwath(struct dml2_core_internal_scratch *scratch, &p->dpte_row_bw[k], &p->meta_row_bw[k]); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, use_one_row_for_frame = %u\n", __func__, k, p->use_one_row_for_frame[k]); - dml2_printf("DML::%s: k=%u, use_one_row_for_frame_flip = %u\n", __func__, k, p->use_one_row_for_frame_flip[k]); - dml2_printf("DML::%s: k=%u, UseMALLForPStateChange = %u\n", __func__, k, p->display_cfg->plane_descriptors[k].overrides.legacy_svp_config); - dml2_printf("DML::%s: k=%u, dpte_row_height_luma = %u\n", __func__, k, p->dpte_row_height_luma[k]); - dml2_printf("DML::%s: k=%u, dpte_row_width_luma_ub = %u\n", __func__, k, p->dpte_row_width_luma_ub[k]); - dml2_printf("DML::%s: k=%u, PixelPTEBytesPerRowY = %u (after hvm level)\n", __func__, k, s->PixelPTEBytesPerRowY[k]); - dml2_printf("DML::%s: k=%u, dpte_row_height_chroma = %u\n", __func__, k, p->dpte_row_height_chroma[k]); - dml2_printf("DML::%s: k=%u, dpte_row_width_chroma_ub = %u\n", __func__, k, p->dpte_row_width_chroma_ub[k]); - dml2_printf("DML::%s: k=%u, PixelPTEBytesPerRowC = %u (after hvm level)\n", __func__, k, s->PixelPTEBytesPerRowC[k]); - dml2_printf("DML::%s: k=%u, PixelPTEBytesPerRow = %u\n", __func__, k, p->PixelPTEBytesPerRow[k]); - dml2_printf("DML::%s: k=%u, PTEBufferSizeNotExceeded = %u\n", __func__, k, p->PTEBufferSizeNotExceeded[k]); - dml2_printf("DML::%s: k=%u, gpuvm_enable = %u\n", __func__, k, p->display_cfg->gpuvm_enable); - dml2_printf("DML::%s: k=%u, PTE_BUFFER_MODE = %u\n", __func__, k, p->PTE_BUFFER_MODE[k]); - dml2_printf("DML::%s: k=%u, BIGK_FRAGMENT_SIZE = %u\n", __func__, k, p->BIGK_FRAGMENT_SIZE[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, use_one_row_for_frame = %u\n", __func__, k, p->use_one_row_for_frame[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, use_one_row_for_frame_flip = %u\n", __func__, k, p->use_one_row_for_frame_flip[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, UseMALLForPStateChange = %u\n", __func__, k, p->display_cfg->plane_descriptors[k].overrides.legacy_svp_config); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_height_luma = %u\n", __func__, k, p->dpte_row_height_luma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_width_luma_ub = %u\n", __func__, k, p->dpte_row_width_luma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowY = %u (after hvm level)\n", __func__, k, s->PixelPTEBytesPerRowY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_height_chroma = %u\n", __func__, k, p->dpte_row_height_chroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_width_chroma_ub = %u\n", __func__, k, p->dpte_row_width_chroma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowC = %u (after hvm level)\n", __func__, k, s->PixelPTEBytesPerRowC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRow = %u\n", __func__, k, p->PixelPTEBytesPerRow[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PTEBufferSizeNotExceeded = %u\n", __func__, k, p->PTEBufferSizeNotExceeded[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, gpuvm_enable = %u\n", __func__, k, p->display_cfg->gpuvm_enable); + DML_LOG_VERBOSE("DML::%s: k=%u, PTE_BUFFER_MODE = %u\n", __func__, k, p->PTE_BUFFER_MODE[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, BIGK_FRAGMENT_SIZE = %u\n", __func__, k, p->BIGK_FRAGMENT_SIZE[k]); #endif } } @@ -3257,19 +3198,19 @@ static double CalculateUrgentLatency( } #ifdef __DML_VBA_DEBUG__ if (qos_type == dml2_qos_param_type_dcn4x) { - dml2_printf("DML::%s: qos_type = %d\n", __func__, qos_type); - dml2_printf("DML::%s: urgent_ramp_uclk_cycles = %d\n", __func__, urgent_ramp_uclk_cycles); - dml2_printf("DML::%s: uclk_freq_mhz = %f\n", __func__, uclk_freq_mhz); - dml2_printf("DML::%s: umc_urgent_ramp_latency_margin = %f\n", __func__, umc_urgent_ramp_latency_margin); + DML_LOG_VERBOSE("DML::%s: qos_type = %d\n", __func__, qos_type); + DML_LOG_VERBOSE("DML::%s: urgent_ramp_uclk_cycles = %d\n", __func__, urgent_ramp_uclk_cycles); + DML_LOG_VERBOSE("DML::%s: uclk_freq_mhz = %f\n", __func__, uclk_freq_mhz); + DML_LOG_VERBOSE("DML::%s: umc_urgent_ramp_latency_margin = %f\n", __func__, umc_urgent_ramp_latency_margin); } else { - dml2_printf("DML::%s: UrgentLatencyPixelDataOnly = %f\n", __func__, UrgentLatencyPixelDataOnly); - dml2_printf("DML::%s: UrgentLatencyPixelMixedWithVMData = %f\n", __func__, UrgentLatencyPixelMixedWithVMData); - dml2_printf("DML::%s: UrgentLatencyVMDataOnly = %f\n", __func__, UrgentLatencyVMDataOnly); - dml2_printf("DML::%s: UrgentLatencyAdjustmentFabricClockComponent = %f\n", __func__, UrgentLatencyAdjustmentFabricClockComponent); - dml2_printf("DML::%s: UrgentLatencyAdjustmentFabricClockReference = %f\n", __func__, UrgentLatencyAdjustmentFabricClockReference); + DML_LOG_VERBOSE("DML::%s: UrgentLatencyPixelDataOnly = %f\n", __func__, UrgentLatencyPixelDataOnly); + DML_LOG_VERBOSE("DML::%s: UrgentLatencyPixelMixedWithVMData = %f\n", __func__, UrgentLatencyPixelMixedWithVMData); + DML_LOG_VERBOSE("DML::%s: UrgentLatencyVMDataOnly = %f\n", __func__, UrgentLatencyVMDataOnly); + DML_LOG_VERBOSE("DML::%s: UrgentLatencyAdjustmentFabricClockComponent = %f\n", __func__, UrgentLatencyAdjustmentFabricClockComponent); + DML_LOG_VERBOSE("DML::%s: UrgentLatencyAdjustmentFabricClockReference = %f\n", __func__, UrgentLatencyAdjustmentFabricClockReference); } - dml2_printf("DML::%s: FabricClock = %f\n", __func__, FabricClock); - dml2_printf("DML::%s: UrgentLatency = %f\n", __func__, urgent_latency); + DML_LOG_VERBOSE("DML::%s: FabricClock = %f\n", __func__, FabricClock); + DML_LOG_VERBOSE("DML::%s: UrgentLatency = %f\n", __func__, urgent_latency); #endif return urgent_latency; } @@ -3296,18 +3237,18 @@ static double CalculateTripToMemory( #ifdef __DML_VBA_DEBUG__ if (qos_type == dml2_qos_param_type_dcn4x) { - dml2_printf("DML::%s: qos_type = %d\n", __func__, qos_type); - dml2_printf("DML::%s: max_round_trip_to_furthest_cs_fclk_cycles = %d\n", __func__, max_round_trip_to_furthest_cs_fclk_cycles); - dml2_printf("DML::%s: mall_overhead_fclk_cycles = %d\n", __func__, mall_overhead_fclk_cycles); - dml2_printf("DML::%s: trip_to_memory_uclk_cycles = %d\n", __func__, trip_to_memory_uclk_cycles); - dml2_printf("DML::%s: uclk_freq_mhz = %f\n", __func__, uclk_freq_mhz); - dml2_printf("DML::%s: FabricClock = %f\n", __func__, FabricClock); - dml2_printf("DML::%s: fabric_max_transport_latency_margin = %f\n", __func__, fabric_max_transport_latency_margin); - dml2_printf("DML::%s: umc_max_latency_margin = %f\n", __func__, umc_max_latency_margin); + DML_LOG_VERBOSE("DML::%s: qos_type = %d\n", __func__, qos_type); + DML_LOG_VERBOSE("DML::%s: max_round_trip_to_furthest_cs_fclk_cycles = %d\n", __func__, max_round_trip_to_furthest_cs_fclk_cycles); + DML_LOG_VERBOSE("DML::%s: mall_overhead_fclk_cycles = %d\n", __func__, mall_overhead_fclk_cycles); + DML_LOG_VERBOSE("DML::%s: trip_to_memory_uclk_cycles = %d\n", __func__, trip_to_memory_uclk_cycles); + DML_LOG_VERBOSE("DML::%s: uclk_freq_mhz = %f\n", __func__, uclk_freq_mhz); + DML_LOG_VERBOSE("DML::%s: FabricClock = %f\n", __func__, FabricClock); + DML_LOG_VERBOSE("DML::%s: fabric_max_transport_latency_margin = %f\n", __func__, fabric_max_transport_latency_margin); + DML_LOG_VERBOSE("DML::%s: umc_max_latency_margin = %f\n", __func__, umc_max_latency_margin); } else { - dml2_printf("DML::%s: UrgLatency = %f\n", __func__, UrgLatency); + DML_LOG_VERBOSE("DML::%s: UrgLatency = %f\n", __func__, UrgLatency); } - dml2_printf("DML::%s: trip_to_memory_us = %f\n", __func__, trip_to_memory_us); + DML_LOG_VERBOSE("DML::%s: trip_to_memory_us = %f\n", __func__, trip_to_memory_us); #endif @@ -3334,14 +3275,14 @@ static double CalculateMetaTripToMemory( #ifdef __DML_VBA_DEBUG__ if (qos_type == dml2_qos_param_type_dcn4x) { - dml2_printf("DML::%s: qos_type = %d\n", __func__, qos_type); - dml2_printf("DML::%s: meta_trip_to_memory_fclk_cycles = %d\n", __func__, meta_trip_to_memory_fclk_cycles); - dml2_printf("DML::%s: meta_trip_to_memory_uclk_cycles = %d\n", __func__, meta_trip_to_memory_uclk_cycles); - dml2_printf("DML::%s: uclk_freq_mhz = %f\n", __func__, uclk_freq_mhz); + DML_LOG_VERBOSE("DML::%s: qos_type = %d\n", __func__, qos_type); + DML_LOG_VERBOSE("DML::%s: meta_trip_to_memory_fclk_cycles = %d\n", __func__, meta_trip_to_memory_fclk_cycles); + DML_LOG_VERBOSE("DML::%s: meta_trip_to_memory_uclk_cycles = %d\n", __func__, meta_trip_to_memory_uclk_cycles); + DML_LOG_VERBOSE("DML::%s: uclk_freq_mhz = %f\n", __func__, uclk_freq_mhz); } else { - dml2_printf("DML::%s: UrgLatency = %f\n", __func__, UrgLatency); + DML_LOG_VERBOSE("DML::%s: UrgLatency = %f\n", __func__, UrgLatency); } - dml2_printf("DML::%s: meta_trip_to_memory_us = %f\n", __func__, meta_trip_to_memory_us); + DML_LOG_VERBOSE("DML::%s: meta_trip_to_memory_us = %f\n", __func__, meta_trip_to_memory_us); #endif @@ -3358,7 +3299,6 @@ static void calculate_cursor_req_attributes( unsigned int *cursor_bytes_per_chunk, unsigned int *cursor_bytes) { - unsigned int cursor_pitch = 0; unsigned int cursor_bytes_per_req = 0; unsigned int cursor_width_bytes = 0; unsigned int cursor_height = 0; @@ -3366,10 +3306,6 @@ static void calculate_cursor_req_attributes( //SW determines the cursor pitch to support the maximum cursor_width that will be used but the following restrictions apply. //- For 2bpp, cursor_pitch = 256 pixels due to min cursor request size of 64B //- For 32 or 64 bpp, cursor_pitch = 64, 128 or 256 pixels depending on the cursor width - if (cursor_bpp == 2) - cursor_pitch = 256; - else - cursor_pitch = (unsigned int)1 << (unsigned int)math_ceil2(math_log((float)cursor_width, 2), 1); //The cursor requestor uses a cursor request size of 64B, 128B, or 256B depending on the cursor_width and cursor_bpp as follows. @@ -3409,8 +3345,8 @@ static void calculate_cursor_req_attributes( *cursor_lines_per_chunk = 1; } else { if (cursor_width > 0) { - dml2_printf("DML::%s: Invalid cursor_bpp = %d\n", __func__, cursor_bpp); - DML2_ASSERT(0); + DML_LOG_VERBOSE("DML::%s: Invalid cursor_bpp = %d\n", __func__, cursor_bpp); + DML_ASSERT(0); } } @@ -3421,15 +3357,15 @@ static void calculate_cursor_req_attributes( cursor_height = cursor_width; *cursor_bytes = *cursor_bytes_per_line * cursor_height; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: cursor_bpp = %d\n", __func__, cursor_bpp); - dml2_printf("DML::%s: cursor_width = %d\n", __func__, cursor_width); - dml2_printf("DML::%s: cursor_width_bytes = %d\n", __func__, cursor_width_bytes); - dml2_printf("DML::%s: cursor_bytes_per_req = %d\n", __func__, cursor_bytes_per_req); - dml2_printf("DML::%s: cursor_lines_per_chunk = %d\n", __func__, *cursor_lines_per_chunk); - dml2_printf("DML::%s: cursor_bytes_per_line = %d\n", __func__, *cursor_bytes_per_line); - dml2_printf("DML::%s: cursor_bytes_per_chunk = %d\n", __func__, *cursor_bytes_per_chunk); - dml2_printf("DML::%s: cursor_bytes = %d\n", __func__, *cursor_bytes); - dml2_printf("DML::%s: cursor_pitch = %d\n", __func__, cursor_pitch); + DML_LOG_VERBOSE("DML::%s: cursor_bpp = %d\n", __func__, cursor_bpp); + DML_LOG_VERBOSE("DML::%s: cursor_width = %d\n", __func__, cursor_width); + DML_LOG_VERBOSE("DML::%s: cursor_width_bytes = %d\n", __func__, cursor_width_bytes); + DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_req = %d\n", __func__, cursor_bytes_per_req); + DML_LOG_VERBOSE("DML::%s: cursor_lines_per_chunk = %d\n", __func__, *cursor_lines_per_chunk); + DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_line = %d\n", __func__, *cursor_bytes_per_line); + DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_chunk = %d\n", __func__, *cursor_bytes_per_chunk); + DML_LOG_VERBOSE("DML::%s: cursor_bytes = %d\n", __func__, *cursor_bytes); + DML_LOG_VERBOSE("DML::%s: cursor_pitch = %d\n", __func__, cursor_bpp == 2 ? 256 : (unsigned int)1 << (unsigned int)math_ceil2(math_log((float)cursor_width, 2), 1)); #endif } @@ -3460,13 +3396,13 @@ static void calculate_cursor_urgent_burst_factor( } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: LinesInCursorBuffer = %u\n", __func__, LinesInCursorBuffer); - dml2_printf("DML::%s: CursorBufferSizeInTime = %f\n", __func__, CursorBufferSizeInTime); - dml2_printf("DML::%s: CursorBufferSize = %u (kbytes)\n", __func__, CursorBufferSize); - dml2_printf("DML::%s: cursor_bytes_per_chunk = %u\n", __func__, cursor_bytes_per_chunk); - dml2_printf("DML::%s: cursor_lines_per_chunk = %u\n", __func__, cursor_lines_per_chunk); - dml2_printf("DML::%s: UrgentBurstFactorCursor = %f\n", __func__, *UrgentBurstFactorCursor); - dml2_printf("DML::%s: NotEnoughUrgentLatencyHiding = %d\n", __func__, *NotEnoughUrgentLatencyHiding); + DML_LOG_VERBOSE("DML::%s: LinesInCursorBuffer = %u\n", __func__, LinesInCursorBuffer); + DML_LOG_VERBOSE("DML::%s: CursorBufferSizeInTime = %f\n", __func__, CursorBufferSizeInTime); + DML_LOG_VERBOSE("DML::%s: CursorBufferSize = %u (kbytes)\n", __func__, CursorBufferSize); + DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_chunk = %u\n", __func__, cursor_bytes_per_chunk); + DML_LOG_VERBOSE("DML::%s: cursor_lines_per_chunk = %u\n", __func__, cursor_lines_per_chunk); + DML_LOG_VERBOSE("DML::%s: UrgentBurstFactorCursor = %f\n", __func__, *UrgentBurstFactorCursor); + DML_LOG_VERBOSE("DML::%s: NotEnoughUrgentLatencyHiding = %d\n", __func__, *NotEnoughUrgentLatencyHiding); #endif } @@ -3501,15 +3437,15 @@ static void CalculateUrgentBurstFactor( *UrgentBurstFactorChroma = 0; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: VRatio = %f\n", __func__, VRatio); - dml2_printf("DML::%s: VRatioC = %f\n", __func__, VRatioC); - dml2_printf("DML::%s: DETBufferSizeY = %d\n", __func__, DETBufferSizeY); - dml2_printf("DML::%s: DETBufferSizeC = %d\n", __func__, DETBufferSizeC); - dml2_printf("DML::%s: BytePerPixelInDETY = %f\n", __func__, BytePerPixelInDETY); - dml2_printf("DML::%s: swath_width_luma_ub = %d\n", __func__, swath_width_luma_ub); - dml2_printf("DML::%s: LineTime = %f\n", __func__, LineTime); + DML_LOG_VERBOSE("DML::%s: VRatio = %f\n", __func__, VRatio); + DML_LOG_VERBOSE("DML::%s: VRatioC = %f\n", __func__, VRatioC); + DML_LOG_VERBOSE("DML::%s: DETBufferSizeY = %d\n", __func__, DETBufferSizeY); + DML_LOG_VERBOSE("DML::%s: DETBufferSizeC = %d\n", __func__, DETBufferSizeC); + DML_LOG_VERBOSE("DML::%s: BytePerPixelInDETY = %f\n", __func__, BytePerPixelInDETY); + DML_LOG_VERBOSE("DML::%s: swath_width_luma_ub = %d\n", __func__, swath_width_luma_ub); + DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, LineTime); #endif - DML2_ASSERT(VRatio > 0); + DML_ASSERT(VRatio > 0); LinesInDETLuma = (dml_is_phantom_pipe(plane_cfg) ? 1024 * 1024 : DETBufferSizeY) / BytePerPixelInDETY / swath_width_luma_ub; @@ -3534,12 +3470,12 @@ static void CalculateUrgentBurstFactor( } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: LinesInDETLuma = %f\n", __func__, LinesInDETLuma); - dml2_printf("DML::%s: UrgentLatency = %f\n", __func__, UrgentLatency); - dml2_printf("DML::%s: DETBufferSizeInTimeLuma = %f\n", __func__, DETBufferSizeInTimeLuma); - dml2_printf("DML::%s: UrgentBurstFactorLuma = %f\n", __func__, *UrgentBurstFactorLuma); - dml2_printf("DML::%s: UrgentBurstFactorChroma = %f\n", __func__, *UrgentBurstFactorChroma); - dml2_printf("DML::%s: NotEnoughUrgentLatencyHiding = %d\n", __func__, *NotEnoughUrgentLatencyHiding); + DML_LOG_VERBOSE("DML::%s: LinesInDETLuma = %f\n", __func__, LinesInDETLuma); + DML_LOG_VERBOSE("DML::%s: UrgentLatency = %f\n", __func__, UrgentLatency); + DML_LOG_VERBOSE("DML::%s: DETBufferSizeInTimeLuma = %f\n", __func__, DETBufferSizeInTimeLuma); + DML_LOG_VERBOSE("DML::%s: UrgentBurstFactorLuma = %f\n", __func__, *UrgentBurstFactorLuma); + DML_LOG_VERBOSE("DML::%s: UrgentBurstFactorChroma = %f\n", __func__, *UrgentBurstFactorChroma); + DML_LOG_VERBOSE("DML::%s: NotEnoughUrgentLatencyHiding = %d\n", __func__, *NotEnoughUrgentLatencyHiding); #endif } @@ -3600,10 +3536,10 @@ static void CalculateDCFCLKDeepSleepTdlut( if (display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut && tdlut_bytes_to_deliver[k] > 0) { double tdlut_required_deepsleep_dcfclk = (double) tdlut_bytes_to_deliver[k] / 64.0 / prefetch_swath_time_us[k]; - dml2_printf("DML::%s: k=%d, DCFClkDeepSleepPerSurface = %f\n", __func__, k, DCFClkDeepSleepPerSurface[k]); - dml2_printf("DML::%s: k=%d, tdlut_bytes_to_deliver = %d\n", __func__, k, tdlut_bytes_to_deliver[k]); - dml2_printf("DML::%s: k=%d, prefetch_swath_time_us = %f\n", __func__, k, prefetch_swath_time_us[k]); - dml2_printf("DML::%s: k=%d, tdlut_required_deepsleep_dcfclk = %f\n", __func__, k, tdlut_required_deepsleep_dcfclk); + DML_LOG_VERBOSE("DML::%s: k=%d, DCFClkDeepSleepPerSurface = %f\n", __func__, k, DCFClkDeepSleepPerSurface[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, tdlut_bytes_to_deliver = %d\n", __func__, k, tdlut_bytes_to_deliver[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, prefetch_swath_time_us = %f\n", __func__, k, prefetch_swath_time_us[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, tdlut_required_deepsleep_dcfclk = %f\n", __func__, k, tdlut_required_deepsleep_dcfclk); // increase the deepsleep dcfclk to match the original dispclk throughput rate if (tdlut_required_deepsleep_dcfclk > DCFClkDeepSleepPerSurface[k]) { @@ -3613,8 +3549,8 @@ static void CalculateDCFCLKDeepSleepTdlut( } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, PixelClock = %f\n", __func__, k, pixel_rate_mhz); - dml2_printf("DML::%s: k=%u, DCFClkDeepSleepPerSurface = %f\n", __func__, k, DCFClkDeepSleepPerSurface[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelClock = %f\n", __func__, k, pixel_rate_mhz); + DML_LOG_VERBOSE("DML::%s: k=%u, DCFClkDeepSleepPerSurface = %f\n", __func__, k, DCFClkDeepSleepPerSurface[k]); #endif } @@ -3625,17 +3561,17 @@ static void CalculateDCFCLKDeepSleepTdlut( *DCFClkDeepSleep = math_max2(8.0, __DML2_CALCS_DCFCLK_FACTOR__ * ReadBandwidth / (double)ReturnBusWidth); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: __DML2_CALCS_DCFCLK_FACTOR__ = %f\n", __func__, __DML2_CALCS_DCFCLK_FACTOR__); - dml2_printf("DML::%s: ReadBandwidth = %f\n", __func__, ReadBandwidth); - dml2_printf("DML::%s: ReturnBusWidth = %u\n", __func__, ReturnBusWidth); - dml2_printf("DML::%s: DCFClkDeepSleep = %f\n", __func__, *DCFClkDeepSleep); + DML_LOG_VERBOSE("DML::%s: __DML2_CALCS_DCFCLK_FACTOR__ = %f\n", __func__, __DML2_CALCS_DCFCLK_FACTOR__); + DML_LOG_VERBOSE("DML::%s: ReadBandwidth = %f\n", __func__, ReadBandwidth); + DML_LOG_VERBOSE("DML::%s: ReturnBusWidth = %u\n", __func__, ReturnBusWidth); + DML_LOG_VERBOSE("DML::%s: DCFClkDeepSleep = %f\n", __func__, *DCFClkDeepSleep); #endif for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { *DCFClkDeepSleep = math_max2(*DCFClkDeepSleep, DCFClkDeepSleepPerSurface[k]); } - dml2_printf("DML::%s: DCFClkDeepSleep = %f (final)\n", __func__, *DCFClkDeepSleep); + DML_LOG_VERBOSE("DML::%s: DCFClkDeepSleep = %f (final)\n", __func__, *DCFClkDeepSleep); } static noinline_for_stack void CalculateDCFCLKDeepSleep( @@ -3731,12 +3667,12 @@ static unsigned int CalculateMaxVStartup( else max_vstartup_lines = vblank_size - (unsigned int)math_max2(1.0, math_ceil2(write_back_delay_us / line_time_us, 1.0)); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: VBlankNom = %u\n", __func__, timing->vblank_nom); - dml2_printf("DML::%s: vblank_nom_default_us = %u\n", __func__, vblank_nom_default_us); - dml2_printf("DML::%s: line_time_us = %f\n", __func__, line_time_us); - dml2_printf("DML::%s: vblank_actual = %u\n", __func__, vblank_actual); - dml2_printf("DML::%s: vblank_avail = %u\n", __func__, vblank_avail); - dml2_printf("DML::%s: max_vstartup_lines = %u\n", __func__, max_vstartup_lines); + DML_LOG_VERBOSE("DML::%s: VBlankNom = %lu\n", __func__, timing->vblank_nom); + DML_LOG_VERBOSE("DML::%s: vblank_nom_default_us = %u\n", __func__, vblank_nom_default_us); + DML_LOG_VERBOSE("DML::%s: line_time_us = %f\n", __func__, line_time_us); + DML_LOG_VERBOSE("DML::%s: vblank_actual = %u\n", __func__, vblank_actual); + DML_LOG_VERBOSE("DML::%s: vblank_avail = %u\n", __func__, vblank_avail); + DML_LOG_VERBOSE("DML::%s: max_vstartup_lines = %u\n", __func__, max_vstartup_lines); #endif max_vstartup_lines = (unsigned int)math_min2(max_vstartup_lines, DML_MAX_VSTARTUP_START); return max_vstartup_lines; @@ -3761,9 +3697,9 @@ static void CalculateSwathAndDETConfiguration(struct dml2_core_internal_scratch const long MAXIMUMCOMPRESSION = 4; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: ForceSingleDPP = %u\n", __func__, p->ForceSingleDPP); + DML_LOG_VERBOSE("DML::%s: ForceSingleDPP = %u\n", __func__, p->ForceSingleDPP); for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - dml2_printf("DML::%s: DPPPerSurface[%u] = %u\n", __func__, k, p->DPPPerSurface[k]); + DML_LOG_VERBOSE("DML::%s: DPPPerSurface[%u] = %u\n", __func__, k, p->DPPPerSurface[k]); } #endif CalculateSwathWidth( @@ -3797,15 +3733,15 @@ static void CalculateSwathAndDETConfiguration(struct dml2_core_internal_scratch p->full_swath_bytes_l[k] = (unsigned int)(p->swath_width_luma_ub[k] * p->BytePerPixDETY[k] * MaximumSwathHeightY[k]); p->full_swath_bytes_c[k] = (unsigned int)(p->swath_width_chroma_ub[k] * p->BytePerPixDETC[k] * MaximumSwathHeightC[k]); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u DPPPerSurface = %u\n", __func__, k, p->DPPPerSurface[k]); - dml2_printf("DML::%s: k=%u swath_width_luma_ub = %u\n", __func__, k, p->swath_width_luma_ub[k]); - dml2_printf("DML::%s: k=%u BytePerPixDETY = %f\n", __func__, k, p->BytePerPixDETY[k]); - dml2_printf("DML::%s: k=%u MaximumSwathHeightY = %u\n", __func__, k, MaximumSwathHeightY[k]); - dml2_printf("DML::%s: k=%u full_swath_bytes_l = %u\n", __func__, k, p->full_swath_bytes_l[k]); - dml2_printf("DML::%s: k=%u swath_width_chroma_ub = %u\n", __func__, k, p->swath_width_chroma_ub[k]); - dml2_printf("DML::%s: k=%u BytePerPixDETC = %f\n", __func__, k, p->BytePerPixDETC[k]); - dml2_printf("DML::%s: k=%u MaximumSwathHeightC = %u\n", __func__, k, MaximumSwathHeightC[k]); - dml2_printf("DML::%s: k=%u full_swath_bytes_c = %u\n", __func__, k, p->full_swath_bytes_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%u DPPPerSurface = %u\n", __func__, k, p->DPPPerSurface[k]); + DML_LOG_VERBOSE("DML::%s: k=%u swath_width_luma_ub = %u\n", __func__, k, p->swath_width_luma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u BytePerPixDETY = %f\n", __func__, k, p->BytePerPixDETY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathHeightY = %u\n", __func__, k, MaximumSwathHeightY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_l = %u\n", __func__, k, p->full_swath_bytes_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u swath_width_chroma_ub = %u\n", __func__, k, p->swath_width_chroma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u BytePerPixDETC = %f\n", __func__, k, p->BytePerPixDETC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathHeightC = %u\n", __func__, k, MaximumSwathHeightC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_c = %u\n", __func__, k, p->full_swath_bytes_c[k]); #endif if (p->display_cfg->plane_descriptors[k].pixel_format == dml2_420_10) { p->full_swath_bytes_l[k] = (unsigned int)(math_ceil2((double)p->full_swath_bytes_l[k], 256)); @@ -3848,11 +3784,11 @@ static void CalculateSwathAndDETConfiguration(struct dml2_core_internal_scratch p->CompressedBufferSizeInkByte); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: TotalActiveDPP = %u\n", __func__, TotalActiveDPP); - dml2_printf("DML::%s: nomDETInKByte = %u\n", __func__, p->nomDETInKByte); - dml2_printf("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, p->ConfigReturnBufferSizeInKByte); - dml2_printf("DML::%s: UnboundedRequestEnabled = %u\n", __func__, *p->UnboundedRequestEnabled); - dml2_printf("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, *p->CompressedBufferSizeInkByte); + DML_LOG_VERBOSE("DML::%s: TotalActiveDPP = %u\n", __func__, TotalActiveDPP); + DML_LOG_VERBOSE("DML::%s: nomDETInKByte = %u\n", __func__, p->nomDETInKByte); + DML_LOG_VERBOSE("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, p->ConfigReturnBufferSizeInKByte); + DML_LOG_VERBOSE("DML::%s: UnboundedRequestEnabled = %u\n", __func__, *p->UnboundedRequestEnabled); + DML_LOG_VERBOSE("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, *p->CompressedBufferSizeInkByte); #endif *p->ViewportSizeSupport = true; @@ -3860,7 +3796,7 @@ static void CalculateSwathAndDETConfiguration(struct dml2_core_internal_scratch DETBufferSizeInKByteForSwathCalculation = (dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k]) ? 1024 : p->DETBufferSizeInKByte[k]); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u DETBufferSizeInKByteForSwathCalculation = %u\n", __func__, k, DETBufferSizeInKByteForSwathCalculation); + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByteForSwathCalculation = %u\n", __func__, k, DETBufferSizeInKByteForSwathCalculation); #endif if (p->display_cfg->plane_descriptors[k].surface.tiling == dml2_sw_linear) { p->SwathHeightY[k] = MaximumSwathHeightY[k]; @@ -3917,13 +3853,13 @@ static void CalculateSwathAndDETConfiguration(struct dml2_core_internal_scratch if ((p->full_swath_bytes_l[k] / 2 + p->full_swath_bytes_c[k] / 2 > DETBufferSizeInKByteForSwathCalculation * 1024 / 2) || p->SwathWidth[k] > p->MaximumSwathWidthLuma[k] || (p->SwathHeightC[k] > 0 && p->SwathWidthChroma[k] > p->MaximumSwathWidthChroma[k])) { *p->ViewportSizeSupport = false; - dml2_printf("DML::%s: k=%u full_swath_bytes_l=%u\n", __func__, k, p->full_swath_bytes_l[k]); - dml2_printf("DML::%s: k=%u full_swath_bytes_c=%u\n", __func__, k, p->full_swath_bytes_c[k]); - dml2_printf("DML::%s: k=%u DETBufferSizeInKByteForSwathCalculation=%u\n", __func__, k, DETBufferSizeInKByteForSwathCalculation); - dml2_printf("DML::%s: k=%u SwathWidth=%u\n", __func__, k, p->SwathWidth[k]); - dml2_printf("DML::%s: k=%u MaximumSwathWidthLuma=%f\n", __func__, k, p->MaximumSwathWidthLuma[k]); - dml2_printf("DML::%s: k=%u SwathWidthChroma=%d\n", __func__, k, p->SwathWidthChroma[k]); - dml2_printf("DML::%s: k=%u MaximumSwathWidthChroma=%f\n", __func__, k, p->MaximumSwathWidthChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_l=%u\n", __func__, k, p->full_swath_bytes_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_c=%u\n", __func__, k, p->full_swath_bytes_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByteForSwathCalculation=%u\n", __func__, k, DETBufferSizeInKByteForSwathCalculation); + DML_LOG_VERBOSE("DML::%s: k=%u SwathWidth=%u\n", __func__, k, p->SwathWidth[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthLuma=%f\n", __func__, k, p->MaximumSwathWidthLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u SwathWidthChroma=%d\n", __func__, k, p->SwathWidthChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthChroma=%f\n", __func__, k, p->MaximumSwathWidthChroma[k]); p->ViewportSizeSupportPerSurface[k] = false; } else { p->ViewportSizeSupportPerSurface[k] = true; @@ -3931,35 +3867,35 @@ static void CalculateSwathAndDETConfiguration(struct dml2_core_internal_scratch if (p->SwathHeightC[k] == 0) { #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, All DET will be used for plane0\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: k=%u, All DET will be used for plane0\n", __func__, k); #endif p->DETBufferSizeY[k] = p->DETBufferSizeInKByte[k] * 1024; p->DETBufferSizeC[k] = 0; } else if (RoundedUpSwathSizeBytesY[k] <= 1.5 * RoundedUpSwathSizeBytesC[k]) { #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, Half DET will be used for plane0, and half for plane1\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: k=%u, Half DET will be used for plane0, and half for plane1\n", __func__, k); #endif p->DETBufferSizeY[k] = p->DETBufferSizeInKByte[k] * 1024 / 2; p->DETBufferSizeC[k] = p->DETBufferSizeInKByte[k] * 1024 / 2; } else { #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, 2/3 DET will be used for plane0, and 1/3 for plane1\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: k=%u, 2/3 DET will be used for plane0, and 1/3 for plane1\n", __func__, k); #endif p->DETBufferSizeY[k] = (unsigned int)(math_floor2(p->DETBufferSizeInKByte[k] * 1024 * 2 / 3, 1024)); p->DETBufferSizeC[k] = p->DETBufferSizeInKByte[k] * 1024 - p->DETBufferSizeY[k]; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u SwathHeightY = %u\n", __func__, k, p->SwathHeightY[k]); - dml2_printf("DML::%s: k=%u SwathHeightC = %u\n", __func__, k, p->SwathHeightC[k]); - dml2_printf("DML::%s: k=%u full_swath_bytes_l = %u\n", __func__, k, p->full_swath_bytes_l[k]); - dml2_printf("DML::%s: k=%u full_swath_bytes_c = %u\n", __func__, k, p->full_swath_bytes_c[k]); - dml2_printf("DML::%s: k=%u RoundedUpSwathSizeBytesY = %u\n", __func__, k, RoundedUpSwathSizeBytesY[k]); - dml2_printf("DML::%s: k=%u RoundedUpSwathSizeBytesC = %u\n", __func__, k, RoundedUpSwathSizeBytesC[k]); - dml2_printf("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, p->DETBufferSizeInKByte[k]); - dml2_printf("DML::%s: k=%u DETBufferSizeY = %u\n", __func__, k, p->DETBufferSizeY[k]); - dml2_printf("DML::%s: k=%u DETBufferSizeC = %u\n", __func__, k, p->DETBufferSizeC[k]); - dml2_printf("DML::%s: k=%u ViewportSizeSupportPerSurface = %u\n", __func__, k, p->ViewportSizeSupportPerSurface[k]); + DML_LOG_VERBOSE("DML::%s: k=%u SwathHeightY = %u\n", __func__, k, p->SwathHeightY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u SwathHeightC = %u\n", __func__, k, p->SwathHeightC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_l = %u\n", __func__, k, p->full_swath_bytes_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_c = %u\n", __func__, k, p->full_swath_bytes_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%u RoundedUpSwathSizeBytesY = %u\n", __func__, k, RoundedUpSwathSizeBytesY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u RoundedUpSwathSizeBytesC = %u\n", __func__, k, RoundedUpSwathSizeBytesC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, p->DETBufferSizeInKByte[k]); + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeY = %u\n", __func__, k, p->DETBufferSizeY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeC = %u\n", __func__, k, p->DETBufferSizeC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u ViewportSizeSupportPerSurface = %u\n", __func__, k, p->ViewportSizeSupportPerSurface[k]); #endif } @@ -3969,12 +3905,12 @@ static void CalculateSwathAndDETConfiguration(struct dml2_core_internal_scratch *p->compbuf_reserved_space_64b = (unsigned int)math_ceil2(math_max2(*p->compbuf_reserved_space_64b, (double)(p->rob_buffer_size_kbytes * 1024 / 64) - (double)(RoundedUpSwathSizeBytesY[SurfaceDoingUnboundedRequest] * TTUFIFODEPTH / (p->mrq_present ? MAXIMUMCOMPRESSION : 1) / 64)), 1.0); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: RoundedUpSwathSizeBytesY[%d] = %u\n", __func__, SurfaceDoingUnboundedRequest, RoundedUpSwathSizeBytesY[SurfaceDoingUnboundedRequest]); - dml2_printf("DML::%s: rob_buffer_size_kbytes = %u\n", __func__, p->rob_buffer_size_kbytes); + DML_LOG_VERBOSE("DML::%s: RoundedUpSwathSizeBytesY[%d] = %u\n", __func__, SurfaceDoingUnboundedRequest, RoundedUpSwathSizeBytesY[SurfaceDoingUnboundedRequest]); + DML_LOG_VERBOSE("DML::%s: rob_buffer_size_kbytes = %u\n", __func__, p->rob_buffer_size_kbytes); #endif } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: compbuf_reserved_space_64b = %u\n", __func__, *p->compbuf_reserved_space_64b); + DML_LOG_VERBOSE("DML::%s: compbuf_reserved_space_64b = %u\n", __func__, *p->compbuf_reserved_space_64b); #endif *p->hw_debug5 = false; @@ -3989,12 +3925,12 @@ static void CalculateSwathAndDETConfiguration(struct dml2_core_internal_scratch + *p->CompressedBufferSizeInkByte * MAXIMUMCOMPRESSION * 1024) > TTUFIFODEPTH * (RoundedUpSwathSizeBytesY[k] + RoundedUpSwathSizeBytesC[k]))) *p->hw_debug5 = true; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u UnboundedRequestEnabled = %u\n", __func__, k, *p->UnboundedRequestEnabled); - dml2_printf("DML::%s: k=%u MAXIMUMCOMPRESSION = %lu\n", __func__, k, MAXIMUMCOMPRESSION); - dml2_printf("DML::%s: k=%u TTUFIFODEPTH = %lu\n", __func__, k, TTUFIFODEPTH); - dml2_printf("DML::%s: k=%u CompressedBufferSizeInkByte = %u\n", __func__, k, *p->CompressedBufferSizeInkByte); - dml2_printf("DML::%s: k=%u RoundedUpSwathSizeBytesC = %u\n", __func__, k, RoundedUpSwathSizeBytesC[k]); - dml2_printf("DML::%s: k=%u hw_debug5 = %u\n", __func__, k, *p->hw_debug5); + DML_LOG_VERBOSE("DML::%s: k=%u UnboundedRequestEnabled = %u\n", __func__, k, *p->UnboundedRequestEnabled); + DML_LOG_VERBOSE("DML::%s: k=%u MAXIMUMCOMPRESSION = %lu\n", __func__, k, MAXIMUMCOMPRESSION); + DML_LOG_VERBOSE("DML::%s: k=%u TTUFIFODEPTH = %lu\n", __func__, k, TTUFIFODEPTH); + DML_LOG_VERBOSE("DML::%s: k=%u CompressedBufferSizeInkByte = %u\n", __func__, k, *p->CompressedBufferSizeInkByte); + DML_LOG_VERBOSE("DML::%s: k=%u RoundedUpSwathSizeBytesC = %u\n", __func__, k, RoundedUpSwathSizeBytesC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u hw_debug5 = %u\n", __func__, k, *p->hw_debug5); #endif } #endif @@ -4192,15 +4128,15 @@ static noinline_for_stack void CalculateODMMode( SurfaceRequiredDISPCLKWithODMCombineThreeToOne = CalculateRequiredDispclk(dml2_odm_mode_combine_3to1, PixelClock); SurfaceRequiredDISPCLKWithODMCombineFourToOne = CalculateRequiredDispclk(dml2_odm_mode_combine_4to1, PixelClock); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: ODMUse = %d\n", __func__, ODMUse); - dml2_printf("DML::%s: Output = %d\n", __func__, Output); - dml2_printf("DML::%s: DSCEnable = %d\n", __func__, DSCEnable); - dml2_printf("DML::%s: MaxDispclk = %f\n", __func__, MaxDispclk); - dml2_printf("DML::%s: MaximumPixelsPerLinePerDSCUnit = %d\n", __func__, MaximumPixelsPerLinePerDSCUnit); - dml2_printf("DML::%s: SurfaceRequiredDISPCLKWithoutODMCombine = %f\n", __func__, SurfaceRequiredDISPCLKWithoutODMCombine); - dml2_printf("DML::%s: SurfaceRequiredDISPCLKWithODMCombineTwoToOne = %f\n", __func__, SurfaceRequiredDISPCLKWithODMCombineTwoToOne); - dml2_printf("DML::%s: SurfaceRequiredDISPCLKWithODMCombineThreeToOne = %f\n", __func__, SurfaceRequiredDISPCLKWithODMCombineThreeToOne); - dml2_printf("DML::%s: SurfaceRequiredDISPCLKWithODMCombineFourToOne = %f\n", __func__, SurfaceRequiredDISPCLKWithODMCombineFourToOne); + DML_LOG_VERBOSE("DML::%s: ODMUse = %d\n", __func__, ODMUse); + DML_LOG_VERBOSE("DML::%s: Output = %d\n", __func__, Output); + DML_LOG_VERBOSE("DML::%s: DSCEnable = %d\n", __func__, DSCEnable); + DML_LOG_VERBOSE("DML::%s: MaxDispclk = %f\n", __func__, MaxDispclk); + DML_LOG_VERBOSE("DML::%s: MaximumPixelsPerLinePerDSCUnit = %d\n", __func__, MaximumPixelsPerLinePerDSCUnit); + DML_LOG_VERBOSE("DML::%s: SurfaceRequiredDISPCLKWithoutODMCombine = %f\n", __func__, SurfaceRequiredDISPCLKWithoutODMCombine); + DML_LOG_VERBOSE("DML::%s: SurfaceRequiredDISPCLKWithODMCombineTwoToOne = %f\n", __func__, SurfaceRequiredDISPCLKWithODMCombineTwoToOne); + DML_LOG_VERBOSE("DML::%s: SurfaceRequiredDISPCLKWithODMCombineThreeToOne = %f\n", __func__, SurfaceRequiredDISPCLKWithODMCombineThreeToOne); + DML_LOG_VERBOSE("DML::%s: SurfaceRequiredDISPCLKWithODMCombineFourToOne = %f\n", __func__, SurfaceRequiredDISPCLKWithODMCombineFourToOne); #endif if (ODMUse == dml2_odm_mode_auto) DecidedODMMode = DecideODMMode(HActive, @@ -4245,10 +4181,10 @@ static noinline_for_stack void CalculateODMMode( *NumberOfDPP = NumberOfDPPRequired; *RequiredDISPCLKPerSurface = success ? DISPCLKRequired : 0; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: ODMMode = %d\n", __func__, *ODMMode); - dml2_printf("DML::%s: NumberOfDPP = %d\n", __func__, *NumberOfDPP); - dml2_printf("DML::%s: TotalAvailablePipesSupport = %d\n", __func__, *TotalAvailablePipesSupport); - dml2_printf("DML::%s: RequiredDISPCLKPerSurface = %f\n", __func__, *RequiredDISPCLKPerSurface); + DML_LOG_VERBOSE("DML::%s: ODMMode = %d\n", __func__, *ODMMode); + DML_LOG_VERBOSE("DML::%s: NumberOfDPP = %d\n", __func__, *NumberOfDPP); + DML_LOG_VERBOSE("DML::%s: TotalAvailablePipesSupport = %d\n", __func__, *TotalAvailablePipesSupport); + DML_LOG_VERBOSE("DML::%s: RequiredDISPCLKPerSurface = %f\n", __func__, *RequiredDISPCLKPerSurface); #endif } @@ -4292,17 +4228,17 @@ static noinline_for_stack void CalculateOutputLink( *OutputRate = dml2_core_internal_output_rate_unknown; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: DSCEnable = %u (dis, en, en_if_necessary)\n", __func__, DSCEnable); - dml2_printf("DML::%s: PHYCLK = %f\n", __func__, PHYCLK); - dml2_printf("DML::%s: PixelClockBackEnd = %f\n", __func__, PixelClockBackEnd); - dml2_printf("DML::%s: AudioSampleRate = %f\n", __func__, AudioSampleRate); - dml2_printf("DML::%s: HActive = %u\n", __func__, HActive); - dml2_printf("DML::%s: HTotal = %u\n", __func__, HTotal); - dml2_printf("DML::%s: ODMModeNoDSC = %u\n", __func__, ODMModeNoDSC); - dml2_printf("DML::%s: ODMModeDSC = %u\n", __func__, ODMModeDSC); - dml2_printf("DML::%s: ForcedOutputLinkBPP = %f\n", __func__, ForcedOutputLinkBPP); - dml2_printf("DML::%s: Output (encoder) = %u\n", __func__, Output); - dml2_printf("DML::%s: OutputLinkDPRate = %u\n", __func__, OutputLinkDPRate); + DML_LOG_VERBOSE("DML::%s: DSCEnable = %u (dis, en, en_if_necessary)\n", __func__, DSCEnable); + DML_LOG_VERBOSE("DML::%s: PHYCLK = %f\n", __func__, PHYCLK); + DML_LOG_VERBOSE("DML::%s: PixelClockBackEnd = %f\n", __func__, PixelClockBackEnd); + DML_LOG_VERBOSE("DML::%s: AudioSampleRate = %f\n", __func__, AudioSampleRate); + DML_LOG_VERBOSE("DML::%s: HActive = %u\n", __func__, HActive); + DML_LOG_VERBOSE("DML::%s: HTotal = %u\n", __func__, HTotal); + DML_LOG_VERBOSE("DML::%s: ODMModeNoDSC = %u\n", __func__, ODMModeNoDSC); + DML_LOG_VERBOSE("DML::%s: ODMModeDSC = %u\n", __func__, ODMModeDSC); + DML_LOG_VERBOSE("DML::%s: ForcedOutputLinkBPP = %f\n", __func__, ForcedOutputLinkBPP); + DML_LOG_VERBOSE("DML::%s: Output (encoder) = %u\n", __func__, Output); + DML_LOG_VERBOSE("DML::%s: OutputLinkDPRate = %u\n", __func__, OutputLinkDPRate); #endif { if (Output == dml2_hdmi) { @@ -4487,9 +4423,9 @@ static noinline_for_stack void CalculateOutputLink( } } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: RequiresDSC = %u\n", __func__, *RequiresDSC); - dml2_printf("DML::%s: RequiresFEC = %u\n", __func__, *RequiresFEC); - dml2_printf("DML::%s: OutBpp = %f\n", __func__, *OutBpp); + DML_LOG_VERBOSE("DML::%s: RequiresDSC = %u\n", __func__, *RequiresDSC); + DML_LOG_VERBOSE("DML::%s: RequiresFEC = %u\n", __func__, *RequiresFEC); + DML_LOG_VERBOSE("DML::%s: OutBpp = %f\n", __func__, *OutBpp); #endif } @@ -4571,17 +4507,17 @@ static unsigned int DSCDelayRequirement( DSCDelayRequirement_val = 0; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: DSCEnabled= %u\n", __func__, DSCEnabled); - dml2_printf("DML::%s: ODMMode = %u\n", __func__, ODMMode); - dml2_printf("DML::%s: OutputBpp = %f\n", __func__, OutputBpp); - dml2_printf("DML::%s: HActive = %u\n", __func__, HActive); - dml2_printf("DML::%s: HTotal = %u\n", __func__, HTotal); - dml2_printf("DML::%s: PixelClock = %f\n", __func__, PixelClock); - dml2_printf("DML::%s: PixelClockBackEnd = %f\n", __func__, PixelClockBackEnd); - dml2_printf("DML::%s: OutputFormat = %u\n", __func__, OutputFormat); - dml2_printf("DML::%s: DSCInputBitPerComponent = %u\n", __func__, DSCInputBitPerComponent); - dml2_printf("DML::%s: NumberOfDSCSlices = %u\n", __func__, NumberOfDSCSlices); - dml2_printf("DML::%s: DSCDelayRequirement_val = %u\n", __func__, DSCDelayRequirement_val); + DML_LOG_VERBOSE("DML::%s: DSCEnabled= %u\n", __func__, DSCEnabled); + DML_LOG_VERBOSE("DML::%s: ODMMode = %u\n", __func__, ODMMode); + DML_LOG_VERBOSE("DML::%s: OutputBpp = %f\n", __func__, OutputBpp); + DML_LOG_VERBOSE("DML::%s: HActive = %u\n", __func__, HActive); + DML_LOG_VERBOSE("DML::%s: HTotal = %u\n", __func__, HTotal); + DML_LOG_VERBOSE("DML::%s: PixelClock = %f\n", __func__, PixelClock); + DML_LOG_VERBOSE("DML::%s: PixelClockBackEnd = %f\n", __func__, PixelClockBackEnd); + DML_LOG_VERBOSE("DML::%s: OutputFormat = %u\n", __func__, OutputFormat); + DML_LOG_VERBOSE("DML::%s: DSCInputBitPerComponent = %u\n", __func__, DSCInputBitPerComponent); + DML_LOG_VERBOSE("DML::%s: NumberOfDSCSlices = %u\n", __func__, NumberOfDSCSlices); + DML_LOG_VERBOSE("DML::%s: DSCDelayRequirement_val = %u\n", __func__, DSCDelayRequirement_val); #endif return DSCDelayRequirement_val; @@ -4654,10 +4590,10 @@ static void CalculateSurfaceSizeInMall( (TotalSurfaceSizeInMALLForSubVP > MALLAllocatedForDCNInBytes); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: MALLAllocatedForDCN = %u\n", __func__, MALLAllocatedForDCN * 1024 * 1024); - dml2_printf("DML::%s: TotalSurfaceSizeInMALLForSubVP = %u\n", __func__, TotalSurfaceSizeInMALLForSubVP); - dml2_printf("DML::%s: TotalSurfaceSizeInMALLForSS = %u\n", __func__, TotalSurfaceSizeInMALLForSS); - dml2_printf("DML::%s: ExceededMALLSize = %u\n", __func__, *ExceededMALLSize); + DML_LOG_VERBOSE("DML::%s: MALLAllocatedForDCN = %u\n", __func__, MALLAllocatedForDCN * 1024 * 1024); + DML_LOG_VERBOSE("DML::%s: TotalSurfaceSizeInMALLForSubVP = %u\n", __func__, TotalSurfaceSizeInMALLForSubVP); + DML_LOG_VERBOSE("DML::%s: TotalSurfaceSizeInMALLForSS = %u\n", __func__, TotalSurfaceSizeInMALLForSS); + DML_LOG_VERBOSE("DML::%s: ExceededMALLSize = %u\n", __func__, *ExceededMALLSize); #endif } @@ -4674,7 +4610,6 @@ static void calculate_tdlut_setting( unsigned int tdlut_vmpg_per_frame; unsigned int tdlut_pte_req_per_frame; unsigned int tdlut_bytes_per_line; - unsigned int tdlut_delivery_cycles; double tdlut_drain_rate; unsigned int tdlut_mpc_width; unsigned int tdlut_bytes_per_group_simple; @@ -4737,44 +4672,47 @@ static void calculate_tdlut_setting( *p->tdlut_bytes_per_frame = tdlut_bytes_per_line * tdlut_mpc_width * tdlut_mpc_width; *p->tdlut_bytes_per_group = tdlut_bytes_per_line * tdlut_mpc_width; //the delivery cycles is DispClk cycles per line * number of lines * number of slices - tdlut_delivery_cycles = (unsigned int)math_ceil2(tdlut_mpc_width/2.0, 1) * tdlut_mpc_width * tdlut_mpc_width; + //tdlut_delivery_cycles = (unsigned int)math_ceil2(tdlut_mpc_width/2.0, 1) * tdlut_mpc_width * tdlut_mpc_width; tdlut_drain_rate = tdlut_bytes_per_line * p->dispclk_mhz / math_ceil2(tdlut_mpc_width/2.0, 1); } else { //tdlut_addressing_mode = tdlut_simple_linear, 3dlut width should be 4*1229=4916 elements *p->tdlut_bytes_per_frame = (unsigned int)math_ceil2(tdlut_width * tdlut_bpe, 256); *p->tdlut_bytes_per_group = tdlut_bytes_per_group_simple; - tdlut_delivery_cycles = (unsigned int)math_ceil2(tdlut_width/2.0, 1); + //tdlut_delivery_cycles = (unsigned int)math_ceil2(tdlut_width/2.0, 1); tdlut_drain_rate = 2 * tdlut_bpe * p->dispclk_mhz; } //the tdlut is fetched during the 2 row times of prefetch. if (p->setup_for_tdlut) { *p->tdlut_groups_per_2row_ub = (unsigned int)math_ceil2((double) *p->tdlut_bytes_per_frame / *p->tdlut_bytes_per_group, 1); - *p->tdlut_opt_time = (*p->tdlut_bytes_per_frame - p->cursor_buffer_size * 1024) / tdlut_drain_rate; + if (*p->tdlut_bytes_per_frame > p->cursor_buffer_size * 1024) + *p->tdlut_opt_time = (*p->tdlut_bytes_per_frame - p->cursor_buffer_size * 1024) / tdlut_drain_rate; + else + *p->tdlut_opt_time = 0; *p->tdlut_drain_time = p->cursor_buffer_size * 1024 / tdlut_drain_rate; *p->tdlut_bytes_to_deliver = (unsigned int) (p->cursor_buffer_size * 1024.0); } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: gpuvm_enable = %d\n", __func__, p->gpuvm_enable); - dml2_printf("DML::%s: vmpg_bytes = %d\n", __func__, vmpg_bytes); - dml2_printf("DML::%s: tdlut_vmpg_per_frame = %d\n", __func__, tdlut_vmpg_per_frame); - dml2_printf("DML::%s: tdlut_pte_req_per_frame = %d\n", __func__, tdlut_pte_req_per_frame); + DML_LOG_VERBOSE("DML::%s: gpuvm_enable = %d\n", __func__, p->gpuvm_enable); + DML_LOG_VERBOSE("DML::%s: vmpg_bytes = %d\n", __func__, vmpg_bytes); + DML_LOG_VERBOSE("DML::%s: tdlut_vmpg_per_frame = %d\n", __func__, tdlut_vmpg_per_frame); + DML_LOG_VERBOSE("DML::%s: tdlut_pte_req_per_frame = %d\n", __func__, tdlut_pte_req_per_frame); - dml2_printf("DML::%s: dispclk_mhz = %f\n", __func__, p->dispclk_mhz); - dml2_printf("DML::%s: tdlut_width = %u\n", __func__, tdlut_width); - dml2_printf("DML::%s: tdlut_addressing_mode = %s\n", __func__, (p->tdlut_addressing_mode == dml2_tdlut_sw_linear) ? "sw_linear" : "simple_linear"); - dml2_printf("DML::%s: tdlut_pitch_bytes = %u\n", __func__, tdlut_pitch_bytes); - dml2_printf("DML::%s: tdlut_footprint_bytes = %u\n", __func__, tdlut_footprint_bytes); - dml2_printf("DML::%s: tdlut_bytes_per_frame = %u\n", __func__, *p->tdlut_bytes_per_frame); - dml2_printf("DML::%s: tdlut_bytes_per_line = %u\n", __func__, tdlut_bytes_per_line); - dml2_printf("DML::%s: tdlut_bytes_per_group = %u\n", __func__, *p->tdlut_bytes_per_group); - dml2_printf("DML::%s: tdlut_drain_rate = %f\n", __func__, tdlut_drain_rate); - dml2_printf("DML::%s: tdlut_delivery_cycles = %u\n", __func__, tdlut_delivery_cycles); - dml2_printf("DML::%s: tdlut_opt_time = %f\n", __func__, *p->tdlut_opt_time); - dml2_printf("DML::%s: tdlut_drain_time = %f\n", __func__, *p->tdlut_drain_time); - dml2_printf("DML::%s: tdlut_bytes_to_deliver = %d\n", __func__, *p->tdlut_bytes_to_deliver); - dml2_printf("DML::%s: tdlut_groups_per_2row_ub = %d\n", __func__, *p->tdlut_groups_per_2row_ub); + DML_LOG_VERBOSE("DML::%s: dispclk_mhz = %f\n", __func__, p->dispclk_mhz); + DML_LOG_VERBOSE("DML::%s: tdlut_width = %u\n", __func__, tdlut_width); + DML_LOG_VERBOSE("DML::%s: tdlut_addressing_mode = %s\n", __func__, (p->tdlut_addressing_mode == dml2_tdlut_sw_linear) ? "sw_linear" : "simple_linear"); + DML_LOG_VERBOSE("DML::%s: tdlut_pitch_bytes = %u\n", __func__, tdlut_pitch_bytes); + DML_LOG_VERBOSE("DML::%s: tdlut_footprint_bytes = %u\n", __func__, tdlut_footprint_bytes); + DML_LOG_VERBOSE("DML::%s: tdlut_bytes_per_frame = %u\n", __func__, *p->tdlut_bytes_per_frame); + DML_LOG_VERBOSE("DML::%s: tdlut_bytes_per_line = %u\n", __func__, tdlut_bytes_per_line); + DML_LOG_VERBOSE("DML::%s: tdlut_bytes_per_group = %u\n", __func__, *p->tdlut_bytes_per_group); + DML_LOG_VERBOSE("DML::%s: tdlut_drain_rate = %f\n", __func__, tdlut_drain_rate); + DML_LOG_VERBOSE("DML::%s: tdlut_delivery_cycles = %u\n", __func__, p->tdlut_addressing_mode == dml2_tdlut_sw_linear ? (unsigned int)math_ceil2(tdlut_mpc_width/2.0, 1) * tdlut_mpc_width * tdlut_mpc_width : (unsigned int)math_ceil2(tdlut_width/2.0, 1)); + DML_LOG_VERBOSE("DML::%s: tdlut_opt_time = %f\n", __func__, *p->tdlut_opt_time); + DML_LOG_VERBOSE("DML::%s: tdlut_drain_time = %f\n", __func__, *p->tdlut_drain_time); + DML_LOG_VERBOSE("DML::%s: tdlut_bytes_to_deliver = %d\n", __func__, *p->tdlut_bytes_to_deliver); + DML_LOG_VERBOSE("DML::%s: tdlut_groups_per_2row_ub = %d\n", __func__, *p->tdlut_groups_per_2row_ub); #endif } @@ -4820,10 +4758,10 @@ static void CalculateTarb( *Tarb = extra_bytes / ReturnBW; *Tarb_prefetch = extra_bytes_prefetch / ReturnBW; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: PixelChunkSizeInKByte = %d\n", __func__, PixelChunkSizeInKByte); - dml2_printf("DML::%s: MetaChunkSize = %d\n", __func__, MetaChunkSize); - dml2_printf("DML::%s: extra_bytes = %f\n", __func__, extra_bytes); - dml2_printf("DML::%s: extra_bytes_prefetch = %f\n", __func__, extra_bytes_prefetch); + DML_LOG_VERBOSE("DML::%s: PixelChunkSizeInKByte = %d\n", __func__, PixelChunkSizeInKByte); + DML_LOG_VERBOSE("DML::%s: MetaChunkSize = %d\n", __func__, MetaChunkSize); + DML_LOG_VERBOSE("DML::%s: extra_bytes = %f\n", __func__, extra_bytes); + DML_LOG_VERBOSE("DML::%s: extra_bytes_prefetch = %f\n", __func__, extra_bytes_prefetch); #endif } @@ -4838,10 +4776,10 @@ static double CalculateTWait( TWait = math_max2(reserved_vblank_time_ns/1000.0, g6_temp_read_blackout_us) + t_urg_trip; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: reserved_vblank_time_ns = %d\n", __func__, reserved_vblank_time_ns); - dml2_printf("DML::%s: UrgentLatency = %f\n", __func__, UrgentLatency); - dml2_printf("DML::%s: Ttrip = %f\n", __func__, Ttrip); - dml2_printf("DML::%s: TWait = %f\n", __func__, TWait); + DML_LOG_VERBOSE("DML::%s: reserved_vblank_time_ns = %ld\n", __func__, reserved_vblank_time_ns); + DML_LOG_VERBOSE("DML::%s: UrgentLatency = %f\n", __func__, UrgentLatency); + DML_LOG_VERBOSE("DML::%s: Ttrip = %f\n", __func__, Ttrip); + DML_LOG_VERBOSE("DML::%s: TWait = %f\n", __func__, TWait); #endif return TWait; } @@ -4887,20 +4825,20 @@ static void CalculateVUpdateAndDynamicMetadataParameters( *Tdmsks = *Tdmsks / 2; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: DynamicMetadataLinesBeforeActiveRequired = %u\n", __func__, DynamicMetadataLinesBeforeActiveRequired); - dml2_printf("DML::%s: VBlank = %u\n", __func__, VBlank); - dml2_printf("DML::%s: HTotal = %u\n", __func__, HTotal); - dml2_printf("DML::%s: PixelClock = %f\n", __func__, PixelClock); - dml2_printf("DML::%s: Dppclk = %f\n", __func__, Dppclk); - dml2_printf("DML::%s: DCFClkDeepSleep = %f\n", __func__, DCFClkDeepSleep); - dml2_printf("DML::%s: MaxInterDCNTileRepeaters = %u\n", __func__, MaxInterDCNTileRepeaters); - dml2_printf("DML::%s: TotalRepeaterDelayTime = %f\n", __func__, TotalRepeaterDelayTime); + DML_LOG_VERBOSE("DML::%s: DynamicMetadataLinesBeforeActiveRequired = %u\n", __func__, DynamicMetadataLinesBeforeActiveRequired); + DML_LOG_VERBOSE("DML::%s: VBlank = %u\n", __func__, VBlank); + DML_LOG_VERBOSE("DML::%s: HTotal = %u\n", __func__, HTotal); + DML_LOG_VERBOSE("DML::%s: PixelClock = %f\n", __func__, PixelClock); + DML_LOG_VERBOSE("DML::%s: Dppclk = %f\n", __func__, Dppclk); + DML_LOG_VERBOSE("DML::%s: DCFClkDeepSleep = %f\n", __func__, DCFClkDeepSleep); + DML_LOG_VERBOSE("DML::%s: MaxInterDCNTileRepeaters = %u\n", __func__, MaxInterDCNTileRepeaters); + DML_LOG_VERBOSE("DML::%s: TotalRepeaterDelayTime = %f\n", __func__, TotalRepeaterDelayTime); - dml2_printf("DML::%s: VUpdateWidthPix = %u\n", __func__, *VUpdateWidthPix); - dml2_printf("DML::%s: VReadyOffsetPix = %u\n", __func__, *VReadyOffsetPix); - dml2_printf("DML::%s: VUpdateOffsetPix = %u\n", __func__, *VUpdateOffsetPix); + DML_LOG_VERBOSE("DML::%s: VUpdateWidthPix = %u\n", __func__, *VUpdateWidthPix); + DML_LOG_VERBOSE("DML::%s: VReadyOffsetPix = %u\n", __func__, *VReadyOffsetPix); + DML_LOG_VERBOSE("DML::%s: VUpdateOffsetPix = %u\n", __func__, *VUpdateOffsetPix); - dml2_printf("DML::%s: Tdmsks = %f\n", __func__, *Tdmsks); + DML_LOG_VERBOSE("DML::%s: Tdmsks = %f\n", __func__, *Tdmsks); #endif } @@ -4923,7 +4861,7 @@ static double get_urgent_bandwidth_required( double ReadBandwidthChroma[], double PrefetchBandwidthLuma[], double PrefetchBandwidthChroma[], - double PrefetchBandwidthOto[], + double PrefetchBandwidthMax[], double excess_vactive_fill_bw_l[], double excess_vactive_fill_bw_c[], double cursor_bw[], @@ -4962,11 +4900,11 @@ static double get_urgent_bandwidth_required( l->adj_factor_cur_pre = UrgentBurstFactorCursorPre[k]; bool is_phantom = dml_is_phantom_pipe(&display_cfg->plane_descriptors[k]); - bool exclude_this_plane = 0; + bool exclude_this_plane = false; // Exclude phantom pipe in bw calculation for non svp prefetch state if (state_type != dml2_core_internal_soc_state_svp_prefetch && is_phantom) - exclude_this_plane = 1; + exclude_this_plane = true; // The qualified row bandwidth, qual_row_bw, accounts for the regular non-flip row bandwidth when there is no possible immediate flip or HostVM invalidation flip. // The qual_row_bw is zero if HostVM is possible and only non-zero and equal to row_bw(i) if immediate flip is not allowed for that pipe. @@ -4987,20 +4925,20 @@ static double get_urgent_bandwidth_required( l->vm_row_bw = NumberOfDPP[k] * prefetch_vmrow_bw[k]; l->flip_and_active_bw = l->per_plane_flip_bw[k] + ReadBandwidthLuma[k] * l->adj_factor_p0 + ReadBandwidthChroma[k] * l->adj_factor_p1 + cursor_bw[k] * l->adj_factor_cur; l->flip_and_prefetch_bw = l->per_plane_flip_bw[k] + NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * l->adj_factor_p0_pre + PrefetchBandwidthChroma[k] * l->adj_factor_p1_pre) + prefetch_cursor_bw[k] * l->adj_factor_cur_pre; - l->flip_and_prefetch_bw_oto = l->per_plane_flip_bw[k] + NumberOfDPP[k] * (PrefetchBandwidthOto[k] * l->adj_factor_p0_pre + PrefetchBandwidthChroma[k] * l->adj_factor_p1_pre) + prefetch_cursor_bw[k] * l->adj_factor_cur_pre; + l->flip_and_prefetch_bw_max = l->per_plane_flip_bw[k] + NumberOfDPP[k] * (PrefetchBandwidthMax[k] * l->adj_factor_p0_pre + PrefetchBandwidthChroma[k] * l->adj_factor_p1_pre) + prefetch_cursor_bw[k] * l->adj_factor_cur_pre; l->active_and_excess_bw = (ReadBandwidthLuma[k] + excess_vactive_fill_bw_l[k]) * l->tmp_nom_adj_factor_p0 + (ReadBandwidthChroma[k] + excess_vactive_fill_bw_c[k]) * l->tmp_nom_adj_factor_p1 + dpte_row_bw[k] + meta_row_bw[k]; - surface_required_bw[k] = math_max5(l->vm_row_bw, l->flip_and_active_bw, l->flip_and_prefetch_bw, l->active_and_excess_bw, l->flip_and_prefetch_bw_oto); + surface_required_bw[k] = math_max5(l->vm_row_bw, l->flip_and_active_bw, l->flip_and_prefetch_bw, l->active_and_excess_bw, l->flip_and_prefetch_bw_max); /* export peak required bandwidth for the surface */ surface_peak_required_bw[k] = math_max2(surface_required_bw[k], surface_peak_required_bw[k]); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%d, max1: vm_row_bw=%f\n", __func__, k, l->vm_row_bw); - dml2_printf("DML::%s: k=%d, max2: flip_and_active_bw=%f\n", __func__, k, l->flip_and_active_bw); - dml2_printf("DML::%s: k=%d, max3: flip_and_prefetch_bw=%f\n", __func__, k, l->flip_and_prefetch_bw); - dml2_printf("DML::%s: k=%d, max4: active_and_excess_bw=%f\n", __func__, k, l->active_and_excess_bw); - dml2_printf("DML::%s: k=%d, surface_required_bw=%f\n", __func__, k, surface_required_bw[k]); - dml2_printf("DML::%s: k=%d, surface_peak_required_bw=%f\n", __func__, k, surface_peak_required_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, max1: vm_row_bw=%f\n", __func__, k, l->vm_row_bw); + DML_LOG_VERBOSE("DML::%s: k=%d, max2: flip_and_active_bw=%f\n", __func__, k, l->flip_and_active_bw); + DML_LOG_VERBOSE("DML::%s: k=%d, max3: flip_and_prefetch_bw=%f\n", __func__, k, l->flip_and_prefetch_bw); + DML_LOG_VERBOSE("DML::%s: k=%d, max4: active_and_excess_bw=%f\n", __func__, k, l->active_and_excess_bw); + DML_LOG_VERBOSE("DML::%s: k=%d, surface_required_bw=%f\n", __func__, k, surface_required_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, surface_peak_required_bw=%f\n", __func__, k, surface_peak_required_bw[k]); #endif } else { surface_required_bw[k] = 0.0; @@ -5009,34 +4947,34 @@ static double get_urgent_bandwidth_required( l->required_bandwidth_mbps += surface_required_bw[k]; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%d, NumberOfDPP=%d\n", __func__, k, NumberOfDPP[k]); - dml2_printf("DML::%s: k=%d, use_qual_row_bw=%d\n", __func__, k, use_qual_row_bw); - dml2_printf("DML::%s: k=%d, immediate_flip=%d\n", __func__, k, display_cfg->plane_descriptors[k].immediate_flip); - dml2_printf("DML::%s: k=%d, mall_svp_prefetch_factor=%f\n", __func__, k, l->mall_svp_prefetch_factor); - dml2_printf("DML::%s: k=%d, adj_factor_p0=%f\n", __func__, k, l->adj_factor_p0); - dml2_printf("DML::%s: k=%d, adj_factor_p1=%f\n", __func__, k, l->adj_factor_p1); - dml2_printf("DML::%s: k=%d, adj_factor_cur=%f\n", __func__, k, l->adj_factor_cur); + DML_LOG_VERBOSE("DML::%s: k=%d, NumberOfDPP=%d\n", __func__, k, NumberOfDPP[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, use_qual_row_bw=%d\n", __func__, k, use_qual_row_bw); + DML_LOG_VERBOSE("DML::%s: k=%d, immediate_flip=%d\n", __func__, k, display_cfg->plane_descriptors[k].immediate_flip); + DML_LOG_VERBOSE("DML::%s: k=%d, mall_svp_prefetch_factor=%f\n", __func__, k, l->mall_svp_prefetch_factor); + DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_p0=%f\n", __func__, k, l->adj_factor_p0); + DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_p1=%f\n", __func__, k, l->adj_factor_p1); + DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_cur=%f\n", __func__, k, l->adj_factor_cur); - dml2_printf("DML::%s: k=%d, adj_factor_p0_pre=%f\n", __func__, k, l->adj_factor_p0_pre); - dml2_printf("DML::%s: k=%d, adj_factor_p1_pre=%f\n", __func__, k, l->adj_factor_p1_pre); - dml2_printf("DML::%s: k=%d, adj_factor_cur_pre=%f\n", __func__, k, l->adj_factor_cur_pre); + DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_p0_pre=%f\n", __func__, k, l->adj_factor_p0_pre); + DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_p1_pre=%f\n", __func__, k, l->adj_factor_p1_pre); + DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_cur_pre=%f\n", __func__, k, l->adj_factor_cur_pre); - dml2_printf("DML::%s: k=%d, per_plane_flip_bw=%f\n", __func__, k, l->per_plane_flip_bw[k]); - dml2_printf("DML::%s: k=%d, prefetch_vmrow_bw=%f\n", __func__, k, prefetch_vmrow_bw[k]); - dml2_printf("DML::%s: k=%d, ReadBandwidthLuma=%f\n", __func__, k, ReadBandwidthLuma[k]); - dml2_printf("DML::%s: k=%d, ReadBandwidthChroma=%f\n", __func__, k, ReadBandwidthChroma[k]); - dml2_printf("DML::%s: k=%d, excess_vactive_fill_bw_l=%f\n", __func__, k, excess_vactive_fill_bw_l[k]); - dml2_printf("DML::%s: k=%d, excess_vactive_fill_bw_c=%f\n", __func__, k, excess_vactive_fill_bw_c[k]); - dml2_printf("DML::%s: k=%d, cursor_bw=%f\n", __func__, k, cursor_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, per_plane_flip_bw=%f\n", __func__, k, l->per_plane_flip_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, prefetch_vmrow_bw=%f\n", __func__, k, prefetch_vmrow_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, ReadBandwidthLuma=%f\n", __func__, k, ReadBandwidthLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, ReadBandwidthChroma=%f\n", __func__, k, ReadBandwidthChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, excess_vactive_fill_bw_l=%f\n", __func__, k, excess_vactive_fill_bw_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, excess_vactive_fill_bw_c=%f\n", __func__, k, excess_vactive_fill_bw_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, cursor_bw=%f\n", __func__, k, cursor_bw[k]); - dml2_printf("DML::%s: k=%d, meta_row_bw=%f\n", __func__, k, meta_row_bw[k]); - dml2_printf("DML::%s: k=%d, dpte_row_bw=%f\n", __func__, k, dpte_row_bw[k]); - dml2_printf("DML::%s: k=%d, PrefetchBandwidthLuma=%f\n", __func__, k, PrefetchBandwidthLuma[k]); - dml2_printf("DML::%s: k=%d, PrefetchBandwidthChroma=%f\n", __func__, k, PrefetchBandwidthChroma[k]); - dml2_printf("DML::%s: k=%d, prefetch_cursor_bw=%f\n", __func__, k, prefetch_cursor_bw[k]); - dml2_printf("DML::%s: k=%d, required_bandwidth_mbps=%f (total), inc_flip_bw=%d, is_phantom=%d exclude_this_plane=%d\n", __func__, k, l->required_bandwidth_mbps, inc_flip_bw, is_phantom, exclude_this_plane); - dml2_printf("DML::%s: k=%d, required_bandwidth_mbps=%f (total), soc_state=%s, inc_flip_bw=%d, is_phantom=%d exclude_this_plane=%d\n", __func__, k, l->required_bandwidth_mbps, dml2_core_internal_soc_state_type_str(state_type), inc_flip_bw, is_phantom, exclude_this_plane); - dml2_printf("DML::%s: k=%d, required_bandwidth_mbps=%f (total), inc_flip_bw=%d, is_phantom=%d exclude_this_plane=%d\n", __func__, k, l->required_bandwidth_mbps, inc_flip_bw, is_phantom, exclude_this_plane); + DML_LOG_VERBOSE("DML::%s: k=%d, meta_row_bw=%f\n", __func__, k, meta_row_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, dpte_row_bw=%f\n", __func__, k, dpte_row_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, PrefetchBandwidthLuma=%f\n", __func__, k, PrefetchBandwidthLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, PrefetchBandwidthChroma=%f\n", __func__, k, PrefetchBandwidthChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, prefetch_cursor_bw=%f\n", __func__, k, prefetch_cursor_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, required_bandwidth_mbps=%f (total), inc_flip_bw=%d, is_phantom=%d exclude_this_plane=%d\n", __func__, k, l->required_bandwidth_mbps, inc_flip_bw, is_phantom, exclude_this_plane); + DML_LOG_VERBOSE("DML::%s: k=%d, required_bandwidth_mbps=%f (total), soc_state=%s, inc_flip_bw=%d, is_phantom=%d exclude_this_plane=%d\n", __func__, k, l->required_bandwidth_mbps, dml2_core_internal_soc_state_type_str(state_type), inc_flip_bw, is_phantom, exclude_this_plane); + DML_LOG_VERBOSE("DML::%s: k=%d, required_bandwidth_mbps=%f (total), inc_flip_bw=%d, is_phantom=%d exclude_this_plane=%d\n", __func__, k, l->required_bandwidth_mbps, inc_flip_bw, is_phantom, exclude_this_plane); #endif } @@ -5120,19 +5058,19 @@ static void CalculateExtraLatency( *ExtraLatency_sr = *ExtraLatency_sr + Tarb; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: qos_type=%u\n", __func__, qos_type); - dml2_printf("DML::%s: hostvm_mode=%u\n", __func__, hostvm_mode); - dml2_printf("DML::%s: Tex_trips=%u\n", __func__, Tex_trips); - dml2_printf("DML::%s: max_outstanding_when_urgent_expected=%u\n", __func__, max_outstanding_when_urgent_expected); - dml2_printf("DML::%s: FabricClock=%f\n", __func__, FabricClock); - dml2_printf("DML::%s: DCFCLK=%f\n", __func__, DCFCLK); - dml2_printf("DML::%s: ReturnBW=%f\n", __func__, ReturnBW); - dml2_printf("DML::%s: RoundTripPingLatencyCycles=%u\n", __func__, RoundTripPingLatencyCycles); - dml2_printf("DML::%s: ReorderingBytes=%u\n", __func__, ReorderingBytes); - dml2_printf("DML::%s: Tarb=%f\n", __func__, Tarb); - dml2_printf("DML::%s: ExtraLatency=%f\n", __func__, *ExtraLatency); - dml2_printf("DML::%s: ExtraLatency_sr=%f\n", __func__, *ExtraLatency_sr); - dml2_printf("DML::%s: ExtraLatencyPrefetch=%f\n", __func__, *ExtraLatencyPrefetch); + DML_LOG_VERBOSE("DML::%s: qos_type=%u\n", __func__, qos_type); + DML_LOG_VERBOSE("DML::%s: hostvm_mode=%u\n", __func__, hostvm_mode); + DML_LOG_VERBOSE("DML::%s: Tex_trips=%f\n", __func__, Tex_trips); + DML_LOG_VERBOSE("DML::%s: max_outstanding_when_urgent_expected=%u\n", __func__, max_outstanding_when_urgent_expected); + DML_LOG_VERBOSE("DML::%s: FabricClock=%f\n", __func__, FabricClock); + DML_LOG_VERBOSE("DML::%s: DCFCLK=%f\n", __func__, DCFCLK); + DML_LOG_VERBOSE("DML::%s: ReturnBW=%f\n", __func__, ReturnBW); + DML_LOG_VERBOSE("DML::%s: RoundTripPingLatencyCycles=%u\n", __func__, RoundTripPingLatencyCycles); + DML_LOG_VERBOSE("DML::%s: ReorderingBytes=%u\n", __func__, ReorderingBytes); + DML_LOG_VERBOSE("DML::%s: Tarb=%f\n", __func__, Tarb); + DML_LOG_VERBOSE("DML::%s: ExtraLatency=%f\n", __func__, *ExtraLatency); + DML_LOG_VERBOSE("DML::%s: ExtraLatency_sr=%f\n", __func__, *ExtraLatency_sr); + DML_LOG_VERBOSE("DML::%s: ExtraLatencyPrefetch=%f\n", __func__, *ExtraLatencyPrefetch); #endif } @@ -5187,7 +5125,7 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch s->Tsw_est3 = 0.0; s->cursor_prefetch_bytes = 0; *p->prefetch_cursor_bw = 0; - *p->RequiredPrefetchBWOTO = 0.0; + *p->RequiredPrefetchBWMax = 0.0; dcc_mrq_enable = (p->dcc_enable && p->mrq_present); @@ -5199,20 +5137,20 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch s->HostVMDynamicLevelsTrips = 0; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: dcc_enable = %u\n", __func__, p->dcc_enable); - dml2_printf("DML::%s: mrq_present = %u\n", __func__, p->mrq_present); - dml2_printf("DML::%s: dcc_mrq_enable = %u\n", __func__, dcc_mrq_enable); - dml2_printf("DML::%s: GPUVMEnable = %u\n", __func__, p->display_cfg->gpuvm_enable); - dml2_printf("DML::%s: GPUVMPageTableLevels = %u\n", __func__, p->display_cfg->gpuvm_max_page_table_levels); - dml2_printf("DML::%s: DCCEnable = %u\n", __func__, p->myPipe->DCCEnable); - dml2_printf("DML::%s: VStartup = %u\n", __func__, p->VStartup); - dml2_printf("DML::%s: HostVMEnable = %u\n", __func__, p->display_cfg->hostvm_enable); - dml2_printf("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); - dml2_printf("DML::%s: TWait = %f\n", __func__, p->TWait); - dml2_printf("DML::%s: TWait_p = %f\n", __func__, s->TWait_p); - dml2_printf("DML::%s: Ttrip = %f\n", __func__, p->Ttrip); - dml2_printf("DML::%s: myPipe->Dppclk = %f\n", __func__, p->myPipe->Dppclk); - dml2_printf("DML::%s: myPipe->Dispclk = %f\n", __func__, p->myPipe->Dispclk); + DML_LOG_VERBOSE("DML::%s: dcc_enable = %u\n", __func__, p->dcc_enable); + DML_LOG_VERBOSE("DML::%s: mrq_present = %u\n", __func__, p->mrq_present); + DML_LOG_VERBOSE("DML::%s: dcc_mrq_enable = %u\n", __func__, dcc_mrq_enable); + DML_LOG_VERBOSE("DML::%s: GPUVMEnable = %u\n", __func__, p->display_cfg->gpuvm_enable); + DML_LOG_VERBOSE("DML::%s: GPUVMPageTableLevels = %u\n", __func__, p->display_cfg->gpuvm_max_page_table_levels); + DML_LOG_VERBOSE("DML::%s: DCCEnable = %u\n", __func__, p->myPipe->DCCEnable); + DML_LOG_VERBOSE("DML::%s: VStartup = %u\n", __func__, p->VStartup); + DML_LOG_VERBOSE("DML::%s: HostVMEnable = %u\n", __func__, p->display_cfg->hostvm_enable); + DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); + DML_LOG_VERBOSE("DML::%s: TWait = %f\n", __func__, p->TWait); + DML_LOG_VERBOSE("DML::%s: TWait_p = %f\n", __func__, s->TWait_p); + DML_LOG_VERBOSE("DML::%s: Ttrip = %f\n", __func__, p->Ttrip); + DML_LOG_VERBOSE("DML::%s: myPipe->Dppclk = %f\n", __func__, p->myPipe->Dppclk); + DML_LOG_VERBOSE("DML::%s: myPipe->Dispclk = %f\n", __func__, p->myPipe->Dispclk); #endif CalculateVUpdateAndDynamicMetadataParameters( p->MaxInterDCNTileRepeaters, @@ -5258,11 +5196,11 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch if (p->DynamicMetadataEnable == true) { if (p->VStartup * s->LineTime < *p->TSetup + *p->Tdmdl + s->Tdmbf + s->Tdmec + s->Tdmsks) { *p->NotEnoughTimeForDynamicMetadata = true; - dml2_printf("DML::%s: Not Enough Time for Dynamic Meta!\n", __func__); - dml2_printf("DML::%s: Tdmbf: %fus - time for dmd transfer from dchub to dio output buffer\n", __func__, s->Tdmbf); - dml2_printf("DML::%s: Tdmec: %fus - time dio takes to transfer dmd\n", __func__, s->Tdmec); - dml2_printf("DML::%s: Tdmsks: %fus - time before active dmd must complete transmission at dio\n", __func__, s->Tdmsks); - dml2_printf("DML::%s: Tdmdl: %fus - time for fabric to become ready and fetch dmd \n", __func__, *p->Tdmdl); + DML_LOG_VERBOSE("DML::%s: Not Enough Time for Dynamic Meta!\n", __func__); + DML_LOG_VERBOSE("DML::%s: Tdmbf: %fus - time for dmd transfer from dchub to dio output buffer\n", __func__, s->Tdmbf); + DML_LOG_VERBOSE("DML::%s: Tdmec: %fus - time dio takes to transfer dmd\n", __func__, s->Tdmec); + DML_LOG_VERBOSE("DML::%s: Tdmsks: %fus - time before active dmd must complete transmission at dio\n", __func__, s->Tdmsks); + DML_LOG_VERBOSE("DML::%s: Tdmdl: %fus - time for fabric to become ready and fetch dmd \n", __func__, *p->Tdmdl); } else { *p->NotEnoughTimeForDynamicMetadata = false; } @@ -5288,21 +5226,21 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch ((p->myPipe->ODMMode == dml2_odm_mode_mso_1to4) ? (double)p->myPipe->HActive * 3.0 / 4.0 : 0)); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: DynamicMetadataVMEnabled = %u\n", __func__, p->DynamicMetadataVMEnabled); - dml2_printf("DML::%s: DPPCycles = %u\n", __func__, s->DPPCycles); - dml2_printf("DML::%s: PixelClock = %f\n", __func__, p->myPipe->PixelClock); - dml2_printf("DML::%s: Dppclk = %f\n", __func__, p->myPipe->Dppclk); - dml2_printf("DML::%s: DISPCLKCycles = %u\n", __func__, s->DISPCLKCycles); - dml2_printf("DML::%s: DISPCLK = %f\n", __func__, p->myPipe->Dispclk); - dml2_printf("DML::%s: DSCDelay = %u\n", __func__, p->DSCDelay); - dml2_printf("DML::%s: ODMMode = %u\n", __func__, p->myPipe->ODMMode); - dml2_printf("DML::%s: DPP_RECOUT_WIDTH = %u\n", __func__, p->DPP_RECOUT_WIDTH); - dml2_printf("DML::%s: DSTXAfterScaler = %u\n", __func__, *p->DSTXAfterScaler); + DML_LOG_VERBOSE("DML::%s: DynamicMetadataVMEnabled = %u\n", __func__, p->DynamicMetadataVMEnabled); + DML_LOG_VERBOSE("DML::%s: DPPCycles = %u\n", __func__, s->DPPCycles); + DML_LOG_VERBOSE("DML::%s: PixelClock = %f\n", __func__, p->myPipe->PixelClock); + DML_LOG_VERBOSE("DML::%s: Dppclk = %f\n", __func__, p->myPipe->Dppclk); + DML_LOG_VERBOSE("DML::%s: DISPCLKCycles = %u\n", __func__, s->DISPCLKCycles); + DML_LOG_VERBOSE("DML::%s: DISPCLK = %f\n", __func__, p->myPipe->Dispclk); + DML_LOG_VERBOSE("DML::%s: DSCDelay = %u\n", __func__, p->DSCDelay); + DML_LOG_VERBOSE("DML::%s: ODMMode = %u\n", __func__, p->myPipe->ODMMode); + DML_LOG_VERBOSE("DML::%s: DPP_RECOUT_WIDTH = %u\n", __func__, p->DPP_RECOUT_WIDTH); + DML_LOG_VERBOSE("DML::%s: DSTXAfterScaler = %u\n", __func__, *p->DSTXAfterScaler); - dml2_printf("DML::%s: setup_for_tdlut = %u\n", __func__, p->setup_for_tdlut); - dml2_printf("DML::%s: tdlut_opt_time = %f\n", __func__, p->tdlut_opt_time); - dml2_printf("DML::%s: tdlut_pte_bytes_per_frame = %u\n", __func__, p->tdlut_pte_bytes_per_frame); - dml2_printf("DML::%s: tdlut_drain_time = %f\n", __func__, p->tdlut_drain_time); + DML_LOG_VERBOSE("DML::%s: setup_for_tdlut = %u\n", __func__, p->setup_for_tdlut); + DML_LOG_VERBOSE("DML::%s: tdlut_opt_time = %f\n", __func__, p->tdlut_opt_time); + DML_LOG_VERBOSE("DML::%s: tdlut_pte_bytes_per_frame = %u\n", __func__, p->tdlut_pte_bytes_per_frame); + DML_LOG_VERBOSE("DML::%s: tdlut_drain_time = %f\n", __func__, p->tdlut_drain_time); #endif if (p->OutputFormat == dml2_420 || (p->myPipe->InterlaceEnable && p->myPipe->ProgressiveToInterlaceUnitInOPP)) @@ -5314,17 +5252,17 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch *p->DSTYAfterScaler = (unsigned int)(math_floor2(s->DSTTotalPixelsAfterScaler / p->myPipe->HTotal, 1)); *p->DSTXAfterScaler = (unsigned int)(s->DSTTotalPixelsAfterScaler - ((double)(*p->DSTYAfterScaler * p->myPipe->HTotal))); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: DSTXAfterScaler = %u (final)\n", __func__, *p->DSTXAfterScaler); - dml2_printf("DML::%s: DSTYAfterScaler = %u (final)\n", __func__, *p->DSTYAfterScaler); + DML_LOG_VERBOSE("DML::%s: DSTXAfterScaler = %u (final)\n", __func__, *p->DSTXAfterScaler); + DML_LOG_VERBOSE("DML::%s: DSTYAfterScaler = %u (final)\n", __func__, *p->DSTYAfterScaler); #endif #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: Tr0_trips = %f\n", __func__, *p->Tr0_trips); - dml2_printf("DML::%s: Tvm_trips = %f\n", __func__, *p->Tvm_trips); - dml2_printf("DML::%s: trip_to_mem = %f\n", __func__, s->trip_to_mem); - dml2_printf("DML::%s: ExtraLatencyPrefetch = %f\n", __func__, p->ExtraLatencyPrefetch); - dml2_printf("DML::%s: GPUVMPageTableLevels = %u\n", __func__, p->display_cfg->gpuvm_max_page_table_levels); - dml2_printf("DML::%s: HostVMDynamicLevelsTrips = %u\n", __func__, s->HostVMDynamicLevelsTrips); + DML_LOG_VERBOSE("DML::%s: Tr0_trips = %f\n", __func__, *p->Tr0_trips); + DML_LOG_VERBOSE("DML::%s: Tvm_trips = %f\n", __func__, *p->Tvm_trips); + DML_LOG_VERBOSE("DML::%s: trip_to_mem = %f\n", __func__, s->trip_to_mem); + DML_LOG_VERBOSE("DML::%s: ExtraLatencyPrefetch = %f\n", __func__, p->ExtraLatencyPrefetch); + DML_LOG_VERBOSE("DML::%s: GPUVMPageTableLevels = %u\n", __func__, p->display_cfg->gpuvm_max_page_table_levels); + DML_LOG_VERBOSE("DML::%s: HostVMDynamicLevelsTrips = %u\n", __func__, s->HostVMDynamicLevelsTrips); #endif if (p->display_cfg->gpuvm_enable) { s->Tvm_trips_rounded = math_ceil2(4.0 * *p->Tvm_trips / s->LineTime, 1.0) / 4.0 * s->LineTime; @@ -5402,7 +5340,7 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch } /* oto prefetch bw should be always be less than total vactive bw */ - //DML2_ASSERT(s->prefetch_bw_oto < s->per_pipe_vactive_sw_bw * p->myPipe->DPPPerSurface); + //DML_ASSERT(s->prefetch_bw_oto < s->per_pipe_vactive_sw_bw * p->myPipe->DPPPerSurface); s->prefetch_bw_oto = math_max2(s->per_pipe_vactive_sw_bw, s->prefetch_bw_oto) * p->mall_prefetch_sdp_overhead_factor; @@ -5418,12 +5356,12 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch * mp will fail if ms decides to use equ schedule and mp decides to use oto schedule * and the required bandwidth increases when going from ms to mp */ - *p->RequiredPrefetchBWOTO = s->prefetch_bw_oto; + *p->RequiredPrefetchBWMax = s->prefetch_bw_oto; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: vactive_sw_bw_l = %f\n", __func__, p->vactive_sw_bw_l); - dml2_printf("DML::%s: vactive_sw_bw_c = %f\n", __func__, p->vactive_sw_bw_c); - dml2_printf("DML::%s: per_pipe_vactive_sw_bw = %f\n", __func__, s->per_pipe_vactive_sw_bw); + DML_LOG_VERBOSE("DML::%s: vactive_sw_bw_l = %f\n", __func__, p->vactive_sw_bw_l); + DML_LOG_VERBOSE("DML::%s: vactive_sw_bw_c = %f\n", __func__, p->vactive_sw_bw_c); + DML_LOG_VERBOSE("DML::%s: per_pipe_vactive_sw_bw = %f\n", __func__, s->per_pipe_vactive_sw_bw); #endif if (p->display_cfg->gpuvm_enable == true) { @@ -5433,9 +5371,9 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch s->LineTime / 4.0); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: Tvm_oto max0 = %f\n", __func__, *p->Tvm_trips); - dml2_printf("DML::%s: Tvm_oto max1 = %f\n", __func__, *p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw_oto); - dml2_printf("DML::%s: Tvm_oto max2 = %f\n", __func__, s->LineTime / 4.0); + DML_LOG_VERBOSE("DML::%s: Tvm_oto max0 = %f\n", __func__, *p->Tvm_trips); + DML_LOG_VERBOSE("DML::%s: Tvm_oto max1 = %f\n", __func__, *p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw_oto); + DML_LOG_VERBOSE("DML::%s: Tvm_oto max2 = %f\n", __func__, s->LineTime / 4.0); #endif } else { s->Tvm_oto = s->Tvm_trips_rounded; @@ -5447,9 +5385,9 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) / s->prefetch_bw_oto, s->LineTime / 4.0); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: Tr0_oto max0 = %f\n", __func__, *p->Tr0_trips); - dml2_printf("DML::%s: Tr0_oto max1 = %f\n", __func__, (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) / s->prefetch_bw_oto); - dml2_printf("DML::%s: Tr0_oto max2 = %f\n", __func__, s->LineTime / 4); + DML_LOG_VERBOSE("DML::%s: Tr0_oto max0 = %f\n", __func__, *p->Tr0_trips); + DML_LOG_VERBOSE("DML::%s: Tr0_oto max1 = %f\n", __func__, (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) / s->prefetch_bw_oto); + DML_LOG_VERBOSE("DML::%s: Tr0_oto max2 = %f\n", __func__, s->LineTime / 4); #endif } else s->Tr0_oto = s->LineTime / 4.0; @@ -5459,11 +5397,11 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch s->dst_y_prefetch_oto = s->Tvm_oto_lines + 2 * s->Tr0_oto_lines + s->Lsw_oto; #ifdef DML_GLOBAL_PREFETCH_CHECK - dml2_printf("DML::%s: impacted_Tpre = %f\n", __func__, p->impacted_dst_y_pre); + DML_LOG_VERBOSE("DML::%s: impacted_Tpre = %f\n", __func__, p->impacted_dst_y_pre); if (p->impacted_dst_y_pre > 0) { - dml2_printf("DML::%s: dst_y_prefetch_oto = %f\n", __func__, s->dst_y_prefetch_oto); + DML_LOG_VERBOSE("DML::%s: dst_y_prefetch_oto = %f\n", __func__, s->dst_y_prefetch_oto); s->dst_y_prefetch_oto = math_max2(s->dst_y_prefetch_oto, p->impacted_dst_y_pre); - dml2_printf("DML::%s: dst_y_prefetch_oto = %f (impacted)\n", __func__, s->dst_y_prefetch_oto); + DML_LOG_VERBOSE("DML::%s: dst_y_prefetch_oto = %f (impacted)\n", __func__, s->dst_y_prefetch_oto); } #endif *p->Tpre_oto = s->dst_y_prefetch_oto * s->LineTime; @@ -5492,72 +5430,71 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch s->dst_y_prefetch_equ = math_min2(s->dst_y_prefetch_equ, 63.75); // limit to the reg limit of U6.2 for DST_Y_PREFETCH #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: HTotal = %u\n", __func__, p->myPipe->HTotal); - dml2_printf("DML::%s: min_Lsw_oto = %f\n", __func__, s->min_Lsw_oto); - dml2_printf("DML::%s: min_Lsw_equ = %f\n", __func__, s->min_Lsw_equ); - dml2_printf("DML::%s: Tno_bw = %f\n", __func__, *p->Tno_bw); - dml2_printf("DML::%s: Tno_bw_flip = %f\n", __func__, *p->Tno_bw_flip); - dml2_printf("DML::%s: ExtraLatencyPrefetch = %f\n", __func__, p->ExtraLatencyPrefetch); - dml2_printf("DML::%s: trip_to_mem = %f\n", __func__, s->trip_to_mem); - dml2_printf("DML::%s: mall_prefetch_sdp_overhead_factor = %f\n", __func__, p->mall_prefetch_sdp_overhead_factor); - dml2_printf("DML::%s: BytePerPixelY = %u\n", __func__, p->myPipe->BytePerPixelY); - dml2_printf("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); - dml2_printf("DML::%s: swath_width_luma_ub = %u\n", __func__, p->swath_width_luma_ub); - dml2_printf("DML::%s: BytePerPixelC = %u\n", __func__, p->myPipe->BytePerPixelC); - dml2_printf("DML::%s: PrefetchSourceLinesC = %f\n", __func__, p->PrefetchSourceLinesC); - dml2_printf("DML::%s: swath_width_chroma_ub = %u\n", __func__, p->swath_width_chroma_ub); - dml2_printf("DML::%s: prefetch_sw_bytes = %f\n", __func__, *p->prefetch_sw_bytes); - dml2_printf("DML::%s: max_Tsw = %f\n", __func__, s->max_Tsw); - dml2_printf("DML::%s: bytes_pp = %f\n", __func__, s->bytes_pp); - dml2_printf("DML::%s: vm_bytes = %u\n", __func__, vm_bytes); - dml2_printf("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, p->PixelPTEBytesPerRow); - dml2_printf("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); - dml2_printf("DML::%s: Tvm_trips = %f\n", __func__, *p->Tvm_trips); - dml2_printf("DML::%s: Tr0_trips = %f\n", __func__, *p->Tr0_trips); - dml2_printf("DML::%s: Tvm_trips_flip = %f\n", __func__, *p->Tvm_trips_flip); - dml2_printf("DML::%s: Tr0_trips_flip = %f\n", __func__, *p->Tr0_trips_flip); - dml2_printf("DML::%s: prefetch_bw_pr = %f\n", __func__, s->prefetch_bw_pr); - dml2_printf("DML::%s: prefetch_bw_oto = %f\n", __func__, s->prefetch_bw_oto); - dml2_printf("DML::%s: Tr0_oto = %f\n", __func__, s->Tr0_oto); - dml2_printf("DML::%s: Tvm_oto = %f\n", __func__, s->Tvm_oto); - dml2_printf("DML::%s: Tvm_oto_lines = %f\n", __func__, s->Tvm_oto_lines); - dml2_printf("DML::%s: Tr0_oto_lines = %f\n", __func__, s->Tr0_oto_lines); - dml2_printf("DML::%s: Lsw_oto = %f\n", __func__, s->Lsw_oto); - dml2_printf("DML::%s: dst_y_prefetch_oto = %f\n", __func__, s->dst_y_prefetch_oto); - dml2_printf("DML::%s: dst_y_prefetch_equ = %f\n", __func__, s->dst_y_prefetch_equ); - dml2_printf("DML::%s: tdlut_row_bytes = %d\n", __func__, tdlut_row_bytes); - dml2_printf("DML::%s: meta_row_bytes = %d\n", __func__, p->meta_row_bytes); -#endif - double Tpre = s->dst_y_prefetch_equ * s->LineTime; + DML_LOG_VERBOSE("DML::%s: HTotal = %u\n", __func__, p->myPipe->HTotal); + DML_LOG_VERBOSE("DML::%s: min_Lsw_oto = %f\n", __func__, s->min_Lsw_oto); + DML_LOG_VERBOSE("DML::%s: min_Lsw_equ = %f\n", __func__, s->min_Lsw_equ); + DML_LOG_VERBOSE("DML::%s: Tno_bw = %f\n", __func__, *p->Tno_bw); + DML_LOG_VERBOSE("DML::%s: Tno_bw_flip = %f\n", __func__, *p->Tno_bw_flip); + DML_LOG_VERBOSE("DML::%s: ExtraLatencyPrefetch = %f\n", __func__, p->ExtraLatencyPrefetch); + DML_LOG_VERBOSE("DML::%s: trip_to_mem = %f\n", __func__, s->trip_to_mem); + DML_LOG_VERBOSE("DML::%s: mall_prefetch_sdp_overhead_factor = %f\n", __func__, p->mall_prefetch_sdp_overhead_factor); + DML_LOG_VERBOSE("DML::%s: BytePerPixelY = %u\n", __func__, p->myPipe->BytePerPixelY); + DML_LOG_VERBOSE("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); + DML_LOG_VERBOSE("DML::%s: swath_width_luma_ub = %u\n", __func__, p->swath_width_luma_ub); + DML_LOG_VERBOSE("DML::%s: BytePerPixelC = %u\n", __func__, p->myPipe->BytePerPixelC); + DML_LOG_VERBOSE("DML::%s: PrefetchSourceLinesC = %f\n", __func__, p->PrefetchSourceLinesC); + DML_LOG_VERBOSE("DML::%s: swath_width_chroma_ub = %u\n", __func__, p->swath_width_chroma_ub); + DML_LOG_VERBOSE("DML::%s: prefetch_sw_bytes = %f\n", __func__, *p->prefetch_sw_bytes); + DML_LOG_VERBOSE("DML::%s: max_Tsw = %f\n", __func__, s->max_Tsw); + DML_LOG_VERBOSE("DML::%s: bytes_pp = %f\n", __func__, s->bytes_pp); + DML_LOG_VERBOSE("DML::%s: vm_bytes = %u\n", __func__, vm_bytes); + DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, p->PixelPTEBytesPerRow); + DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); + DML_LOG_VERBOSE("DML::%s: Tvm_trips = %f\n", __func__, *p->Tvm_trips); + DML_LOG_VERBOSE("DML::%s: Tr0_trips = %f\n", __func__, *p->Tr0_trips); + DML_LOG_VERBOSE("DML::%s: Tvm_trips_flip = %f\n", __func__, *p->Tvm_trips_flip); + DML_LOG_VERBOSE("DML::%s: Tr0_trips_flip = %f\n", __func__, *p->Tr0_trips_flip); + DML_LOG_VERBOSE("DML::%s: prefetch_bw_pr = %f\n", __func__, s->prefetch_bw_pr); + DML_LOG_VERBOSE("DML::%s: prefetch_bw_oto = %f\n", __func__, s->prefetch_bw_oto); + DML_LOG_VERBOSE("DML::%s: Tr0_oto = %f\n", __func__, s->Tr0_oto); + DML_LOG_VERBOSE("DML::%s: Tvm_oto = %f\n", __func__, s->Tvm_oto); + DML_LOG_VERBOSE("DML::%s: Tvm_oto_lines = %f\n", __func__, s->Tvm_oto_lines); + DML_LOG_VERBOSE("DML::%s: Tr0_oto_lines = %f\n", __func__, s->Tr0_oto_lines); + DML_LOG_VERBOSE("DML::%s: Lsw_oto = %f\n", __func__, s->Lsw_oto); + DML_LOG_VERBOSE("DML::%s: dst_y_prefetch_oto = %f\n", __func__, s->dst_y_prefetch_oto); + DML_LOG_VERBOSE("DML::%s: dst_y_prefetch_equ = %f\n", __func__, s->dst_y_prefetch_equ); + DML_LOG_VERBOSE("DML::%s: tdlut_row_bytes = %d\n", __func__, tdlut_row_bytes); + DML_LOG_VERBOSE("DML::%s: meta_row_bytes = %d\n", __func__, p->meta_row_bytes); +#endif s->dst_y_prefetch_equ = math_floor2(4.0 * (s->dst_y_prefetch_equ + 0.125), 1) / 4.0; *p->Tpre_rounded = s->dst_y_prefetch_equ * s->LineTime; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: dst_y_prefetch_equ: %f (after round)\n", __func__, s->dst_y_prefetch_equ); - dml2_printf("DML::%s: LineTime: %f\n", __func__, s->LineTime); - dml2_printf("DML::%s: VStartup: %u\n", __func__, p->VStartup); - dml2_printf("DML::%s: Tvstartup: %fus - time between vstartup and first pixel of active\n", __func__, p->VStartup * s->LineTime); - dml2_printf("DML::%s: TSetup: %fus - time from vstartup to vready\n", __func__, *p->TSetup); - dml2_printf("DML::%s: TCalc: %fus - time for calculations in dchub starting at vready\n", __func__, p->TCalc); - dml2_printf("DML::%s: TWait: %fus - time for fabric to become ready max(pstate exit,cstate enter/exit, urgent latency) after TCalc\n", __func__, p->TWait); - dml2_printf("DML::%s: Tdmbf: %fus - time for dmd transfer from dchub to dio output buffer\n", __func__, s->Tdmbf); - dml2_printf("DML::%s: Tdmec: %fus - time dio takes to transfer dmd\n", __func__, s->Tdmec); - dml2_printf("DML::%s: Tdmsks: %fus - time before active dmd must complete transmission at dio\n", __func__, s->Tdmsks); - dml2_printf("DML::%s: TWait = %f\n", __func__, p->TWait); - dml2_printf("DML::%s: TWait_p = %f\n", __func__, s->TWait_p); - dml2_printf("DML::%s: Ttrip = %f\n", __func__, p->Ttrip); - dml2_printf("DML::%s: Tex = %f\n", __func__, p->ExtraLatencyPrefetch); - dml2_printf("DML::%s: Tdmdl_vm: %fus - time for vm stages of dmd \n", __func__, *p->Tdmdl_vm); - dml2_printf("DML::%s: Tdmdl: %fus - time for fabric to become ready and fetch dmd \n", __func__, *p->Tdmdl); - dml2_printf("DML::%s: TWait_p: %fus\n", __func__, s->TWait_p); - dml2_printf("DML::%s: Ttrip: %fus\n", __func__, p->Ttrip); - dml2_printf("DML::%s: DSTXAfterScaler: %u pixels - number of pixel clocks pipeline and buffer delay after scaler \n", __func__, *p->DSTXAfterScaler); - dml2_printf("DML::%s: DSTYAfterScaler: %u lines - number of lines of pipeline and buffer delay after scaler \n", __func__, *p->DSTYAfterScaler); - dml2_printf("DML::%s: vm_bytes: %f (hvm inefficiency scaled)\n", __func__, vm_bytes*p->HostVMInefficiencyFactor); - dml2_printf("DML::%s: row_bytes: %f (hvm inefficiency scaled, 1 row)\n", __func__, p->PixelPTEBytesPerRow*p->HostVMInefficiencyFactor+p->meta_row_bytes+tdlut_row_bytes); - dml2_printf("DML::%s: Tno_bw: %f\n", __func__, *p->Tno_bw); - dml2_printf("DML::%s: Tpre=%f Tpre_rounded: %f, delta=%f\n", __func__, Tpre, *p->Tpre_rounded, (*p->Tpre_rounded - Tpre)); - dml2_printf("DML::%s: Tvm_trips=%f Tvm_trips_rounded: %f, delta=%f\n", __func__, *p->Tvm_trips, s->Tvm_trips_rounded, (s->Tvm_trips_rounded - *p->Tvm_trips)); + DML_LOG_VERBOSE("DML::%s: dst_y_prefetch_equ: %f (after round)\n", __func__, s->dst_y_prefetch_equ); + DML_LOG_VERBOSE("DML::%s: LineTime: %f\n", __func__, s->LineTime); + DML_LOG_VERBOSE("DML::%s: VStartup: %u\n", __func__, p->VStartup); + DML_LOG_VERBOSE("DML::%s: Tvstartup: %fus - time between vstartup and first pixel of active\n", __func__, p->VStartup * s->LineTime); + DML_LOG_VERBOSE("DML::%s: TSetup: %fus - time from vstartup to vready\n", __func__, *p->TSetup); + DML_LOG_VERBOSE("DML::%s: TCalc: %fus - time for calculations in dchub starting at vready\n", __func__, p->TCalc); + DML_LOG_VERBOSE("DML::%s: TWait: %fus - time for fabric to become ready max(pstate exit,cstate enter/exit, urgent latency) after TCalc\n", __func__, p->TWait); + DML_LOG_VERBOSE("DML::%s: Tdmbf: %fus - time for dmd transfer from dchub to dio output buffer\n", __func__, s->Tdmbf); + DML_LOG_VERBOSE("DML::%s: Tdmec: %fus - time dio takes to transfer dmd\n", __func__, s->Tdmec); + DML_LOG_VERBOSE("DML::%s: Tdmsks: %fus - time before active dmd must complete transmission at dio\n", __func__, s->Tdmsks); + DML_LOG_VERBOSE("DML::%s: TWait = %f\n", __func__, p->TWait); + DML_LOG_VERBOSE("DML::%s: TWait_p = %f\n", __func__, s->TWait_p); + DML_LOG_VERBOSE("DML::%s: Ttrip = %f\n", __func__, p->Ttrip); + DML_LOG_VERBOSE("DML::%s: Tex = %f\n", __func__, p->ExtraLatencyPrefetch); + DML_LOG_VERBOSE("DML::%s: Tdmdl_vm: %fus - time for vm stages of dmd \n", __func__, *p->Tdmdl_vm); + DML_LOG_VERBOSE("DML::%s: Tdmdl: %fus - time for fabric to become ready and fetch dmd \n", __func__, *p->Tdmdl); + DML_LOG_VERBOSE("DML::%s: TWait_p: %fus\n", __func__, s->TWait_p); + DML_LOG_VERBOSE("DML::%s: Ttrip: %fus\n", __func__, p->Ttrip); + DML_LOG_VERBOSE("DML::%s: DSTXAfterScaler: %u pixels - number of pixel clocks pipeline and buffer delay after scaler \n", __func__, *p->DSTXAfterScaler); + DML_LOG_VERBOSE("DML::%s: DSTYAfterScaler: %u lines - number of lines of pipeline and buffer delay after scaler \n", __func__, *p->DSTYAfterScaler); + DML_LOG_VERBOSE("DML::%s: vm_bytes: %f (hvm inefficiency scaled)\n", __func__, vm_bytes*p->HostVMInefficiencyFactor); + DML_LOG_VERBOSE("DML::%s: row_bytes: %f (hvm inefficiency scaled, 1 row)\n", __func__, p->PixelPTEBytesPerRow*p->HostVMInefficiencyFactor+p->meta_row_bytes+tdlut_row_bytes); + DML_LOG_VERBOSE("DML::%s: Tno_bw: %f\n", __func__, *p->Tno_bw); + DML_LOG_VERBOSE("DML::%s: Tpre=%f Tpre_rounded: %f, delta=%f\n", __func__, (s->dst_y_prefetch_equ * s->LineTime), *p->Tpre_rounded, (*p->Tpre_rounded - (s->dst_y_prefetch_equ * s->LineTime))); + DML_LOG_VERBOSE("DML::%s: Tvm_trips=%f Tvm_trips_rounded: %f, delta=%f\n", __func__, *p->Tvm_trips, s->Tvm_trips_rounded, (s->Tvm_trips_rounded - *p->Tvm_trips)); #endif *p->dst_y_per_vm_vblank = 0; @@ -5596,19 +5533,19 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch } else s->prefetch_bw1 = 0; - dml2_printf("DML::%s: prefetch_bw1: %f\n", __func__, s->prefetch_bw1); + DML_LOG_VERBOSE("DML::%s: prefetch_bw1: %f\n", __func__, s->prefetch_bw1); if ((s->Tsw_est1 < s->min_Lsw_equ * s->LineTime) && (*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.75 * s->LineTime - *p->Tno_bw > 0)) { s->prefetch_bw1 = (vm_bytes * p->HostVMInefficiencyFactor + 2 * (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes)) / (*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.75 * s->LineTime - *p->Tno_bw); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: vm and 2 rows bytes = %f\n", __func__, (vm_bytes * p->HostVMInefficiencyFactor + 2 * (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes))); - dml2_printf("DML::%s: Tpre_rounded = %f\n", __func__, *p->Tpre_rounded); - dml2_printf("DML::%s: minus term = %f\n", __func__, s->min_Lsw_equ * s->LineTime + 0.75 * s->LineTime + *p->Tno_bw); - dml2_printf("DML::%s: min_Lsw_equ = %f\n", __func__, s->min_Lsw_equ); - dml2_printf("DML::%s: LineTime = %f\n", __func__, s->LineTime); - dml2_printf("DML::%s: Tno_bw = %f\n", __func__, *p->Tno_bw); - dml2_printf("DML::%s: Time to fetch vm and 2 rows = %f\n", __func__, (*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.75 * s->LineTime - *p->Tno_bw)); - dml2_printf("DML::%s: prefetch_bw1: %f (updated)\n", __func__, s->prefetch_bw1); + DML_LOG_VERBOSE("DML::%s: vm and 2 rows bytes = %f\n", __func__, (vm_bytes * p->HostVMInefficiencyFactor + 2 * (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes))); + DML_LOG_VERBOSE("DML::%s: Tpre_rounded = %f\n", __func__, *p->Tpre_rounded); + DML_LOG_VERBOSE("DML::%s: minus term = %f\n", __func__, s->min_Lsw_equ * s->LineTime + 0.75 * s->LineTime + *p->Tno_bw); + DML_LOG_VERBOSE("DML::%s: min_Lsw_equ = %f\n", __func__, s->min_Lsw_equ); + DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, s->LineTime); + DML_LOG_VERBOSE("DML::%s: Tno_bw = %f\n", __func__, *p->Tno_bw); + DML_LOG_VERBOSE("DML::%s: Time to fetch vm and 2 rows = %f\n", __func__, (*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.75 * s->LineTime - *p->Tno_bw)); + DML_LOG_VERBOSE("DML::%s: prefetch_bw1: %f (updated)\n", __func__, s->prefetch_bw1); #endif } @@ -5620,10 +5557,10 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch } else s->prefetch_bw2 = 0; - dml2_printf("DML::%s: prefetch_bw2: %f\n", __func__, s->prefetch_bw2); + DML_LOG_VERBOSE("DML::%s: prefetch_bw2: %f\n", __func__, s->prefetch_bw2); if ((s->Tsw_est2 < s->min_Lsw_equ * s->LineTime) && ((*p->Tpre_rounded - *p->Tno_bw - 2.0 * s->Tr0_trips_rounded - s->min_Lsw_equ * s->LineTime - 0.25 * s->LineTime) > 0)) { s->prefetch_bw2 = vm_bytes * p->HostVMInefficiencyFactor / (*p->Tpre_rounded - *p->Tno_bw - 2.0 * s->Tr0_trips_rounded - s->min_Lsw_equ * s->LineTime - 0.25 * s->LineTime); - dml2_printf("DML::%s: prefetch_bw2: %f (updated)\n", __func__, s->prefetch_bw2); + DML_LOG_VERBOSE("DML::%s: prefetch_bw2: %f (updated)\n", __func__, s->prefetch_bw2); } // prefetch_bw3: 2*R0 + SW @@ -5634,10 +5571,10 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch } else s->prefetch_bw3 = 0; - dml2_printf("DML::%s: prefetch_bw3: %f\n", __func__, s->prefetch_bw3); + DML_LOG_VERBOSE("DML::%s: prefetch_bw3: %f\n", __func__, s->prefetch_bw3); if ((s->Tsw_est3 < s->min_Lsw_equ * s->LineTime) && ((*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.5 * s->LineTime - s->Tvm_trips_rounded) > 0)) { s->prefetch_bw3 = (2 * (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes)) / (*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.5 * s->LineTime - s->Tvm_trips_rounded); - dml2_printf("DML::%s: prefetch_bw3: %f (updated)\n", __func__, s->prefetch_bw3); + DML_LOG_VERBOSE("DML::%s: prefetch_bw3: %f (updated)\n", __func__, s->prefetch_bw3); } // prefetch_bw4: SW @@ -5647,17 +5584,17 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch s->prefetch_bw4 = 0; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: Tno_bw: %f\n", __func__, *p->Tno_bw); - dml2_printf("DML::%s: Tpre=%f Tpre_rounded: %f, delta=%f\n", __func__, Tpre, *p->Tpre_rounded, (*p->Tpre_rounded - Tpre)); - dml2_printf("DML::%s: Tvm_trips=%f Tvm_trips_rounded: %f, delta=%f\n", __func__, *p->Tvm_trips, s->Tvm_trips_rounded, (s->Tvm_trips_rounded - *p->Tvm_trips)); - dml2_printf("DML::%s: Tr0_trips=%f Tr0_trips_rounded: %f, delta=%f\n", __func__, *p->Tr0_trips, s->Tr0_trips_rounded, (s->Tr0_trips_rounded - *p->Tr0_trips)); - dml2_printf("DML::%s: Tsw_est1: %f\n", __func__, s->Tsw_est1); - dml2_printf("DML::%s: Tsw_est2: %f\n", __func__, s->Tsw_est2); - dml2_printf("DML::%s: Tsw_est3: %f\n", __func__, s->Tsw_est3); - dml2_printf("DML::%s: prefetch_bw1: %f (final)\n", __func__, s->prefetch_bw1); - dml2_printf("DML::%s: prefetch_bw2: %f (final)\n", __func__, s->prefetch_bw2); - dml2_printf("DML::%s: prefetch_bw3: %f (final)\n", __func__, s->prefetch_bw3); - dml2_printf("DML::%s: prefetch_bw4: %f (final)\n", __func__, s->prefetch_bw4); + DML_LOG_VERBOSE("DML::%s: Tno_bw: %f\n", __func__, *p->Tno_bw); + DML_LOG_VERBOSE("DML::%s: Tpre=%f Tpre_rounded: %f, delta=%f\n", __func__, s->dst_y_prefetch_equ * s->LineTime, *p->Tpre_rounded, (*p->Tpre_rounded - (s->dst_y_prefetch_equ * s->LineTime))); + DML_LOG_VERBOSE("DML::%s: Tvm_trips=%f Tvm_trips_rounded: %f, delta=%f\n", __func__, *p->Tvm_trips, s->Tvm_trips_rounded, (s->Tvm_trips_rounded - *p->Tvm_trips)); + DML_LOG_VERBOSE("DML::%s: Tr0_trips=%f Tr0_trips_rounded: %f, delta=%f\n", __func__, *p->Tr0_trips, s->Tr0_trips_rounded, (s->Tr0_trips_rounded - *p->Tr0_trips)); + DML_LOG_VERBOSE("DML::%s: Tsw_est1: %f\n", __func__, s->Tsw_est1); + DML_LOG_VERBOSE("DML::%s: Tsw_est2: %f\n", __func__, s->Tsw_est2); + DML_LOG_VERBOSE("DML::%s: Tsw_est3: %f\n", __func__, s->Tsw_est3); + DML_LOG_VERBOSE("DML::%s: prefetch_bw1: %f (final)\n", __func__, s->prefetch_bw1); + DML_LOG_VERBOSE("DML::%s: prefetch_bw2: %f (final)\n", __func__, s->prefetch_bw2); + DML_LOG_VERBOSE("DML::%s: prefetch_bw3: %f (final)\n", __func__, s->prefetch_bw3); + DML_LOG_VERBOSE("DML::%s: prefetch_bw4: %f (final)\n", __func__, s->prefetch_bw4); #endif { bool Case1OK = false; @@ -5676,14 +5613,14 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch double total_row_bytes = (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes); - dml2_printf("DML::%s: Tvm_trips_rounded = %f\n", __func__, s->Tvm_trips_rounded); - dml2_printf("DML::%s: Tr0_trips_rounded = %f\n", __func__, s->Tr0_trips_rounded); + DML_LOG_VERBOSE("DML::%s: Tvm_trips_rounded = %f\n", __func__, s->Tvm_trips_rounded); + DML_LOG_VERBOSE("DML::%s: Tr0_trips_rounded = %f\n", __func__, s->Tr0_trips_rounded); if (s->prefetch_bw1 > 0) { double vm_transfer_time = *p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw1; double row_transfer_time = total_row_bytes / s->prefetch_bw1; - dml2_printf("DML::%s: Case1: vm_transfer_time = %f\n", __func__, vm_transfer_time); - dml2_printf("DML::%s: Case1: row_transfer_time = %f\n", __func__, row_transfer_time); + DML_LOG_VERBOSE("DML::%s: Case1: vm_transfer_time = %f\n", __func__, vm_transfer_time); + DML_LOG_VERBOSE("DML::%s: Case1: row_transfer_time = %f\n", __func__, row_transfer_time); if (vm_transfer_time >= s->Tvm_trips_rounded && row_transfer_time >= s->Tr0_trips_rounded) { Case1OK = true; } @@ -5696,8 +5633,8 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch if (s->prefetch_bw2 > 0) { double vm_transfer_time = *p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw2; double row_transfer_time = total_row_bytes / s->prefetch_bw2; - dml2_printf("DML::%s: Case2: vm_transfer_time = %f\n", __func__, vm_transfer_time); - dml2_printf("DML::%s: Case2: row_transfer_time = %f\n", __func__, row_transfer_time); + DML_LOG_VERBOSE("DML::%s: Case2: vm_transfer_time = %f\n", __func__, vm_transfer_time); + DML_LOG_VERBOSE("DML::%s: Case2: row_transfer_time = %f\n", __func__, row_transfer_time); if (vm_transfer_time >= s->Tvm_trips_rounded && row_transfer_time < s->Tr0_trips_rounded) { Case2OK = true; } @@ -5709,8 +5646,8 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch if (s->prefetch_bw3 > 0) { double vm_transfer_time = *p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw3; double row_transfer_time = total_row_bytes / s->prefetch_bw3; - dml2_printf("DML::%s: Case3: vm_transfer_time = %f\n", __func__, vm_transfer_time); - dml2_printf("DML::%s: Case3: row_transfer_time = %f\n", __func__, row_transfer_time); + DML_LOG_VERBOSE("DML::%s: Case3: vm_transfer_time = %f\n", __func__, vm_transfer_time); + DML_LOG_VERBOSE("DML::%s: Case3: row_transfer_time = %f\n", __func__, row_transfer_time); if (vm_transfer_time < s->Tvm_trips_rounded && row_transfer_time >= s->Tr0_trips_rounded) { Case3OK = true; } @@ -5730,10 +5667,10 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch p->vm_bytes * p->HostVMInefficiencyFactor / (31 * s->LineTime) - *p->Tno_bw, (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) / (15 * s->LineTime)); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: Case1OK: %u\n", __func__, Case1OK); - dml2_printf("DML::%s: Case2OK: %u\n", __func__, Case2OK); - dml2_printf("DML::%s: Case3OK: %u\n", __func__, Case3OK); - dml2_printf("DML::%s: prefetch_bw_equ: %f\n", __func__, s->prefetch_bw_equ); + DML_LOG_VERBOSE("DML::%s: Case1OK: %u\n", __func__, Case1OK); + DML_LOG_VERBOSE("DML::%s: Case2OK: %u\n", __func__, Case2OK); + DML_LOG_VERBOSE("DML::%s: Case3OK: %u\n", __func__, Case3OK); + DML_LOG_VERBOSE("DML::%s: prefetch_bw_equ: %f\n", __func__, s->prefetch_bw_equ); #endif if (s->prefetch_bw_equ > 0) { @@ -5753,12 +5690,12 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch } else { s->Tvm_equ = 0; s->Tr0_equ = 0; - dml2_printf("DML::%s: prefetch_bw_equ equals 0!\n", __func__); + DML_LOG_VERBOSE("DML::%s: prefetch_bw_equ equals 0!\n", __func__); } } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: Tvm_equ = %f\n", __func__, s->Tvm_equ); - dml2_printf("DML::%s: Tr0_equ = %f\n", __func__, s->Tr0_equ); + DML_LOG_VERBOSE("DML::%s: Tvm_equ = %f\n", __func__, s->Tvm_equ); + DML_LOG_VERBOSE("DML::%s: Tr0_equ = %f\n", __func__, s->Tr0_equ); #endif // Use the more stressful prefetch schedule if (s->dst_y_prefetch_oto < s->dst_y_prefetch_equ) { @@ -5769,7 +5706,7 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch *p->dst_y_per_vm_vblank = math_ceil2(4.0 * s->TimeForFetchingVM / s->LineTime, 1.0) / 4.0; *p->dst_y_per_row_vblank = math_ceil2(4.0 * s->TimeForFetchingRowInVBlank / s->LineTime, 1.0) / 4.0; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: Using oto scheduling for prefetch\n", __func__); + DML_LOG_VERBOSE("DML::%s: Using oto scheduling for prefetch\n", __func__); #endif } else { @@ -5781,11 +5718,17 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch s->TimeForFetchingVM = s->Tvm_equ; s->TimeForFetchingRowInVBlank = s->Tr0_equ; - *p->dst_y_per_vm_vblank = math_ceil2(4.0 * s->TimeForFetchingVM / s->LineTime, 1.0) / 4.0; - *p->dst_y_per_row_vblank = math_ceil2(4.0 * s->TimeForFetchingRowInVBlank / s->LineTime, 1.0) / 4.0; + *p->dst_y_per_vm_vblank = math_ceil2(4.0 * s->TimeForFetchingVM / s->LineTime, 1.0) / 4.0; + *p->dst_y_per_row_vblank = math_ceil2(4.0 * s->TimeForFetchingRowInVBlank / s->LineTime, 1.0) / 4.0; + + /* equ bw should be propagated so a ceiling of the equ bw is accounted for prior to mode programming. + * Overall bandwidth may be lower when going from mode support to mode programming but final pixel data + * bandwidth may end up higher than what was calculated in mode support. + */ + *p->RequiredPrefetchBWMax = math_max2(s->prefetch_bw_equ, *p->RequiredPrefetchBWMax); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: Using equ bw scheduling for prefetch\n", __func__); + DML_LOG_VERBOSE("DML::%s: Using equ bw scheduling for prefetch\n", __func__); #endif } @@ -5797,31 +5740,31 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch *p->prefetch_swath_time_us = (s->LinesToRequestPrefetchPixelData * s->LineTime); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: TimeForFetchingVM = %f\n", __func__, s->TimeForFetchingVM); - dml2_printf("DML::%s: TimeForFetchingRowInVBlank = %f\n", __func__, s->TimeForFetchingRowInVBlank); - dml2_printf("DML::%s: LineTime = %f\n", __func__, s->LineTime); - dml2_printf("DML::%s: dst_y_prefetch = %f\n", __func__, *p->dst_y_prefetch); - dml2_printf("DML::%s: dst_y_per_vm_vblank = %f\n", __func__, *p->dst_y_per_vm_vblank); - dml2_printf("DML::%s: dst_y_per_row_vblank = %f\n", __func__, *p->dst_y_per_row_vblank); - dml2_printf("DML::%s: LinesToRequestPrefetchPixelData = %f\n", __func__, s->LinesToRequestPrefetchPixelData); - dml2_printf("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); - dml2_printf("DML::%s: prefetch_swath_time_us = %f\n", __func__, *p->prefetch_swath_time_us); + DML_LOG_VERBOSE("DML::%s: TimeForFetchingVM = %f\n", __func__, s->TimeForFetchingVM); + DML_LOG_VERBOSE("DML::%s: TimeForFetchingRowInVBlank = %f\n", __func__, s->TimeForFetchingRowInVBlank); + DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, s->LineTime); + DML_LOG_VERBOSE("DML::%s: dst_y_prefetch = %f\n", __func__, *p->dst_y_prefetch); + DML_LOG_VERBOSE("DML::%s: dst_y_per_vm_vblank = %f\n", __func__, *p->dst_y_per_vm_vblank); + DML_LOG_VERBOSE("DML::%s: dst_y_per_row_vblank = %f\n", __func__, *p->dst_y_per_row_vblank); + DML_LOG_VERBOSE("DML::%s: LinesToRequestPrefetchPixelData = %f\n", __func__, s->LinesToRequestPrefetchPixelData); + DML_LOG_VERBOSE("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); + DML_LOG_VERBOSE("DML::%s: prefetch_swath_time_us = %f\n", __func__, *p->prefetch_swath_time_us); - dml2_printf("DML::%s: cursor_bytes_per_chunk = %d\n", __func__, p->cursor_bytes_per_chunk); - dml2_printf("DML::%s: cursor_bytes_per_line = %d\n", __func__, p->cursor_bytes_per_line); - dml2_printf("DML::%s: cursor_prefetch_bytes = %d\n", __func__, s->cursor_prefetch_bytes); - dml2_printf("DML::%s: prefetch_cursor_bw = %f\n", __func__, *p->prefetch_cursor_bw); + DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_chunk = %d\n", __func__, p->cursor_bytes_per_chunk); + DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_line = %d\n", __func__, p->cursor_bytes_per_line); + DML_LOG_VERBOSE("DML::%s: cursor_prefetch_bytes = %d\n", __func__, s->cursor_prefetch_bytes); + DML_LOG_VERBOSE("DML::%s: prefetch_cursor_bw = %f\n", __func__, *p->prefetch_cursor_bw); #endif - DML2_ASSERT(*p->dst_y_prefetch < 64); + DML_ASSERT(*p->dst_y_prefetch < 64); unsigned int min_lsw_required = (unsigned int)math_max2(2, p->tdlut_drain_time / s->LineTime); if (s->LinesToRequestPrefetchPixelData >= min_lsw_required && s->prefetch_bw_equ > 0) { *p->VRatioPrefetchY = (double)p->PrefetchSourceLinesY / s->LinesToRequestPrefetchPixelData; *p->VRatioPrefetchY = math_max2(*p->VRatioPrefetchY, 1.0); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: VRatioPrefetchY = %f\n", __func__, *p->VRatioPrefetchY); - dml2_printf("DML::%s: SwathHeightY = %u\n", __func__, p->SwathHeightY); - dml2_printf("DML::%s: VInitPreFillY = %u\n", __func__, p->VInitPreFillY); + DML_LOG_VERBOSE("DML::%s: VRatioPrefetchY = %f\n", __func__, *p->VRatioPrefetchY); + DML_LOG_VERBOSE("DML::%s: SwathHeightY = %u\n", __func__, p->SwathHeightY); + DML_LOG_VERBOSE("DML::%s: VInitPreFillY = %u\n", __func__, p->VInitPreFillY); #endif if ((p->SwathHeightY > 4) && (p->VInitPreFillY > 3)) { if (s->LinesToRequestPrefetchPixelData > (p->VInitPreFillY - 3.0) / 2.0) { @@ -5829,13 +5772,13 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch (double)p->MaxNumSwathY * p->SwathHeightY / (s->LinesToRequestPrefetchPixelData - (p->VInitPreFillY - 3.0) / 2.0)); } else { s->NoTimeToPrefetch = true; - dml2_printf("DML::%s: No time to prefetch!. LinesToRequestPrefetchPixelData=%f VinitPreFillY=%u\n", __func__, s->LinesToRequestPrefetchPixelData, p->VInitPreFillY); + DML_LOG_VERBOSE("DML::%s: No time to prefetch!. LinesToRequestPrefetchPixelData=%f VinitPreFillY=%u\n", __func__, s->LinesToRequestPrefetchPixelData, p->VInitPreFillY); *p->VRatioPrefetchY = 0; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: VRatioPrefetchY = %f\n", __func__, *p->VRatioPrefetchY); - dml2_printf("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); - dml2_printf("DML::%s: MaxNumSwathY = %u\n", __func__, p->MaxNumSwathY); + DML_LOG_VERBOSE("DML::%s: VRatioPrefetchY = %f\n", __func__, *p->VRatioPrefetchY); + DML_LOG_VERBOSE("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); + DML_LOG_VERBOSE("DML::%s: MaxNumSwathY = %u\n", __func__, p->MaxNumSwathY); #endif } @@ -5843,22 +5786,22 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch *p->VRatioPrefetchC = math_max2(*p->VRatioPrefetchC, 1.0); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: VRatioPrefetchC = %f\n", __func__, *p->VRatioPrefetchC); - dml2_printf("DML::%s: SwathHeightC = %u\n", __func__, p->SwathHeightC); - dml2_printf("DML::%s: VInitPreFillC = %u\n", __func__, p->VInitPreFillC); + DML_LOG_VERBOSE("DML::%s: VRatioPrefetchC = %f\n", __func__, *p->VRatioPrefetchC); + DML_LOG_VERBOSE("DML::%s: SwathHeightC = %u\n", __func__, p->SwathHeightC); + DML_LOG_VERBOSE("DML::%s: VInitPreFillC = %u\n", __func__, p->VInitPreFillC); #endif if ((p->SwathHeightC > 4) && (p->VInitPreFillC > 3)) { if (s->LinesToRequestPrefetchPixelData > (p->VInitPreFillC - 3.0) / 2.0) { *p->VRatioPrefetchC = math_max2(*p->VRatioPrefetchC, (double)p->MaxNumSwathC * p->SwathHeightC / (s->LinesToRequestPrefetchPixelData - (p->VInitPreFillC - 3.0) / 2.0)); } else { s->NoTimeToPrefetch = true; - dml2_printf("DML::%s: No time to prefetch!. LinesToRequestPrefetchPixelData=%f VInitPreFillC=%u\n", __func__, s->LinesToRequestPrefetchPixelData, p->VInitPreFillC); + DML_LOG_VERBOSE("DML::%s: No time to prefetch!. LinesToRequestPrefetchPixelData=%f VInitPreFillC=%u\n", __func__, s->LinesToRequestPrefetchPixelData, p->VInitPreFillC); *p->VRatioPrefetchC = 0; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: VRatioPrefetchC = %f\n", __func__, *p->VRatioPrefetchC); - dml2_printf("DML::%s: PrefetchSourceLinesC = %f\n", __func__, p->PrefetchSourceLinesC); - dml2_printf("DML::%s: MaxNumSwathC = %u\n", __func__, p->MaxNumSwathC); + DML_LOG_VERBOSE("DML::%s: VRatioPrefetchC = %f\n", __func__, *p->VRatioPrefetchC); + DML_LOG_VERBOSE("DML::%s: PrefetchSourceLinesC = %f\n", __func__, p->PrefetchSourceLinesC); + DML_LOG_VERBOSE("DML::%s: MaxNumSwathC = %u\n", __func__, p->MaxNumSwathC); #endif } @@ -5866,36 +5809,34 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch *p->RequiredPrefetchPixelDataBWChroma = (double)p->PrefetchSourceLinesC / s->LinesToRequestPrefetchPixelData * p->myPipe->BytePerPixelC * p->swath_width_chroma_ub / s->LineTime; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: BytePerPixelY = %u\n", __func__, p->myPipe->BytePerPixelY); - dml2_printf("DML::%s: swath_width_luma_ub = %u\n", __func__, p->swath_width_luma_ub); - dml2_printf("DML::%s: LineTime = %f\n", __func__, s->LineTime); - dml2_printf("DML::%s: RequiredPrefetchPixelDataBWLuma = %f\n", __func__, *p->RequiredPrefetchPixelDataBWLuma); - dml2_printf("DML::%s: RequiredPrefetchPixelDataBWChroma = %f\n", __func__, *p->RequiredPrefetchPixelDataBWChroma); + DML_LOG_VERBOSE("DML::%s: BytePerPixelY = %u\n", __func__, p->myPipe->BytePerPixelY); + DML_LOG_VERBOSE("DML::%s: swath_width_luma_ub = %u\n", __func__, p->swath_width_luma_ub); + DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, s->LineTime); + DML_LOG_VERBOSE("DML::%s: RequiredPrefetchPixelDataBWLuma = %f\n", __func__, *p->RequiredPrefetchPixelDataBWLuma); + DML_LOG_VERBOSE("DML::%s: RequiredPrefetchPixelDataBWChroma = %f\n", __func__, *p->RequiredPrefetchPixelDataBWChroma); #endif } else { s->NoTimeToPrefetch = true; - dml2_printf("DML::%s: No time to prefetch!, LinesToRequestPrefetchPixelData: %f, should be >= %d\n", __func__, s->LinesToRequestPrefetchPixelData, min_lsw_required); - dml2_printf("DML::%s: No time to prefetch!, prefetch_bw_equ: %f, should be > 0\n", __func__, s->prefetch_bw_equ); + DML_LOG_VERBOSE("DML::%s: No time to prefetch!, LinesToRequestPrefetchPixelData: %f, should be >= %d\n", __func__, s->LinesToRequestPrefetchPixelData, min_lsw_required); + DML_LOG_VERBOSE("DML::%s: No time to prefetch!, prefetch_bw_equ: %f, should be > 0\n", __func__, s->prefetch_bw_equ); *p->VRatioPrefetchY = 0; *p->VRatioPrefetchC = 0; *p->RequiredPrefetchPixelDataBWLuma = 0; *p->RequiredPrefetchPixelDataBWChroma = 0; } - dml2_printf("DML: Tpre: %fus - sum of time to request 2 x data pte, swaths\n", (double)s->LinesToRequestPrefetchPixelData * s->LineTime + 2.0 * s->TimeForFetchingRowInVBlank + s->TimeForFetchingVM); - dml2_printf("DML: Tvm: %fus - time to fetch vm\n", s->TimeForFetchingVM); - dml2_printf("DML: Tr0: %fus - time to fetch first row of data pagetables\n", s->TimeForFetchingRowInVBlank); - dml2_printf("DML: Tsw: %fus = time to fetch enough pixel data and cursor data to feed the scalers init position and detile\n", (double)s->LinesToRequestPrefetchPixelData * s->LineTime); - dml2_printf("DML: To: %fus - time for propagation from scaler to optc\n", (*p->DSTYAfterScaler + ((double)(*p->DSTXAfterScaler) / (double)p->myPipe->HTotal)) * s->LineTime); - dml2_printf("DML: Tvstartup - TSetup - Tcalc - TWait - Tpre - To > 0\n"); - dml2_printf("DML: Tslack(pre): %fus - time left over in schedule\n", p->VStartup * s->LineTime - s->TimeForFetchingVM - 2 * s->TimeForFetchingRowInVBlank - (*p->DSTYAfterScaler + ((double)(*p->DSTXAfterScaler) / (double)p->myPipe->HTotal)) * s->LineTime - p->TWait - p->TCalc - *p->TSetup); - dml2_printf("DML: row_bytes = dpte_row_bytes (per_pipe) = PixelPTEBytesPerRow = : %u\n", p->PixelPTEBytesPerRow); + DML_LOG_VERBOSE("DML: Tpre: %fus - sum of time to request 2 x data pte, swaths\n", (double)s->LinesToRequestPrefetchPixelData * s->LineTime + 2.0 * s->TimeForFetchingRowInVBlank + s->TimeForFetchingVM); + DML_LOG_VERBOSE("DML: Tvm: %fus - time to fetch vm\n", s->TimeForFetchingVM); + DML_LOG_VERBOSE("DML: Tr0: %fus - time to fetch first row of data pagetables\n", s->TimeForFetchingRowInVBlank); + DML_LOG_VERBOSE("DML: Tsw: %fus = time to fetch enough pixel data and cursor data to feed the scalers init position and detile\n", (double)s->LinesToRequestPrefetchPixelData * s->LineTime); + DML_LOG_VERBOSE("DML: To: %fus - time for propagation from scaler to optc\n", (*p->DSTYAfterScaler + ((double)(*p->DSTXAfterScaler) / (double)p->myPipe->HTotal)) * s->LineTime); + DML_LOG_VERBOSE("DML: Tvstartup - TSetup - Tcalc - TWait - Tpre - To > 0\n"); + DML_LOG_VERBOSE("DML: Tslack(pre): %fus - time left over in schedule\n", p->VStartup * s->LineTime - s->TimeForFetchingVM - 2 * s->TimeForFetchingRowInVBlank - (*p->DSTYAfterScaler + ((double)(*p->DSTXAfterScaler) / (double)p->myPipe->HTotal)) * s->LineTime - p->TWait - p->TCalc - *p->TSetup); + DML_LOG_VERBOSE("DML: row_bytes = dpte_row_bytes (per_pipe) = PixelPTEBytesPerRow = : %u\n", p->PixelPTEBytesPerRow); } else { - dml2_printf("DML::%s: No time to prefetch! dst_y_prefetch_equ = %f (should be > 1)\n", __func__, s->dst_y_prefetch_equ); - dml2_printf("DML::%s: No time to prefetch! min_Lsw_equ_ok = %d, Tpre_rounded (%f) should be >= Tvm_trips_rounded (%f) + 2.0*Tr0_trips_rounded (%f) + min_Tsw_equ (%f)\n", + DML_LOG_VERBOSE("DML::%s: No time to prefetch! dst_y_prefetch_equ = %f (should be > 1)\n", __func__, s->dst_y_prefetch_equ); + DML_LOG_VERBOSE("DML::%s: No time to prefetch! min_Lsw_equ_ok = %d, Tpre_rounded (%f) should be >= Tvm_trips_rounded (%f) + 2.0*Tr0_trips_rounded (%f) + min_Tsw_equ (%f)\n", __func__, min_Lsw_equ_ok, *p->Tpre_rounded, s->Tvm_trips_rounded, 2.0*s->Tr0_trips_rounded, s->min_Lsw_equ*s->LineTime); - dml2_printf("DML::%s: No time to prefetch! min_Lsw_equ_ok = %d, Tpre_rounded+Tvm_trips_rounded+2.0*Tr0_trips_rounded+min_Tsw_equ (%f) should be > \n", - __func__, tpre_gt_req_latency, (s->min_Lsw_equ*s->LineTime + s->Tvm_trips_rounded + 2.0*s->Tr0_trips_rounded), p->Turg, s->trip_to_mem, p->ExtraLatencyPrefetch); s->NoTimeToPrefetch = true; s->TimeForFetchingVM = 0; s->TimeForFetchingRowInVBlank = 0; @@ -5916,18 +5857,18 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch prefetch_vm_bw = 0; } else if (*p->dst_y_per_vm_vblank > 0) { #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); - dml2_printf("DML::%s: dst_y_per_vm_vblank = %f\n", __func__, *p->dst_y_per_vm_vblank); - dml2_printf("DML::%s: LineTime = %f\n", __func__, s->LineTime); + DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); + DML_LOG_VERBOSE("DML::%s: dst_y_per_vm_vblank = %f\n", __func__, *p->dst_y_per_vm_vblank); + DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, s->LineTime); #endif prefetch_vm_bw = vm_bytes * p->HostVMInefficiencyFactor / (*p->dst_y_per_vm_vblank * s->LineTime); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: prefetch_vm_bw = %f\n", __func__, prefetch_vm_bw); + DML_LOG_VERBOSE("DML::%s: prefetch_vm_bw = %f\n", __func__, prefetch_vm_bw); #endif } else { prefetch_vm_bw = 0; s->NoTimeToPrefetch = true; - dml2_printf("DML::%s: No time to prefetch!. dst_y_per_vm_vblank=%f (should be > 0)\n", __func__, *p->dst_y_per_vm_vblank); + DML_LOG_VERBOSE("DML::%s: No time to prefetch!. dst_y_per_vm_vblank=%f (should be > 0)\n", __func__, *p->dst_y_per_vm_vblank); } if (p->PixelPTEBytesPerRow == 0 && tdlut_row_bytes == 0) { @@ -5936,14 +5877,14 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch prefetch_row_bw = (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + tdlut_row_bytes) / (*p->dst_y_per_row_vblank * s->LineTime); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, p->PixelPTEBytesPerRow); - dml2_printf("DML::%s: dst_y_per_row_vblank = %f\n", __func__, *p->dst_y_per_row_vblank); - dml2_printf("DML::%s: prefetch_row_bw = %f\n", __func__, prefetch_row_bw); + DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, p->PixelPTEBytesPerRow); + DML_LOG_VERBOSE("DML::%s: dst_y_per_row_vblank = %f\n", __func__, *p->dst_y_per_row_vblank); + DML_LOG_VERBOSE("DML::%s: prefetch_row_bw = %f\n", __func__, prefetch_row_bw); #endif } else { prefetch_row_bw = 0; s->NoTimeToPrefetch = true; - dml2_printf("DML::%s: No time to prefetch!. dst_y_per_row_vblank=%f (should be > 0)\n", __func__, *p->dst_y_per_row_vblank); + DML_LOG_VERBOSE("DML::%s: No time to prefetch!. dst_y_per_row_vblank=%f (should be > 0)\n", __func__, *p->dst_y_per_row_vblank); } *p->prefetch_vmrow_bw = math_max2(prefetch_vm_bw, prefetch_row_bw); @@ -5963,12 +5904,12 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch *p->prefetch_vmrow_bw = 0; } - dml2_printf("DML::%s: dst_y_per_vm_vblank = %f (final)\n", __func__, *p->dst_y_per_vm_vblank); - dml2_printf("DML::%s: dst_y_per_row_vblank = %f (final)\n", __func__, *p->dst_y_per_row_vblank); - dml2_printf("DML::%s: prefetch_vmrow_bw = %f (final)\n", __func__, *p->prefetch_vmrow_bw); - dml2_printf("DML::%s: RequiredPrefetchPixelDataBWLuma = %f (final)\n", __func__, *p->RequiredPrefetchPixelDataBWLuma); - dml2_printf("DML::%s: RequiredPrefetchPixelDataBWChroma = %f (final)\n", __func__, *p->RequiredPrefetchPixelDataBWChroma); - dml2_printf("DML::%s: NoTimeToPrefetch=%d\n", __func__, s->NoTimeToPrefetch); + DML_LOG_VERBOSE("DML::%s: dst_y_per_vm_vblank = %f (final)\n", __func__, *p->dst_y_per_vm_vblank); + DML_LOG_VERBOSE("DML::%s: dst_y_per_row_vblank = %f (final)\n", __func__, *p->dst_y_per_row_vblank); + DML_LOG_VERBOSE("DML::%s: prefetch_vmrow_bw = %f (final)\n", __func__, *p->prefetch_vmrow_bw); + DML_LOG_VERBOSE("DML::%s: RequiredPrefetchPixelDataBWLuma = %f (final)\n", __func__, *p->RequiredPrefetchPixelDataBWLuma); + DML_LOG_VERBOSE("DML::%s: RequiredPrefetchPixelDataBWChroma = %f (final)\n", __func__, *p->RequiredPrefetchPixelDataBWChroma); + DML_LOG_VERBOSE("DML::%s: NoTimeToPrefetch=%d\n", __func__, s->NoTimeToPrefetch); return s->NoTimeToPrefetch; } @@ -6005,7 +5946,7 @@ static unsigned int find_max_impact_plane(unsigned int this_plane_idx, unsigned } } if (max_idx <= 0) { - DML2_ASSERT(max_idx >= 0); + DML_ASSERT(max_idx >= 0); max_idx = this_plane_idx; } @@ -6037,12 +5978,12 @@ static noinline_for_stack bool CheckGlobalPrefetchAdmissibility(struct dml2_core // worst case if the rob and cdb is fully hogged s->max_Trpd_dcfclk_cycles = (unsigned int) math_ceil2((p->rob_buffer_size_kbytes*1024 + p->compressed_buffer_size_kbytes*DML_MAX_COMPRESSION_RATIO*1024)/64.0, 1.0); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: num_active_planes = %d\n", __func__, p->num_active_planes); - dml2_printf("DML::%s: rob_buffer_size_kbytes = %d\n", __func__, p->rob_buffer_size_kbytes); - dml2_printf("DML::%s: compressed_buffer_size_kbytes = %d\n", __func__, p->compressed_buffer_size_kbytes); - dml2_printf("DML::%s: estimated_urg_bandwidth_required_mbps = %f\n", __func__, p->estimated_urg_bandwidth_required_mbps); - dml2_printf("DML::%s: estimated_dcfclk_mhz = %f\n", __func__, p->estimated_dcfclk_mhz); - dml2_printf("DML::%s: max_Trpd_dcfclk_cycles = %u\n", __func__, s->max_Trpd_dcfclk_cycles); + DML_LOG_VERBOSE("DML::%s: num_active_planes = %d\n", __func__, p->num_active_planes); + DML_LOG_VERBOSE("DML::%s: rob_buffer_size_kbytes = %d\n", __func__, p->rob_buffer_size_kbytes); + DML_LOG_VERBOSE("DML::%s: compressed_buffer_size_kbytes = %d\n", __func__, p->compressed_buffer_size_kbytes); + DML_LOG_VERBOSE("DML::%s: estimated_urg_bandwidth_required_mbps = %f\n", __func__, p->estimated_urg_bandwidth_required_mbps); + DML_LOG_VERBOSE("DML::%s: estimated_dcfclk_mhz = %f\n", __func__, p->estimated_dcfclk_mhz); + DML_LOG_VERBOSE("DML::%s: max_Trpd_dcfclk_cycles = %u\n", __func__, s->max_Trpd_dcfclk_cycles); #endif // calculate the return impact from each plane, request is 256B per dcfclk @@ -6063,12 +6004,12 @@ static noinline_for_stack bool CheckGlobalPrefetchAdmissibility(struct dml2_core s->burst_bytes_to_fill_det += (unsigned int) (math_floor2(p->lb_source_lines_l[i] / p->swath_height_l[i], 1) * s->src_swath_bytes_l[i]); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: i=%u pixel_format = %d\n", __func__, i, p->pixel_format[i]); - dml2_printf("DML::%s: i=%u chunk_bytes_l = %d\n", __func__, i, p->chunk_bytes_l); - dml2_printf("DML::%s: i=%u lb_source_lines_l = %d\n", __func__, i, p->lb_source_lines_l[i]); - dml2_printf("DML::%s: i=%u src_detile_buf_size_bytes_l=%d\n", __func__, i, s->src_detile_buf_size_bytes_l[i]); - dml2_printf("DML::%s: i=%u src_swath_bytes_l=%d\n", __func__, i, s->src_swath_bytes_l[i]); - dml2_printf("DML::%s: i=%u burst_bytes_to_fill_det=%d (luma)\n", __func__, i, s->burst_bytes_to_fill_det); + DML_LOG_VERBOSE("DML::%s: i=%u pixel_format = %d\n", __func__, i, p->pixel_format[i]); + DML_LOG_VERBOSE("DML::%s: i=%u chunk_bytes_l = %d\n", __func__, i, p->chunk_bytes_l); + DML_LOG_VERBOSE("DML::%s: i=%u lb_source_lines_l = %d\n", __func__, i, p->lb_source_lines_l[i]); + DML_LOG_VERBOSE("DML::%s: i=%u src_detile_buf_size_bytes_l=%d\n", __func__, i, s->src_detile_buf_size_bytes_l[i]); + DML_LOG_VERBOSE("DML::%s: i=%u src_swath_bytes_l=%d\n", __func__, i, s->src_swath_bytes_l[i]); + DML_LOG_VERBOSE("DML::%s: i=%u burst_bytes_to_fill_det=%d (luma)\n", __func__, i, s->burst_bytes_to_fill_det); #endif if (s->src_swath_bytes_c[i] > 0) { // dual_plane @@ -6079,10 +6020,10 @@ static noinline_for_stack bool CheckGlobalPrefetchAdmissibility(struct dml2_core } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: i=%u chunk_bytes_c = %d\n", __func__, i, p->chunk_bytes_c); - dml2_printf("DML::%s: i=%u lb_source_lines_c = %d\n", __func__, i, p->lb_source_lines_c[i]); - dml2_printf("DML::%s: i=%u src_detile_buf_size_bytes_c=%d\n", __func__, i, s->src_detile_buf_size_bytes_c[i]); - dml2_printf("DML::%s: i=%u src_swath_bytes_c=%d\n", __func__, i, s->src_swath_bytes_c[i]); + DML_LOG_VERBOSE("DML::%s: i=%u chunk_bytes_c = %d\n", __func__, i, p->chunk_bytes_c); + DML_LOG_VERBOSE("DML::%s: i=%u lb_source_lines_c = %d\n", __func__, i, p->lb_source_lines_c[i]); + DML_LOG_VERBOSE("DML::%s: i=%u src_detile_buf_size_bytes_c=%d\n", __func__, i, s->src_detile_buf_size_bytes_c[i]); + DML_LOG_VERBOSE("DML::%s: i=%u src_swath_bytes_c=%d\n", __func__, i, s->src_swath_bytes_c[i]); #endif } @@ -6090,9 +6031,9 @@ static noinline_for_stack bool CheckGlobalPrefetchAdmissibility(struct dml2_core s->accumulated_return_path_dcfclk_cycles[i] = (unsigned int) math_ceil2(((DML_MAX_COMPRESSION_RATIO-1) * 64 * p->estimated_dcfclk_mhz) * s->time_to_fill_det_us / 64.0, 1.0); //for 64B per DCFClk #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: i=%u burst_bytes_to_fill_det=%d\n", __func__, i, s->burst_bytes_to_fill_det); - dml2_printf("DML::%s: i=%u time_to_fill_det_us=%f\n", __func__, i, s->time_to_fill_det_us); - dml2_printf("DML::%s: i=%u accumulated_return_path_dcfclk_cycles=%u\n", __func__, i, s->accumulated_return_path_dcfclk_cycles[i]); + DML_LOG_VERBOSE("DML::%s: i=%u burst_bytes_to_fill_det=%d\n", __func__, i, s->burst_bytes_to_fill_det); + DML_LOG_VERBOSE("DML::%s: i=%u time_to_fill_det_us=%f\n", __func__, i, s->time_to_fill_det_us); + DML_LOG_VERBOSE("DML::%s: i=%u accumulated_return_path_dcfclk_cycles=%u\n", __func__, i, s->accumulated_return_path_dcfclk_cycles[i]); #endif // clamping to worst case delay which is one which occupy the full rob+cdb if (s->accumulated_return_path_dcfclk_cycles[i] > s->max_Trpd_dcfclk_cycles) @@ -6109,7 +6050,7 @@ static noinline_for_stack bool CheckGlobalPrefetchAdmissibility(struct dml2_core p->impacted_dst_y_pre[i] = math_ceil2(p->impacted_dst_y_pre[i] / p->line_time[i], 0.25); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: i=%u impacted_Tpre=%f (k=%u)\n", __func__, i, p->impacted_dst_y_pre[i], k); + DML_LOG_VERBOSE("DML::%s: i=%u impacted_Tpre=%f (k=%u)\n", __func__, i, p->impacted_dst_y_pre[i], k); #endif } @@ -6120,8 +6061,8 @@ static noinline_for_stack bool CheckGlobalPrefetchAdmissibility(struct dml2_core *p->recalc_prefetch_schedule = 1; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: i=%u Tpre_rounded=%f\n", __func__, i, p->Tpre_rounded[i]); - dml2_printf("DML::%s: i=%u Tpre_oto=%f\n", __func__, i, p->Tpre_oto[i]); + DML_LOG_VERBOSE("DML::%s: i=%u Tpre_rounded=%f\n", __func__, i, p->Tpre_rounded[i]); + DML_LOG_VERBOSE("DML::%s: i=%u Tpre_oto=%f\n", __func__, i, p->Tpre_oto[i]); #endif } } else { @@ -6131,8 +6072,8 @@ static noinline_for_stack bool CheckGlobalPrefetchAdmissibility(struct dml2_core } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: prefetch_global_check_passed=%u\n", __func__, s->prefetch_global_check_passed); - dml2_printf("DML::%s: recalc_prefetch_schedule=%u\n", __func__, *p->recalc_prefetch_schedule); + DML_LOG_VERBOSE("DML::%s: prefetch_global_check_passed=%u\n", __func__, s->prefetch_global_check_passed); + DML_LOG_VERBOSE("DML::%s: recalc_prefetch_schedule=%u\n", __func__, *p->recalc_prefetch_schedule); #endif return s->prefetch_global_check_passed; @@ -6150,8 +6091,8 @@ static void calculate_peak_bandwidth_required( memset(l, 0, sizeof(struct dml2_core_shared_calculate_peak_bandwidth_required_locals)); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: inc_flip_bw = %d\n", __func__, p->inc_flip_bw); - dml2_printf("DML::%s: NumberOfActiveSurfaces = %d\n", __func__, p->num_active_planes); + DML_LOG_VERBOSE("DML::%s: inc_flip_bw = %d\n", __func__, p->inc_flip_bw); + DML_LOG_VERBOSE("DML::%s: NumberOfActiveSurfaces = %d\n", __func__, p->num_active_planes); #endif for (unsigned int k = 0; k < p->num_active_planes; ++k) { @@ -6180,7 +6121,7 @@ static void calculate_peak_bandwidth_required( p->surface_read_bandwidth_c, l->zero_array, //PrefetchBandwidthLuma, l->zero_array, //PrefetchBandwidthChroma, - l->zero_array, //PrefetchBWOTO + l->zero_array, //PrefetchBWMax l->zero_array, l->zero_array, l->zero_array, @@ -6217,7 +6158,7 @@ static void calculate_peak_bandwidth_required( p->surface_read_bandwidth_c, l->zero_array, //PrefetchBandwidthLuma, l->zero_array, //PrefetchBandwidthChroma, - l->zero_array, //PrefetchBWOTO + l->zero_array, //PrefetchBWMax p->excess_vactive_fill_bw_l, p->excess_vactive_fill_bw_c, p->cursor_bw, @@ -6254,7 +6195,7 @@ static void calculate_peak_bandwidth_required( p->surface_read_bandwidth_c, p->prefetch_bandwidth_l, p->prefetch_bandwidth_c, - p->prefetch_bandwidth_oto, // to prevent ms/mp mismatch when oto bw > total vactive bw + p->prefetch_bandwidth_max, // to prevent ms/mp mismatches where mp prefetch bw > ms prefetch bw p->excess_vactive_fill_bw_l, p->excess_vactive_fill_bw_c, p->cursor_bw, @@ -6291,7 +6232,7 @@ static void calculate_peak_bandwidth_required( p->surface_read_bandwidth_c, p->prefetch_bandwidth_l, p->prefetch_bandwidth_c, - p->prefetch_bandwidth_oto, // to prevent ms/mp mismatch when oto bw > total vactive bw + p->prefetch_bandwidth_max, // to prevent ms/mp mismatch where mp prefetch bw > ms prefetch bw p->excess_vactive_fill_bw_l, p->excess_vactive_fill_bw_c, p->cursor_bw, @@ -6328,7 +6269,7 @@ static void calculate_peak_bandwidth_required( p->surface_read_bandwidth_c, p->prefetch_bandwidth_l, p->prefetch_bandwidth_c, - p->prefetch_bandwidth_oto, // to prevent ms/mp mismatch when oto bw > total vactive bw + p->prefetch_bandwidth_max, // to prevent ms/mp mismatches where mp prefetch bw > ms prefetch bw p->excess_vactive_fill_bw_l, p->excess_vactive_fill_bw_c, p->cursor_bw, @@ -6347,12 +6288,12 @@ static void calculate_peak_bandwidth_required( p->surface_peak_required_bw[m][n]); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: urg_vactive_bandwidth_required%s[%s][%s]=%f\n", __func__, (p->inc_flip_bw ? "_flip" : ""), dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->urg_vactive_bandwidth_required[m][n]); - dml2_printf("DML::%s: urg_bandwidth_required%s[%s][%s]=%f\n", __func__, (p->inc_flip_bw ? "_flip" : ""), dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->urg_bandwidth_required[m][n]); - dml2_printf("DML::%s: urg_bandwidth_required_qual[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->urg_bandwidth_required[m][n]); - dml2_printf("DML::%s: non_urg_bandwidth_required%s[%s][%s]=%f\n", __func__, (p->inc_flip_bw ? "_flip" : ""), dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->non_urg_bandwidth_required[m][n]); + DML_LOG_VERBOSE("DML::%s: urg_vactive_bandwidth_required%s[%s][%s]=%f\n", __func__, (p->inc_flip_bw ? "_flip" : ""), dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->urg_vactive_bandwidth_required[m][n]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_required%s[%s][%s]=%f\n", __func__, (p->inc_flip_bw ? "_flip" : ""), dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->urg_bandwidth_required[m][n]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_required_qual[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->urg_bandwidth_required[m][n]); + DML_LOG_VERBOSE("DML::%s: non_urg_bandwidth_required%s[%s][%s]=%f\n", __func__, (p->inc_flip_bw ? "_flip" : ""), dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->non_urg_bandwidth_required[m][n]); #endif - DML2_ASSERT(p->urg_bandwidth_required[m][n] >= p->non_urg_bandwidth_required[m][n]); + DML_ASSERT(p->urg_bandwidth_required[m][n] >= p->non_urg_bandwidth_required[m][n]); } } } @@ -6414,18 +6355,18 @@ static void check_urgent_bandwidth_support( } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: frac_urg_bandwidth_nom_sdp = %f\n", __func__, frac_urg_bandwidth_nom_sdp); - dml2_printf("DML::%s: frac_urg_bandwidth_nom_dram = %f\n", __func__, frac_urg_bandwidth_nom_dram); - dml2_printf("DML::%s: frac_urg_bandwidth_nom = %f\n", __func__, *frac_urg_bandwidth_nom); + DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_nom_sdp = %f\n", __func__, frac_urg_bandwidth_nom_sdp); + DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_nom_dram = %f\n", __func__, frac_urg_bandwidth_nom_dram); + DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_nom = %f\n", __func__, *frac_urg_bandwidth_nom); - dml2_printf("DML::%s: frac_urg_bandwidth_mall_sdp = %f\n", __func__, frac_urg_bandwidth_mall_sdp); - dml2_printf("DML::%s: frac_urg_bandwidth_mall_dram = %f\n", __func__, frac_urg_bandwidth_mall_dram); - dml2_printf("DML::%s: frac_urg_bandwidth_mall = %f\n", __func__, *frac_urg_bandwidth_mall); - dml2_printf("DML::%s: bandwidth_support_ok = %d\n", __func__, *bandwidth_support_ok); + DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_mall_sdp = %f\n", __func__, frac_urg_bandwidth_mall_sdp); + DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_mall_dram = %f\n", __func__, frac_urg_bandwidth_mall_dram); + DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_mall = %f\n", __func__, *frac_urg_bandwidth_mall); + DML_LOG_VERBOSE("DML::%s: bandwidth_support_ok = %d\n", __func__, *bandwidth_support_ok); for (unsigned int m = 0; m < dml2_core_internal_soc_state_max; m++) { for (unsigned int n = 0; n < dml2_core_internal_bw_max; n++) { - dml2_printf("DML::%s: state:%s bw_type:%s urg_bandwidth_available=%f %s urg_bandwidth_required=%f\n", + DML_LOG_VERBOSE("DML::%s: state:%s bw_type:%s urg_bandwidth_available=%f %s urg_bandwidth_required=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), urg_bandwidth_available[m][n], (urg_bandwidth_available[m][n] < urg_bandwidth_required[m][n]) ? "<" : ">=", urg_bandwidth_required[m][n]); } @@ -6446,14 +6387,14 @@ static double get_bandwidth_available_for_immediate_flip(enum dml2_core_internal flip_bw_available_mbps = flip_bw_available_sdp_mbps < flip_bw_available_dram_mbps ? flip_bw_available_sdp_mbps : flip_bw_available_dram_mbps; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: eval_state = %s\n", __func__, dml2_core_internal_soc_state_type_str(eval_state)); - dml2_printf("DML::%s: urg_bandwidth_available_sdp_mbps = %f\n", __func__, urg_bandwidth_available[eval_state][dml2_core_internal_bw_sdp]); - dml2_printf("DML::%s: urg_bandwidth_available_dram_mbps = %f\n", __func__, urg_bandwidth_available[eval_state][dml2_core_internal_bw_dram]); - dml2_printf("DML::%s: urg_bandwidth_required_sdp_mbps = %f\n", __func__, urg_bandwidth_required[eval_state][dml2_core_internal_bw_sdp]); - dml2_printf("DML::%s: urg_bandwidth_required_dram_mbps = %f\n", __func__, urg_bandwidth_required[eval_state][dml2_core_internal_bw_dram]); - dml2_printf("DML::%s: flip_bw_available_sdp_mbps = %f\n", __func__, flip_bw_available_sdp_mbps); - dml2_printf("DML::%s: flip_bw_available_dram_mbps = %f\n", __func__, flip_bw_available_dram_mbps); - dml2_printf("DML::%s: flip_bw_available_mbps = %f\n", __func__, flip_bw_available_mbps); + DML_LOG_VERBOSE("DML::%s: eval_state = %s\n", __func__, dml2_core_internal_soc_state_type_str(eval_state)); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available_sdp_mbps = %f\n", __func__, urg_bandwidth_available[eval_state][dml2_core_internal_bw_sdp]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available_dram_mbps = %f\n", __func__, urg_bandwidth_available[eval_state][dml2_core_internal_bw_dram]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_required_sdp_mbps = %f\n", __func__, urg_bandwidth_required[eval_state][dml2_core_internal_bw_sdp]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_required_dram_mbps = %f\n", __func__, urg_bandwidth_required[eval_state][dml2_core_internal_bw_dram]); + DML_LOG_VERBOSE("DML::%s: flip_bw_available_sdp_mbps = %f\n", __func__, flip_bw_available_sdp_mbps); + DML_LOG_VERBOSE("DML::%s: flip_bw_available_dram_mbps = %f\n", __func__, flip_bw_available_dram_mbps); + DML_LOG_VERBOSE("DML::%s: flip_bw_available_mbps = %f\n", __func__, flip_bw_available_mbps); #endif return flip_bw_available_mbps; @@ -6478,28 +6419,28 @@ static void calculate_immediate_flip_bandwidth_support( *flip_bandwidth_support_ok &= urg_bandwidth_available[eval_state][n] >= urg_bandwidth_required_flip[eval_state][n]; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: n = %s\n", __func__, dml2_core_internal_bw_type_str(n)); - dml2_printf("DML::%s: urg_bandwidth_available = %f\n", __func__, urg_bandwidth_available[eval_state][n]); - dml2_printf("DML::%s: non_urg_bandwidth_required_flip = %f\n", __func__, non_urg_bandwidth_required_flip[eval_state][n]); - dml2_printf("DML::%s: urg_bandwidth_required_flip = %f\n", __func__, urg_bandwidth_required_flip[eval_state][n]); - dml2_printf("DML::%s: flip_bandwidth_support_ok = %d\n", __func__, *flip_bandwidth_support_ok); + DML_LOG_VERBOSE("DML::%s: n = %s\n", __func__, dml2_core_internal_bw_type_str(n)); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available = %f\n", __func__, urg_bandwidth_available[eval_state][n]); + DML_LOG_VERBOSE("DML::%s: non_urg_bandwidth_required_flip = %f\n", __func__, non_urg_bandwidth_required_flip[eval_state][n]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_required_flip = %f\n", __func__, urg_bandwidth_required_flip[eval_state][n]); + DML_LOG_VERBOSE("DML::%s: flip_bandwidth_support_ok = %d\n", __func__, *flip_bandwidth_support_ok); #endif - DML2_ASSERT(urg_bandwidth_required_flip[eval_state][n] >= non_urg_bandwidth_required_flip[eval_state][n]); + DML_ASSERT(urg_bandwidth_required_flip[eval_state][n] >= non_urg_bandwidth_required_flip[eval_state][n]); } *frac_urg_bandwidth_flip = (frac_urg_bw_flip_sdp > frac_urg_bw_flip_dram) ? frac_urg_bw_flip_sdp : frac_urg_bw_flip_dram; *flip_bandwidth_support_ok &= (*frac_urg_bandwidth_flip <= 1.0); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: eval_state = %s\n", __func__, dml2_core_internal_soc_state_type_str(eval_state)); - dml2_printf("DML::%s: frac_urg_bw_flip_sdp = %f\n", __func__, frac_urg_bw_flip_sdp); - dml2_printf("DML::%s: frac_urg_bw_flip_dram = %f\n", __func__, frac_urg_bw_flip_dram); - dml2_printf("DML::%s: frac_urg_bandwidth_flip = %f\n", __func__, *frac_urg_bandwidth_flip); - dml2_printf("DML::%s: flip_bandwidth_support_ok = %d\n", __func__, *flip_bandwidth_support_ok); + DML_LOG_VERBOSE("DML::%s: eval_state = %s\n", __func__, dml2_core_internal_soc_state_type_str(eval_state)); + DML_LOG_VERBOSE("DML::%s: frac_urg_bw_flip_sdp = %f\n", __func__, frac_urg_bw_flip_sdp); + DML_LOG_VERBOSE("DML::%s: frac_urg_bw_flip_dram = %f\n", __func__, frac_urg_bw_flip_dram); + DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_flip = %f\n", __func__, *frac_urg_bandwidth_flip); + DML_LOG_VERBOSE("DML::%s: flip_bandwidth_support_ok = %d\n", __func__, *flip_bandwidth_support_ok); for (unsigned int m = 0; m < dml2_core_internal_soc_state_max; m++) { for (unsigned int n = 0; n < dml2_core_internal_bw_max; n++) { - dml2_printf("DML::%s: state:%s bw_type:%s, urg_bandwidth_available=%f %s urg_bandwidth_required=%f\n", + DML_LOG_VERBOSE("DML::%s: state:%s bw_type:%s, urg_bandwidth_available=%f %s urg_bandwidth_required=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), urg_bandwidth_available[m][n], (urg_bandwidth_available[m][n] < urg_bandwidth_required_flip[m][n]) ? "<" : ">=", urg_bandwidth_required_flip[m][n]); } @@ -6549,27 +6490,27 @@ static void CalculateFlipSchedule( l->dpte_row_bytes = DPTEBytesPerRow; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: GPUVMEnable = %u\n", __func__, GPUVMEnable); - dml2_printf("DML::%s: ip.max_flip_time_us = %d\n", __func__, max_flip_time_us); - dml2_printf("DML::%s: ip.max_flip_time_lines = %d\n", __func__, max_flip_time_lines); - dml2_printf("DML::%s: BandwidthAvailableForImmediateFlip = %f\n", __func__, BandwidthAvailableForImmediateFlip); - dml2_printf("DML::%s: TotImmediateFlipBytes = %u\n", __func__, TotImmediateFlipBytes); - dml2_printf("DML::%s: use_lb_flip_bw = %u\n", __func__, use_lb_flip_bw); - dml2_printf("DML::%s: iflip_enable = %u\n", __func__, iflip_enable); - dml2_printf("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, HostVMInefficiencyFactor); - dml2_printf("DML::%s: LineTime = %f\n", __func__, LineTime); - dml2_printf("DML::%s: Tno_bw_flip = %f\n", __func__, Tno_bw_flip); - dml2_printf("DML::%s: Tvm_trips_flip = %f\n", __func__, Tvm_trips_flip); - dml2_printf("DML::%s: Tr0_trips_flip = %f\n", __func__, Tr0_trips_flip); - dml2_printf("DML::%s: Tvm_trips_flip_rounded = %f\n", __func__, Tvm_trips_flip_rounded); - dml2_printf("DML::%s: Tr0_trips_flip_rounded = %f\n", __func__, Tr0_trips_flip_rounded); - dml2_printf("DML::%s: vm_bytes = %f\n", __func__, vm_bytes); - dml2_printf("DML::%s: DPTEBytesPerRow = %f\n", __func__, DPTEBytesPerRow); - dml2_printf("DML::%s: meta_row_bytes = %d\n", __func__, meta_row_bytes); - dml2_printf("DML::%s: dpte_row_bytes = %f\n", __func__, l->dpte_row_bytes); - dml2_printf("DML::%s: dpte_row_height = %d\n", __func__, dpte_row_height); - dml2_printf("DML::%s: meta_row_height = %d\n", __func__, meta_row_height); - dml2_printf("DML::%s: VRatio = %f\n", __func__, VRatio); + DML_LOG_VERBOSE("DML::%s: GPUVMEnable = %u\n", __func__, GPUVMEnable); + DML_LOG_VERBOSE("DML::%s: ip.max_flip_time_us = %d\n", __func__, max_flip_time_us); + DML_LOG_VERBOSE("DML::%s: ip.max_flip_time_lines = %d\n", __func__, max_flip_time_lines); + DML_LOG_VERBOSE("DML::%s: BandwidthAvailableForImmediateFlip = %f\n", __func__, BandwidthAvailableForImmediateFlip); + DML_LOG_VERBOSE("DML::%s: TotImmediateFlipBytes = %u\n", __func__, TotImmediateFlipBytes); + DML_LOG_VERBOSE("DML::%s: use_lb_flip_bw = %u\n", __func__, use_lb_flip_bw); + DML_LOG_VERBOSE("DML::%s: iflip_enable = %u\n", __func__, iflip_enable); + DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, HostVMInefficiencyFactor); + DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, LineTime); + DML_LOG_VERBOSE("DML::%s: Tno_bw_flip = %f\n", __func__, Tno_bw_flip); + DML_LOG_VERBOSE("DML::%s: Tvm_trips_flip = %f\n", __func__, Tvm_trips_flip); + DML_LOG_VERBOSE("DML::%s: Tr0_trips_flip = %f\n", __func__, Tr0_trips_flip); + DML_LOG_VERBOSE("DML::%s: Tvm_trips_flip_rounded = %f\n", __func__, Tvm_trips_flip_rounded); + DML_LOG_VERBOSE("DML::%s: Tr0_trips_flip_rounded = %f\n", __func__, Tr0_trips_flip_rounded); + DML_LOG_VERBOSE("DML::%s: vm_bytes = %f\n", __func__, vm_bytes); + DML_LOG_VERBOSE("DML::%s: DPTEBytesPerRow = %f\n", __func__, DPTEBytesPerRow); + DML_LOG_VERBOSE("DML::%s: meta_row_bytes = %d\n", __func__, meta_row_bytes); + DML_LOG_VERBOSE("DML::%s: dpte_row_bytes = %f\n", __func__, l->dpte_row_bytes); + DML_LOG_VERBOSE("DML::%s: dpte_row_height = %d\n", __func__, dpte_row_height); + DML_LOG_VERBOSE("DML::%s: meta_row_height = %d\n", __func__, meta_row_height); + DML_LOG_VERBOSE("DML::%s: VRatio = %f\n", __func__, VRatio); #endif if (TotImmediateFlipBytes > 0 && (GPUVMEnable || dcc_mrq_enable)) { @@ -6596,9 +6537,9 @@ static void CalculateFlipSchedule( l->min_row_time = l->min_row_height * LineTime / VRatio; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: min_row_time = %f\n", __func__, l->min_row_time); + DML_LOG_VERBOSE("DML::%s: min_row_time = %f\n", __func__, l->min_row_time); #endif - DML2_ASSERT(l->min_row_time > 0); + DML_ASSERT(l->min_row_time > 0); if (use_lb_flip_bw) { // For mode check, calculation the flip bw requirement with worst case flip time @@ -6619,20 +6560,20 @@ static void CalculateFlipSchedule( l->hvm_scaled_vm_bytes / (l->max_flip_time - Tno_bw_flip - 2 * Tr0_trips_flip_rounded), l->hvm_scaled_row_bytes / (l->max_flip_time - Tvm_trips_flip_rounded)); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: max_flip_time = %f\n", __func__, l->max_flip_time); - dml2_printf("DML::%s: total vm bytes (hvm ineff scaled) = %f\n", __func__, l->hvm_scaled_vm_bytes); - dml2_printf("DML::%s: total row bytes (%d row, hvm ineff scaled) = %f\n", __func__, l->num_rows, l->hvm_scaled_row_bytes); - dml2_printf("DML::%s: total vm+row bytes (hvm ineff scaled) = %f\n", __func__, l->hvm_scaled_vm_row_bytes); - dml2_printf("DML::%s: lb_flip_bw for vm and row = %f\n", __func__, l->hvm_scaled_vm_row_bytes / (l->max_flip_time - Tno_bw_flip)); - dml2_printf("DML::%s: lb_flip_bw for vm = %f\n", __func__, l->hvm_scaled_vm_bytes / (l->max_flip_time - Tno_bw_flip - 2 * Tr0_trips_flip_rounded)); - dml2_printf("DML::%s: lb_flip_bw for row = %f\n", __func__, l->hvm_scaled_row_bytes / (l->max_flip_time - Tvm_trips_flip_rounded)); + DML_LOG_VERBOSE("DML::%s: max_flip_time = %f\n", __func__, l->max_flip_time); + DML_LOG_VERBOSE("DML::%s: total vm bytes (hvm ineff scaled) = %f\n", __func__, l->hvm_scaled_vm_bytes); + DML_LOG_VERBOSE("DML::%s: total row bytes (%f row, hvm ineff scaled) = %f\n", __func__, l->num_rows, l->hvm_scaled_row_bytes); + DML_LOG_VERBOSE("DML::%s: total vm+row bytes (hvm ineff scaled) = %f\n", __func__, l->hvm_scaled_vm_row_bytes); + DML_LOG_VERBOSE("DML::%s: lb_flip_bw for vm and row = %f\n", __func__, l->hvm_scaled_vm_row_bytes / (l->max_flip_time - Tno_bw_flip)); + DML_LOG_VERBOSE("DML::%s: lb_flip_bw for vm = %f\n", __func__, l->hvm_scaled_vm_bytes / (l->max_flip_time - Tno_bw_flip - 2 * Tr0_trips_flip_rounded)); + DML_LOG_VERBOSE("DML::%s: lb_flip_bw for row = %f\n", __func__, l->hvm_scaled_row_bytes / (l->max_flip_time - Tvm_trips_flip_rounded)); if (l->lb_flip_bw > 0) { - dml2_printf("DML::%s: mode_support est Tvm_flip = %f (bw-based)\n", __func__, Tno_bw_flip + l->hvm_scaled_vm_bytes / l->lb_flip_bw); - dml2_printf("DML::%s: mode_support est Tr0_flip = %f (bw-based)\n", __func__, l->hvm_scaled_row_bytes / l->lb_flip_bw / l->num_rows); - dml2_printf("DML::%s: mode_support est dst_y_per_vm_flip = %f (bw-based)\n", __func__, Tno_bw_flip + l->hvm_scaled_vm_bytes / l->lb_flip_bw / LineTime); - dml2_printf("DML::%s: mode_support est dst_y_per_row_flip = %f (bw-based)\n", __func__, l->hvm_scaled_row_bytes / l->lb_flip_bw / LineTime / l->num_rows); - dml2_printf("DML::%s: Tvm_trips_flip_rounded + 2*Tr0_trips_flip_rounded = %f\n", __func__, (Tvm_trips_flip_rounded + 2 * Tr0_trips_flip_rounded)); + DML_LOG_VERBOSE("DML::%s: mode_support est Tvm_flip = %f (bw-based)\n", __func__, Tno_bw_flip + l->hvm_scaled_vm_bytes / l->lb_flip_bw); + DML_LOG_VERBOSE("DML::%s: mode_support est Tr0_flip = %f (bw-based)\n", __func__, l->hvm_scaled_row_bytes / l->lb_flip_bw / l->num_rows); + DML_LOG_VERBOSE("DML::%s: mode_support est dst_y_per_vm_flip = %f (bw-based)\n", __func__, Tno_bw_flip + l->hvm_scaled_vm_bytes / l->lb_flip_bw / LineTime); + DML_LOG_VERBOSE("DML::%s: mode_support est dst_y_per_row_flip = %f (bw-based)\n", __func__, l->hvm_scaled_row_bytes / l->lb_flip_bw / LineTime / l->num_rows); + DML_LOG_VERBOSE("DML::%s: Tvm_trips_flip_rounded + 2*Tr0_trips_flip_rounded = %f\n", __func__, (Tvm_trips_flip_rounded + 2 * Tr0_trips_flip_rounded)); } #endif l->lb_flip_bw = math_max3(l->lb_flip_bw, @@ -6640,8 +6581,8 @@ static void CalculateFlipSchedule( (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes) / (15 * LineTime)); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: lb_flip_bw for vm reg limit = %f\n", __func__, l->hvm_scaled_vm_bytes / (31 * LineTime) - Tno_bw_flip); - dml2_printf("DML::%s: lb_flip_bw for row reg limit = %f\n", __func__, (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes) / (15 * LineTime)); + DML_LOG_VERBOSE("DML::%s: lb_flip_bw for vm reg limit = %f\n", __func__, l->hvm_scaled_vm_bytes / (31 * LineTime) - Tno_bw_flip); + DML_LOG_VERBOSE("DML::%s: lb_flip_bw for row reg limit = %f\n", __func__, (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes) / (15 * LineTime)); #endif } @@ -6653,13 +6594,12 @@ static void CalculateFlipSchedule( } else { if (iflip_enable) { l->ImmediateFlipBW = (double)per_pipe_flip_bytes * BandwidthAvailableForImmediateFlip / (double)TotImmediateFlipBytes; // flip_bw(i) - double portion = (double)per_pipe_flip_bytes / (double)TotImmediateFlipBytes; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: per_pipe_flip_bytes = %d\n", __func__, per_pipe_flip_bytes); - dml2_printf("DML::%s: BandwidthAvailableForImmediateFlip = %f\n", __func__, BandwidthAvailableForImmediateFlip); - dml2_printf("DML::%s: ImmediateFlipBW = %f\n", __func__, l->ImmediateFlipBW); - dml2_printf("DML::%s: portion of flip bw = %f\n", __func__, portion); + DML_LOG_VERBOSE("DML::%s: per_pipe_flip_bytes = %d\n", __func__, per_pipe_flip_bytes); + DML_LOG_VERBOSE("DML::%s: BandwidthAvailableForImmediateFlip = %f\n", __func__, BandwidthAvailableForImmediateFlip); + DML_LOG_VERBOSE("DML::%s: ImmediateFlipBW = %f\n", __func__, l->ImmediateFlipBW); + DML_LOG_VERBOSE("DML::%s: portion of flip bw = %f\n", __func__, (double)per_pipe_flip_bytes / (double)TotImmediateFlipBytes); #endif if (l->ImmediateFlipBW == 0) { l->Tvm_flip = 0; @@ -6674,11 +6614,11 @@ static void CalculateFlipSchedule( LineTime / 4.0); } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: total vm bytes (hvm ineff scaled) = %f\n", __func__, vm_bytes * HostVMInefficiencyFactor); - dml2_printf("DML::%s: total row bytes (hvm ineff scaled, one row) = %f\n", __func__, (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes)); + DML_LOG_VERBOSE("DML::%s: total vm bytes (hvm ineff scaled) = %f\n", __func__, vm_bytes * HostVMInefficiencyFactor); + DML_LOG_VERBOSE("DML::%s: total row bytes (hvm ineff scaled, one row) = %f\n", __func__, (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes)); - dml2_printf("DML::%s: Tvm_flip = %f (bw-based), Tvm_trips_flip = %f (latency-based)\n", __func__, Tno_bw_flip + vm_bytes * HostVMInefficiencyFactor / l->ImmediateFlipBW, Tvm_trips_flip); - dml2_printf("DML::%s: Tr0_flip = %f (bw-based), Tr0_trips_flip = %f (latency-based)\n", __func__, (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes) / l->ImmediateFlipBW, Tr0_trips_flip); + DML_LOG_VERBOSE("DML::%s: Tvm_flip = %f (bw-based), Tvm_trips_flip = %f (latency-based)\n", __func__, Tno_bw_flip + vm_bytes * HostVMInefficiencyFactor / l->ImmediateFlipBW, Tvm_trips_flip); + DML_LOG_VERBOSE("DML::%s: Tr0_flip = %f (bw-based), Tr0_trips_flip = %f (latency-based)\n", __func__, (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes) / l->ImmediateFlipBW, Tr0_trips_flip); #endif *dst_y_per_vm_flip = math_ceil2(4.0 * (l->Tvm_flip / LineTime), 1.0) / 4.0; *dst_y_per_row_flip = math_ceil2(4.0 * (l->Tr0_flip / LineTime), 1.0) / 4.0; @@ -6711,14 +6651,14 @@ static void CalculateFlipSchedule( #ifdef __DML_VBA_DEBUG__ if (!use_lb_flip_bw) { - dml2_printf("DML::%s: dst_y_per_vm_flip = %f (should be < 32)\n", __func__, *dst_y_per_vm_flip); - dml2_printf("DML::%s: dst_y_per_row_flip = %f (should be < 16)\n", __func__, *dst_y_per_row_flip); - dml2_printf("DML::%s: Tvm_flip = %f (final)\n", __func__, l->Tvm_flip); - dml2_printf("DML::%s: Tr0_flip = %f (final)\n", __func__, l->Tr0_flip); - dml2_printf("DML::%s: Tvm_flip + 2*Tr0_flip = %f (should be <= min_row_time=%f)\n", __func__, l->Tvm_flip + 2 * l->Tr0_flip, l->min_row_time); + DML_LOG_VERBOSE("DML::%s: dst_y_per_vm_flip = %f (should be < 32)\n", __func__, *dst_y_per_vm_flip); + DML_LOG_VERBOSE("DML::%s: dst_y_per_row_flip = %f (should be < 16)\n", __func__, *dst_y_per_row_flip); + DML_LOG_VERBOSE("DML::%s: Tvm_flip = %f (final)\n", __func__, l->Tvm_flip); + DML_LOG_VERBOSE("DML::%s: Tr0_flip = %f (final)\n", __func__, l->Tr0_flip); + DML_LOG_VERBOSE("DML::%s: Tvm_flip + 2*Tr0_flip = %f (should be <= min_row_time=%f)\n", __func__, l->Tvm_flip + 2 * l->Tr0_flip, l->min_row_time); } - dml2_printf("DML::%s: final_flip_bw = %f\n", __func__, *final_flip_bw); - dml2_printf("DML::%s: ImmediateFlipSupportedForPipe = %u\n", __func__, *ImmediateFlipSupportedForPipe); + DML_LOG_VERBOSE("DML::%s: final_flip_bw = %f\n", __func__, *final_flip_bw); + DML_LOG_VERBOSE("DML::%s: ImmediateFlipSupportedForPipe = %u\n", __func__, *ImmediateFlipSupportedForPipe); #endif } @@ -6736,7 +6676,7 @@ static void CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( p->Watermark->UrgentWatermark = p->mmSOCParameters.UrgentLatency + p->mmSOCParameters.ExtraLatency; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: UrgentWatermark = %f\n", __func__, p->Watermark->UrgentWatermark); + DML_LOG_VERBOSE("DML::%s: UrgentWatermark = %f\n", __func__, p->Watermark->UrgentWatermark); #endif p->Watermark->USRRetrainingWatermark = p->mmSOCParameters.UrgentLatency + p->mmSOCParameters.ExtraLatency + p->mmSOCParameters.USRRetrainingLatency + p->mmSOCParameters.SMNLatency; @@ -6755,20 +6695,20 @@ static void CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( p->Watermark->temp_read_or_ppt_watermark_us = p->mmSOCParameters.g6_temp_read_blackout_us + p->Watermark->UrgentWatermark; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: UrgentLatency = %f\n", __func__, p->mmSOCParameters.UrgentLatency); - dml2_printf("DML::%s: ExtraLatency = %f\n", __func__, p->mmSOCParameters.ExtraLatency); - dml2_printf("DML::%s: DRAMClockChangeLatency = %f\n", __func__, p->mmSOCParameters.DRAMClockChangeLatency); - dml2_printf("DML::%s: SREnterPlusExitZ8Time = %f\n", __func__, p->mmSOCParameters.SREnterPlusExitZ8Time); - dml2_printf("DML::%s: SREnterPlusExitTime = %f\n", __func__, p->mmSOCParameters.SREnterPlusExitTime); - dml2_printf("DML::%s: UrgentWatermark = %f\n", __func__, p->Watermark->UrgentWatermark); - dml2_printf("DML::%s: USRRetrainingWatermark = %f\n", __func__, p->Watermark->USRRetrainingWatermark); - dml2_printf("DML::%s: DRAMClockChangeWatermark = %f\n", __func__, p->Watermark->DRAMClockChangeWatermark); - dml2_printf("DML::%s: FCLKChangeWatermark = %f\n", __func__, p->Watermark->FCLKChangeWatermark); - dml2_printf("DML::%s: StutterExitWatermark = %f\n", __func__, p->Watermark->StutterExitWatermark); - dml2_printf("DML::%s: StutterEnterPlusExitWatermark = %f\n", __func__, p->Watermark->StutterEnterPlusExitWatermark); - dml2_printf("DML::%s: Z8StutterExitWatermark = %f\n", __func__, p->Watermark->Z8StutterExitWatermark); - dml2_printf("DML::%s: Z8StutterEnterPlusExitWatermark = %f\n", __func__, p->Watermark->Z8StutterEnterPlusExitWatermark); - dml2_printf("DML::%s: temp_read_or_ppt_watermark_us = %f\n", __func__, p->Watermark->temp_read_or_ppt_watermark_us); + DML_LOG_VERBOSE("DML::%s: UrgentLatency = %f\n", __func__, p->mmSOCParameters.UrgentLatency); + DML_LOG_VERBOSE("DML::%s: ExtraLatency = %f\n", __func__, p->mmSOCParameters.ExtraLatency); + DML_LOG_VERBOSE("DML::%s: DRAMClockChangeLatency = %f\n", __func__, p->mmSOCParameters.DRAMClockChangeLatency); + DML_LOG_VERBOSE("DML::%s: SREnterPlusExitZ8Time = %f\n", __func__, p->mmSOCParameters.SREnterPlusExitZ8Time); + DML_LOG_VERBOSE("DML::%s: SREnterPlusExitTime = %f\n", __func__, p->mmSOCParameters.SREnterPlusExitTime); + DML_LOG_VERBOSE("DML::%s: UrgentWatermark = %f\n", __func__, p->Watermark->UrgentWatermark); + DML_LOG_VERBOSE("DML::%s: USRRetrainingWatermark = %f\n", __func__, p->Watermark->USRRetrainingWatermark); + DML_LOG_VERBOSE("DML::%s: DRAMClockChangeWatermark = %f\n", __func__, p->Watermark->DRAMClockChangeWatermark); + DML_LOG_VERBOSE("DML::%s: FCLKChangeWatermark = %f\n", __func__, p->Watermark->FCLKChangeWatermark); + DML_LOG_VERBOSE("DML::%s: StutterExitWatermark = %f\n", __func__, p->Watermark->StutterExitWatermark); + DML_LOG_VERBOSE("DML::%s: StutterEnterPlusExitWatermark = %f\n", __func__, p->Watermark->StutterEnterPlusExitWatermark); + DML_LOG_VERBOSE("DML::%s: Z8StutterExitWatermark = %f\n", __func__, p->Watermark->Z8StutterExitWatermark); + DML_LOG_VERBOSE("DML::%s: Z8StutterEnterPlusExitWatermark = %f\n", __func__, p->Watermark->Z8StutterEnterPlusExitWatermark); + DML_LOG_VERBOSE("DML::%s: temp_read_or_ppt_watermark_us = %f\n", __func__, p->Watermark->temp_read_or_ppt_watermark_us); #endif s->TotalActiveWriteback = 0; @@ -6801,11 +6741,11 @@ static void CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( p->Watermark->WritebackFCLKChangeWatermark = p->Watermark->WritebackFCLKChangeWatermark + p->mmSOCParameters.USRRetrainingLatency; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: WritebackDRAMClockChangeWatermark = %f\n", __func__, p->Watermark->WritebackDRAMClockChangeWatermark); - dml2_printf("DML::%s: WritebackFCLKChangeWatermark = %f\n", __func__, p->Watermark->WritebackFCLKChangeWatermark); - dml2_printf("DML::%s: WritebackUrgentWatermark = %f\n", __func__, p->Watermark->WritebackUrgentWatermark); - dml2_printf("DML::%s: USRRetrainingRequired = %u\n", __func__, p->USRRetrainingRequired); - dml2_printf("DML::%s: USRRetrainingLatency = %f\n", __func__, p->mmSOCParameters.USRRetrainingLatency); + DML_LOG_VERBOSE("DML::%s: WritebackDRAMClockChangeWatermark = %f\n", __func__, p->Watermark->WritebackDRAMClockChangeWatermark); + DML_LOG_VERBOSE("DML::%s: WritebackFCLKChangeWatermark = %f\n", __func__, p->Watermark->WritebackFCLKChangeWatermark); + DML_LOG_VERBOSE("DML::%s: WritebackUrgentWatermark = %f\n", __func__, p->Watermark->WritebackUrgentWatermark); + DML_LOG_VERBOSE("DML::%s: USRRetrainingRequired = %u\n", __func__, p->USRRetrainingRequired); + DML_LOG_VERBOSE("DML::%s: USRRetrainingLatency = %f\n", __func__, p->mmSOCParameters.USRRetrainingLatency); #endif s->TotalPixelBW = 0.0; @@ -6836,11 +6776,11 @@ static void CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( s->LBLatencyHidingSourceLinesC[k] = (unsigned int)(math_min2((double)p->MaxLineBufferLines, math_floor2((double)p->LineBufferSize / LBBitPerPixel / ((double)p->SwathWidthC[k] / math_max2(h_ratio_c, 1.0)), 1)) - (v_taps_c - 1)); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, MaxLineBufferLines = %u\n", __func__, k, p->MaxLineBufferLines); - dml2_printf("DML::%s: k=%u, LineBufferSize = %u\n", __func__, k, p->LineBufferSize); - dml2_printf("DML::%s: k=%u, LBBitPerPixel = %u\n", __func__, k, LBBitPerPixel); - dml2_printf("DML::%s: k=%u, HRatio = %f\n", __func__, k, h_ratio); - dml2_printf("DML::%s: k=%u, VTaps = %f\n", __func__, k, v_taps); + DML_LOG_VERBOSE("DML::%s: k=%u, MaxLineBufferLines = %u\n", __func__, k, p->MaxLineBufferLines); + DML_LOG_VERBOSE("DML::%s: k=%u, LineBufferSize = %u\n", __func__, k, p->LineBufferSize); + DML_LOG_VERBOSE("DML::%s: k=%u, LBBitPerPixel = %f\n", __func__, k, LBBitPerPixel); + DML_LOG_VERBOSE("DML::%s: k=%u, HRatio = %f\n", __func__, k, h_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u, VTaps = %f\n", __func__, k, v_taps); #endif s->EffectiveLBLatencyHidingY = s->LBLatencyHidingSourceLinesY[k] / v_ratio * (h_total / pixel_clock_mhz); @@ -6943,16 +6883,16 @@ static void CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( s->sub_vp_lines_l = s->src_y_pstate_l + s->src_y_ahead_l + p->meta_row_height_l[k]; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, DETBufferSizeY = %u\n", __func__, k, p->DETBufferSizeY[k]); - dml2_printf("DML::%s: k=%u, BytePerPixelDETY = %f\n", __func__, k, p->BytePerPixelDETY[k]); - dml2_printf("DML::%s: k=%u, SwathWidthY = %u\n", __func__, k, p->SwathWidthY[k]); - dml2_printf("DML::%s: k=%u, SwathHeightY = %u\n", __func__, k, p->SwathHeightY[k]); - dml2_printf("DML::%s: k=%u, LBLatencyHidingSourceLinesY = %u\n", __func__, k, s->LBLatencyHidingSourceLinesY[k]); - dml2_printf("DML::%s: k=%u, dst_y_pstate = %u\n", __func__, k, s->dst_y_pstate); - dml2_printf("DML::%s: k=%u, src_y_pstate_l = %u\n", __func__, k, s->src_y_pstate_l); - dml2_printf("DML::%s: k=%u, src_y_ahead_l = %u\n", __func__, k, s->src_y_ahead_l); - dml2_printf("DML::%s: k=%u, meta_row_height_l = %u\n", __func__, k, p->meta_row_height_l[k]); - dml2_printf("DML::%s: k=%u, sub_vp_lines_l = %u\n", __func__, k, s->sub_vp_lines_l); + DML_LOG_VERBOSE("DML::%s: k=%u, DETBufferSizeY = %u\n", __func__, k, p->DETBufferSizeY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, BytePerPixelDETY = %f\n", __func__, k, p->BytePerPixelDETY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, SwathWidthY = %u\n", __func__, k, p->SwathWidthY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, SwathHeightY = %u\n", __func__, k, p->SwathHeightY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, LBLatencyHidingSourceLinesY = %u\n", __func__, k, s->LBLatencyHidingSourceLinesY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dst_y_pstate = %u\n", __func__, k, s->dst_y_pstate); + DML_LOG_VERBOSE("DML::%s: k=%u, src_y_pstate_l = %u\n", __func__, k, s->src_y_pstate_l); + DML_LOG_VERBOSE("DML::%s: k=%u, src_y_ahead_l = %u\n", __func__, k, s->src_y_ahead_l); + DML_LOG_VERBOSE("DML::%s: k=%u, meta_row_height_l = %u\n", __func__, k, p->meta_row_height_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, sub_vp_lines_l = %u\n", __func__, k, s->sub_vp_lines_l); #endif p->SubViewportLinesNeededInMALL[k] = s->sub_vp_lines_l; @@ -6967,10 +6907,10 @@ static void CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( p->SubViewportLinesNeededInMALL[k] = (unsigned int)(math_max2(s->sub_vp_lines_l, s->sub_vp_lines_c)); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, meta_row_height_c = %u\n", __func__, k, p->meta_row_height_c[k]); - dml2_printf("DML::%s: k=%u, src_y_pstate_c = %u\n", __func__, k, s->src_y_pstate_c); - dml2_printf("DML::%s: k=%u, src_y_ahead_c = %u\n", __func__, k, s->src_y_ahead_c); - dml2_printf("DML::%s: k=%u, sub_vp_lines_c = %u\n", __func__, k, s->sub_vp_lines_c); + DML_LOG_VERBOSE("DML::%s: k=%u, meta_row_height_c = %u\n", __func__, k, p->meta_row_height_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, src_y_pstate_c = %u\n", __func__, k, s->src_y_pstate_c); + DML_LOG_VERBOSE("DML::%s: k=%u, src_y_ahead_c = %u\n", __func__, k, s->src_y_ahead_c); + DML_LOG_VERBOSE("DML::%s: k=%u, sub_vp_lines_c = %u\n", __func__, k, s->sub_vp_lines_c); #endif } } @@ -6992,10 +6932,10 @@ static void CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: DRAMClockChangeSupport = %u\n", __func__, *p->global_dram_clock_change_supported); - dml2_printf("DML::%s: FCLKChangeSupport = %u\n", __func__, *p->global_fclk_change_supported); - dml2_printf("DML::%s: MaxActiveFCLKChangeLatencySupported = %f\n", __func__, *p->MaxActiveFCLKChangeLatencySupported); - dml2_printf("DML::%s: USRRetrainingSupport = %u\n", __func__, *p->USRRetrainingSupport); + DML_LOG_VERBOSE("DML::%s: DRAMClockChangeSupport = %u\n", __func__, *p->global_dram_clock_change_supported); + DML_LOG_VERBOSE("DML::%s: FCLKChangeSupport = %u\n", __func__, *p->global_fclk_change_supported); + DML_LOG_VERBOSE("DML::%s: MaxActiveFCLKChangeLatencySupported = %f\n", __func__, *p->MaxActiveFCLKChangeLatencySupported); + DML_LOG_VERBOSE("DML::%s: USRRetrainingSupport = %u\n", __func__, *p->USRRetrainingSupport); #endif } @@ -7141,7 +7081,7 @@ static unsigned int get_qos_param_index(unsigned long uclk_freq_khz, const struc unsigned int index = 0; for (i = 0; i < DML_MAX_CLK_TABLE_SIZE; i++) { - dml2_printf("DML::%s: per_uclk_dpm_params[%d].minimum_uclk_khz = %d\n", __func__, i, per_uclk_dpm_params[i].minimum_uclk_khz); + DML_LOG_VERBOSE("DML::%s: per_uclk_dpm_params[%d].minimum_uclk_khz = %ld\n", __func__, i, per_uclk_dpm_params[i].minimum_uclk_khz); if (i == 0) index = 0; @@ -7153,32 +7093,30 @@ static unsigned int get_qos_param_index(unsigned long uclk_freq_khz, const struc break; } } -#if defined(__DML_VBA_DEBUG__) - dml2_printf("DML::%s: uclk_freq_khz = %d\n", __func__, uclk_freq_khz); - dml2_printf("DML::%s: index = %d\n", __func__, index); -#endif + DML_LOG_VERBOSE("DML::%s: uclk_freq_khz = %ld\n", __func__, uclk_freq_khz); + DML_LOG_VERBOSE("DML::%s: index = %d\n", __func__, index); return index; } static unsigned int get_active_min_uclk_dpm_index(unsigned long uclk_freq_khz, const struct dml2_soc_state_table *clk_table) { unsigned int i; - bool clk_entry_found = 0; + bool clk_entry_found = false; for (i = 0; i < clk_table->uclk.num_clk_values; i++) { - dml2_printf("DML::%s: clk_table.uclk.clk_values_khz[%d] = %d\n", __func__, i, clk_table->uclk.clk_values_khz[i]); + DML_LOG_VERBOSE("DML::%s: clk_table.uclk.clk_values_khz[%d] = %ld\n", __func__, i, clk_table->uclk.clk_values_khz[i]); if (uclk_freq_khz == clk_table->uclk.clk_values_khz[i]) { - clk_entry_found = 1; + clk_entry_found = true; break; } } if (!clk_entry_found) - DML2_ASSERT(clk_entry_found); + DML_ASSERT(clk_entry_found); #if defined(__DML_VBA_DEBUG__) - dml2_printf("DML::%s: uclk_freq_khz = %ld\n", __func__, uclk_freq_khz); - dml2_printf("DML::%s: index = %d\n", __func__, i); + DML_LOG_VERBOSE("DML::%s: uclk_freq_khz = %ld\n", __func__, uclk_freq_khz); + DML_LOG_VERBOSE("DML::%s: index = %d\n", __func__, i); #endif return i; } @@ -7218,10 +7156,10 @@ static void calculate_hostvm_inefficiency_factor( if ((*HostVMInefficiencyFactorPrefetch < 4) && (remote_iommu_outstanding_translations < max_outstanding_reqs)) *HostVMInefficiencyFactorPrefetch = 4; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: urg_bandwidth_avail_active_pixel_and_vm = %f\n", __func__, urg_bandwidth_avail_active_pixel_and_vm); - dml2_printf("DML::%s: urg_bandwidth_avail_active_vm_only = %f\n", __func__, urg_bandwidth_avail_active_vm_only); - dml2_printf("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, *HostVMInefficiencyFactor); - dml2_printf("DML::%s: HostVMInefficiencyFactorPrefetch = %f\n", __func__, *HostVMInefficiencyFactorPrefetch); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_avail_active_pixel_and_vm = %f\n", __func__, urg_bandwidth_avail_active_pixel_and_vm); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_avail_active_vm_only = %f\n", __func__, urg_bandwidth_avail_active_vm_only); + DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, *HostVMInefficiencyFactor); + DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactorPrefetch = %f\n", __func__, *HostVMInefficiencyFactorPrefetch); #endif } } @@ -7335,30 +7273,660 @@ static void calculate_pstate_keepout_dst_lines( } } +static noinline_for_stack void dml_core_ms_prefetch_check(struct dml2_core_internal_display_mode_lib *mode_lib, + const struct dml2_display_cfg *display_cfg) +{ + struct dml2_core_calcs_mode_support_locals *s = &mode_lib->scratch.dml_core_mode_support_locals; + struct dml2_core_calcs_calculate_tdlut_setting_params *calculate_tdlut_setting_params = &mode_lib->scratch.calculate_tdlut_setting_params; + struct dml2_core_calcs_CalculatePrefetchSchedule_params *CalculatePrefetchSchedule_params = &mode_lib->scratch.CalculatePrefetchSchedule_params; + struct dml2_core_calcs_calculate_peak_bandwidth_required_params *calculate_peak_bandwidth_params = &mode_lib->scratch.calculate_peak_bandwidth_params; +#ifdef DML_GLOBAL_PREFETCH_CHECK + struct dml2_core_calcs_CheckGlobalPrefetchAdmissibility_params *CheckGlobalPrefetchAdmissibility_params = &mode_lib->scratch.CheckGlobalPrefetchAdmissibility_params; +#endif + struct dml2_core_calcs_CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params *CalculateWatermarks_params = &mode_lib->scratch.CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params; + + double min_return_bw_for_latency; + unsigned int k; + + mode_lib->ms.TimeCalc = 24 / mode_lib->ms.dcfclk_deepsleep; + + calculate_hostvm_inefficiency_factor( + &s->HostVMInefficiencyFactor, + &s->HostVMInefficiencyFactorPrefetch, + + display_cfg->gpuvm_enable, + display_cfg->hostvm_enable, + mode_lib->ip.remote_iommu_outstanding_translations, + mode_lib->soc.max_outstanding_reqs, + mode_lib->ms.support.urg_bandwidth_available_pixel_and_vm[dml2_core_internal_soc_state_sys_active], + mode_lib->ms.support.urg_bandwidth_available_vm_only[dml2_core_internal_soc_state_sys_active]); + + mode_lib->ms.Total3dlutActive = 0; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut) + mode_lib->ms.Total3dlutActive = mode_lib->ms.Total3dlutActive + 1; + + // Calculate tdlut schedule related terms + calculate_tdlut_setting_params->dispclk_mhz = mode_lib->ms.RequiredDISPCLK; + calculate_tdlut_setting_params->setup_for_tdlut = display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut; + calculate_tdlut_setting_params->tdlut_width_mode = display_cfg->plane_descriptors[k].tdlut.tdlut_width_mode; + calculate_tdlut_setting_params->tdlut_addressing_mode = display_cfg->plane_descriptors[k].tdlut.tdlut_addressing_mode; + calculate_tdlut_setting_params->cursor_buffer_size = mode_lib->ip.cursor_buffer_size; + calculate_tdlut_setting_params->gpuvm_enable = display_cfg->gpuvm_enable; + calculate_tdlut_setting_params->gpuvm_page_size_kbytes = display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes; + calculate_tdlut_setting_params->tdlut_mpc_width_flag = display_cfg->plane_descriptors[k].tdlut.tdlut_mpc_width_flag; + calculate_tdlut_setting_params->is_gfx11 = dml_get_gfx_version(display_cfg->plane_descriptors[k].surface.tiling); + + // output + calculate_tdlut_setting_params->tdlut_pte_bytes_per_frame = &s->tdlut_pte_bytes_per_frame[k]; + calculate_tdlut_setting_params->tdlut_bytes_per_frame = &s->tdlut_bytes_per_frame[k]; + calculate_tdlut_setting_params->tdlut_groups_per_2row_ub = &s->tdlut_groups_per_2row_ub[k]; + calculate_tdlut_setting_params->tdlut_opt_time = &s->tdlut_opt_time[k]; + calculate_tdlut_setting_params->tdlut_drain_time = &s->tdlut_drain_time[k]; + calculate_tdlut_setting_params->tdlut_bytes_to_deliver = &s->tdlut_bytes_to_deliver[k]; + calculate_tdlut_setting_params->tdlut_bytes_per_group = &s->tdlut_bytes_per_group[k]; + + calculate_tdlut_setting(&mode_lib->scratch, calculate_tdlut_setting_params); + } + + min_return_bw_for_latency = mode_lib->ms.support.urg_bandwidth_available_min_latency[dml2_core_internal_soc_state_sys_active]; + + if (mode_lib->soc.qos_parameters.qos_type == dml2_qos_param_type_dcn3) + s->ReorderingBytes = (unsigned int)(mode_lib->soc.clk_table.dram_config.channel_count * math_max3(mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_pixel_only_bytes, + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_pixel_and_vm_bytes, + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_vm_only_bytes)); + + CalculateExtraLatency( + display_cfg, + mode_lib->ip.rob_buffer_size_kbytes, + mode_lib->soc.qos_parameters.qos_params.dcn32x.loaded_round_trip_latency_fclk_cycles, + s->ReorderingBytes, + mode_lib->ms.DCFCLK, + mode_lib->ms.FabricClock, + mode_lib->ip.pixel_chunk_size_kbytes, + min_return_bw_for_latency, + mode_lib->ms.num_active_planes, + mode_lib->ms.NoOfDPP, + mode_lib->ms.dpte_group_bytes, + s->tdlut_bytes_per_group, + s->HostVMInefficiencyFactor, + s->HostVMInefficiencyFactorPrefetch, + mode_lib->soc.hostvm_min_page_size_kbytes, + mode_lib->soc.qos_parameters.qos_type, + !(display_cfg->overrides.max_outstanding_when_urgent_expected_disable), + mode_lib->soc.max_outstanding_reqs, + mode_lib->ms.support.request_size_bytes_luma, + mode_lib->ms.support.request_size_bytes_chroma, + mode_lib->ip.meta_chunk_size_kbytes, + mode_lib->ip.dchub_arb_to_ret_delay, + mode_lib->ms.TripToMemory, + mode_lib->ip.hostvm_mode, + + // output + &mode_lib->ms.ExtraLatency, + &mode_lib->ms.ExtraLatency_sr, + &mode_lib->ms.ExtraLatencyPrefetch); + + for (k = 0; k < mode_lib->ms.num_active_planes; k++) + s->impacted_dst_y_pre[k] = 0; + + s->recalc_prefetch_schedule = 0; + s->recalc_prefetch_done = 0; + do { + mode_lib->ms.support.PrefetchSupported = true; + + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + s->line_times[k] = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + s->pixel_format[k] = display_cfg->plane_descriptors[k].pixel_format; + + s->lb_source_lines_l[k] = get_num_lb_source_lines(mode_lib->ip.max_line_buffer_lines, mode_lib->ip.line_buffer_size_bits, + mode_lib->ms.NoOfDPP[k], + display_cfg->plane_descriptors[k].composition.viewport.plane0.width, + display_cfg->plane_descriptors[k].composition.viewport.plane0.height, + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio, + display_cfg->plane_descriptors[k].composition.rotation_angle); + + s->lb_source_lines_c[k] = get_num_lb_source_lines(mode_lib->ip.max_line_buffer_lines, mode_lib->ip.line_buffer_size_bits, + mode_lib->ms.NoOfDPP[k], + display_cfg->plane_descriptors[k].composition.viewport.plane1.width, + display_cfg->plane_descriptors[k].composition.viewport.plane1.height, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio, + display_cfg->plane_descriptors[k].composition.rotation_angle); + + struct dml2_core_internal_DmlPipe *myPipe = &s->myPipe; + + mode_lib->ms.TWait[k] = CalculateTWait( + display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns, + mode_lib->ms.UrgLatency, + mode_lib->ms.TripToMemory, + !dml_is_phantom_pipe(&display_cfg->plane_descriptors[k]) && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.drr_config.enabled ? + get_g6_temp_read_blackout_us(&mode_lib->soc, (unsigned int)(mode_lib->ms.uclk_freq_mhz * 1000), mode_lib->ms.state_idx) : 0.0); + + myPipe->Dppclk = mode_lib->ms.RequiredDPPCLK[k]; + myPipe->Dispclk = mode_lib->ms.RequiredDISPCLK; + myPipe->PixelClock = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + myPipe->DCFClkDeepSleep = mode_lib->ms.dcfclk_deepsleep; + myPipe->DPPPerSurface = mode_lib->ms.NoOfDPP[k]; + myPipe->ScalerEnabled = display_cfg->plane_descriptors[k].composition.scaler_info.enabled; + myPipe->VRatio = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; + myPipe->VRatioChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; + myPipe->VTaps = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps; + myPipe->VTapsChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps; + myPipe->RotationAngle = display_cfg->plane_descriptors[k].composition.rotation_angle; + myPipe->mirrored = display_cfg->plane_descriptors[k].composition.mirrored; + myPipe->BlockWidth256BytesY = mode_lib->ms.Read256BlockWidthY[k]; + myPipe->BlockHeight256BytesY = mode_lib->ms.Read256BlockHeightY[k]; + myPipe->BlockWidth256BytesC = mode_lib->ms.Read256BlockWidthC[k]; + myPipe->BlockHeight256BytesC = mode_lib->ms.Read256BlockHeightC[k]; + myPipe->InterlaceEnable = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced; + myPipe->NumberOfCursors = display_cfg->plane_descriptors[k].cursor.num_cursors; + myPipe->VBlank = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active; + myPipe->HTotal = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total; + myPipe->HActive = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active; + myPipe->DCCEnable = display_cfg->plane_descriptors[k].surface.dcc.enable; + myPipe->ODMMode = mode_lib->ms.ODMMode[k]; + myPipe->SourcePixelFormat = display_cfg->plane_descriptors[k].pixel_format; + myPipe->BytePerPixelY = mode_lib->ms.BytePerPixelY[k]; + myPipe->BytePerPixelC = mode_lib->ms.BytePerPixelC[k]; + myPipe->ProgressiveToInterlaceUnitInOPP = mode_lib->ip.ptoi_supported; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: Calling CalculatePrefetchSchedule for k=%u\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: MaximumVStartup = %u\n", __func__, s->MaximumVStartup[k]); +#endif + CalculatePrefetchSchedule_params->display_cfg = display_cfg; + CalculatePrefetchSchedule_params->HostVMInefficiencyFactor = s->HostVMInefficiencyFactorPrefetch; + CalculatePrefetchSchedule_params->myPipe = myPipe; + CalculatePrefetchSchedule_params->DSCDelay = mode_lib->ms.DSCDelay[k]; + CalculatePrefetchSchedule_params->DPPCLKDelaySubtotalPlusCNVCFormater = mode_lib->ip.dppclk_delay_subtotal + mode_lib->ip.dppclk_delay_cnvc_formatter; + CalculatePrefetchSchedule_params->DPPCLKDelaySCL = mode_lib->ip.dppclk_delay_scl; + CalculatePrefetchSchedule_params->DPPCLKDelaySCLLBOnly = mode_lib->ip.dppclk_delay_scl_lb_only; + CalculatePrefetchSchedule_params->DPPCLKDelayCNVCCursor = mode_lib->ip.dppclk_delay_cnvc_cursor; + CalculatePrefetchSchedule_params->DISPCLKDelaySubtotal = mode_lib->ip.dispclk_delay_subtotal; + CalculatePrefetchSchedule_params->DPP_RECOUT_WIDTH = (unsigned int)(mode_lib->ms.SwathWidthY[k] / display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio); + CalculatePrefetchSchedule_params->OutputFormat = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format; + CalculatePrefetchSchedule_params->MaxInterDCNTileRepeaters = mode_lib->ip.max_inter_dcn_tile_repeaters; + CalculatePrefetchSchedule_params->VStartup = s->MaximumVStartup[k]; + CalculatePrefetchSchedule_params->HostVMMinPageSize = mode_lib->soc.hostvm_min_page_size_kbytes; + CalculatePrefetchSchedule_params->DynamicMetadataEnable = display_cfg->plane_descriptors[k].dynamic_meta_data.enable; + CalculatePrefetchSchedule_params->DynamicMetadataVMEnabled = mode_lib->ip.dynamic_metadata_vm_enabled; + CalculatePrefetchSchedule_params->DynamicMetadataLinesBeforeActiveRequired = display_cfg->plane_descriptors[k].dynamic_meta_data.lines_before_active_required; + CalculatePrefetchSchedule_params->DynamicMetadataTransmittedBytes = display_cfg->plane_descriptors[k].dynamic_meta_data.transmitted_bytes; + CalculatePrefetchSchedule_params->UrgentLatency = mode_lib->ms.UrgLatency; + CalculatePrefetchSchedule_params->ExtraLatencyPrefetch = mode_lib->ms.ExtraLatencyPrefetch; + CalculatePrefetchSchedule_params->TCalc = mode_lib->ms.TimeCalc; + CalculatePrefetchSchedule_params->vm_bytes = mode_lib->ms.vm_bytes[k]; + CalculatePrefetchSchedule_params->PixelPTEBytesPerRow = mode_lib->ms.DPTEBytesPerRow[k]; + CalculatePrefetchSchedule_params->PrefetchSourceLinesY = mode_lib->ms.PrefetchLinesY[k]; + CalculatePrefetchSchedule_params->VInitPreFillY = mode_lib->ms.PrefillY[k]; + CalculatePrefetchSchedule_params->MaxNumSwathY = mode_lib->ms.MaxNumSwathY[k]; + CalculatePrefetchSchedule_params->PrefetchSourceLinesC = mode_lib->ms.PrefetchLinesC[k]; + CalculatePrefetchSchedule_params->VInitPreFillC = mode_lib->ms.PrefillC[k]; + CalculatePrefetchSchedule_params->MaxNumSwathC = mode_lib->ms.MaxNumSwathC[k]; + CalculatePrefetchSchedule_params->swath_width_luma_ub = mode_lib->ms.swath_width_luma_ub[k]; + CalculatePrefetchSchedule_params->swath_width_chroma_ub = mode_lib->ms.swath_width_chroma_ub[k]; + CalculatePrefetchSchedule_params->SwathHeightY = mode_lib->ms.SwathHeightY[k]; + CalculatePrefetchSchedule_params->SwathHeightC = mode_lib->ms.SwathHeightC[k]; + CalculatePrefetchSchedule_params->TWait = mode_lib->ms.TWait[k]; + CalculatePrefetchSchedule_params->Ttrip = mode_lib->ms.TripToMemory; + CalculatePrefetchSchedule_params->Turg = mode_lib->ms.UrgLatency; + CalculatePrefetchSchedule_params->setup_for_tdlut = display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut; + CalculatePrefetchSchedule_params->tdlut_pte_bytes_per_frame = s->tdlut_pte_bytes_per_frame[k]; + CalculatePrefetchSchedule_params->tdlut_bytes_per_frame = s->tdlut_bytes_per_frame[k]; + CalculatePrefetchSchedule_params->tdlut_opt_time = s->tdlut_opt_time[k]; + CalculatePrefetchSchedule_params->tdlut_drain_time = s->tdlut_drain_time[k]; + CalculatePrefetchSchedule_params->num_cursors = (display_cfg->plane_descriptors[k].cursor.cursor_width > 0); + CalculatePrefetchSchedule_params->cursor_bytes_per_chunk = s->cursor_bytes_per_chunk[k]; + CalculatePrefetchSchedule_params->cursor_bytes_per_line = s->cursor_bytes_per_line[k]; + CalculatePrefetchSchedule_params->dcc_enable = display_cfg->plane_descriptors[k].surface.dcc.enable; + CalculatePrefetchSchedule_params->mrq_present = mode_lib->ip.dcn_mrq_present; + CalculatePrefetchSchedule_params->meta_row_bytes = mode_lib->ms.meta_row_bytes[k]; + CalculatePrefetchSchedule_params->mall_prefetch_sdp_overhead_factor = mode_lib->ms.mall_prefetch_sdp_overhead_factor[k]; + CalculatePrefetchSchedule_params->impacted_dst_y_pre = s->impacted_dst_y_pre[k]; + CalculatePrefetchSchedule_params->vactive_sw_bw_l = mode_lib->ms.vactive_sw_bw_l[k]; + CalculatePrefetchSchedule_params->vactive_sw_bw_c = mode_lib->ms.vactive_sw_bw_c[k]; + + // output + CalculatePrefetchSchedule_params->DSTXAfterScaler = &s->DSTXAfterScaler[k]; + CalculatePrefetchSchedule_params->DSTYAfterScaler = &s->DSTYAfterScaler[k]; + CalculatePrefetchSchedule_params->dst_y_prefetch = &mode_lib->ms.dst_y_prefetch[k]; + CalculatePrefetchSchedule_params->dst_y_per_vm_vblank = &mode_lib->ms.LinesForVM[k]; + CalculatePrefetchSchedule_params->dst_y_per_row_vblank = &mode_lib->ms.LinesForDPTERow[k]; + CalculatePrefetchSchedule_params->VRatioPrefetchY = &mode_lib->ms.VRatioPreY[k]; + CalculatePrefetchSchedule_params->VRatioPrefetchC = &mode_lib->ms.VRatioPreC[k]; + CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWLuma = &mode_lib->ms.RequiredPrefetchPixelDataBWLuma[k]; // prefetch_sw_bw_l + CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWChroma = &mode_lib->ms.RequiredPrefetchPixelDataBWChroma[k]; // prefetch_sw_bw_c + CalculatePrefetchSchedule_params->RequiredPrefetchBWMax = &mode_lib->ms.RequiredPrefetchBWMax[k]; + CalculatePrefetchSchedule_params->NotEnoughTimeForDynamicMetadata = &mode_lib->ms.NoTimeForDynamicMetadata[k]; + CalculatePrefetchSchedule_params->Tno_bw = &mode_lib->ms.Tno_bw[k]; + CalculatePrefetchSchedule_params->Tno_bw_flip = &mode_lib->ms.Tno_bw_flip[k]; + CalculatePrefetchSchedule_params->prefetch_vmrow_bw = &mode_lib->ms.prefetch_vmrow_bw[k]; + CalculatePrefetchSchedule_params->Tdmdl_vm = &s->dummy_single[0]; + CalculatePrefetchSchedule_params->Tdmdl = &s->dummy_single[1]; + CalculatePrefetchSchedule_params->TSetup = &s->dummy_single[2]; + CalculatePrefetchSchedule_params->Tvm_trips = &s->Tvm_trips[k]; + CalculatePrefetchSchedule_params->Tr0_trips = &s->Tr0_trips[k]; + CalculatePrefetchSchedule_params->Tvm_trips_flip = &s->Tvm_trips_flip[k]; + CalculatePrefetchSchedule_params->Tr0_trips_flip = &s->Tr0_trips_flip[k]; + CalculatePrefetchSchedule_params->Tvm_trips_flip_rounded = &s->Tvm_trips_flip_rounded[k]; + CalculatePrefetchSchedule_params->Tr0_trips_flip_rounded = &s->Tr0_trips_flip_rounded[k]; + CalculatePrefetchSchedule_params->VUpdateOffsetPix = &s->dummy_integer[0]; + CalculatePrefetchSchedule_params->VUpdateWidthPix = &s->dummy_integer[1]; + CalculatePrefetchSchedule_params->VReadyOffsetPix = &s->dummy_integer[2]; + CalculatePrefetchSchedule_params->prefetch_cursor_bw = &mode_lib->ms.prefetch_cursor_bw[k]; + CalculatePrefetchSchedule_params->prefetch_sw_bytes = &s->prefetch_sw_bytes[k]; + CalculatePrefetchSchedule_params->Tpre_rounded = &s->Tpre_rounded[k]; + CalculatePrefetchSchedule_params->Tpre_oto = &s->Tpre_oto[k]; + CalculatePrefetchSchedule_params->prefetch_swath_time_us = &s->prefetch_swath_time_us[k]; + + mode_lib->ms.NoTimeForPrefetch[k] = CalculatePrefetchSchedule(&mode_lib->scratch, CalculatePrefetchSchedule_params); + + mode_lib->ms.support.PrefetchSupported &= !mode_lib->ms.NoTimeForPrefetch[k]; + DML_LOG_VERBOSE("DML::%s: k=%d, dst_y_per_vm_vblank = %f\n", __func__, k, *CalculatePrefetchSchedule_params->dst_y_per_vm_vblank); + DML_LOG_VERBOSE("DML::%s: k=%d, dst_y_per_row_vblank = %f\n", __func__, k, *CalculatePrefetchSchedule_params->dst_y_per_row_vblank); + } // for k num_planes + + CalculateDCFCLKDeepSleepTdlut( + display_cfg, + mode_lib->ms.num_active_planes, + mode_lib->ms.BytePerPixelY, + mode_lib->ms.BytePerPixelC, + mode_lib->ms.SwathWidthY, + mode_lib->ms.SwathWidthC, + mode_lib->ms.NoOfDPP, + mode_lib->ms.PSCL_FACTOR, + mode_lib->ms.PSCL_FACTOR_CHROMA, + mode_lib->ms.RequiredDPPCLK, + mode_lib->ms.vactive_sw_bw_l, + mode_lib->ms.vactive_sw_bw_c, + mode_lib->soc.return_bus_width_bytes, + mode_lib->ms.RequiredDISPCLK, + s->tdlut_bytes_to_deliver, + s->prefetch_swath_time_us, + + /* Output */ + &mode_lib->ms.dcfclk_deepsleep); + + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + if (mode_lib->ms.dst_y_prefetch[k] < 2.0 + || mode_lib->ms.LinesForVM[k] >= 32.0 + || mode_lib->ms.LinesForDPTERow[k] >= 16.0 + || mode_lib->ms.NoTimeForPrefetch[k] == true + || s->DSTYAfterScaler[k] > 8) { + mode_lib->ms.support.PrefetchSupported = false; + DML_LOG_VERBOSE("DML::%s: k=%d, dst_y_prefetch=%f (should not be < 2)\n", __func__, k, mode_lib->ms.dst_y_prefetch[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, LinesForVM=%f (should not be >= 32)\n", __func__, k, mode_lib->ms.LinesForVM[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, LinesForDPTERow=%f (should not be >= 16)\n", __func__, k, mode_lib->ms.LinesForDPTERow[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, DSTYAfterScaler=%d (should be <= 8)\n", __func__, k, s->DSTYAfterScaler[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, NoTimeForPrefetch=%d\n", __func__, k, mode_lib->ms.NoTimeForPrefetch[k]); + } + } + + mode_lib->ms.support.DynamicMetadataSupported = true; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.NoTimeForDynamicMetadata[k] == true) { + mode_lib->ms.support.DynamicMetadataSupported = false; + } + } + + mode_lib->ms.support.VRatioInPrefetchSupported = true; + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + if (mode_lib->ms.VRatioPreY[k] > __DML2_CALCS_MAX_VRATIO_PRE__ || + mode_lib->ms.VRatioPreC[k] > __DML2_CALCS_MAX_VRATIO_PRE__) { + mode_lib->ms.support.VRatioInPrefetchSupported = false; + DML_LOG_VERBOSE("DML::%s: k=%d VRatioPreY = %f (should be <= %f)\n", __func__, k, mode_lib->ms.VRatioPreY[k], __DML2_CALCS_MAX_VRATIO_PRE__); + DML_LOG_VERBOSE("DML::%s: k=%d VRatioPreC = %f (should be <= %f)\n", __func__, k, mode_lib->ms.VRatioPreC[k], __DML2_CALCS_MAX_VRATIO_PRE__); + DML_LOG_VERBOSE("DML::%s: VRatioInPrefetchSupported = %u\n", __func__, mode_lib->ms.support.VRatioInPrefetchSupported); + } + } + + mode_lib->ms.support.PrefetchSupported &= mode_lib->ms.support.VRatioInPrefetchSupported; + + // By default, do not recalc prefetch schedule + s->recalc_prefetch_schedule = 0; + + // Only do urg vs prefetch bandwidth check, flip schedule check, power saving feature support check IF the Prefetch Schedule Check is ok + if (mode_lib->ms.support.PrefetchSupported) { + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + // Calculate Urgent burst factor for prefetch +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%d, Calling CalculateUrgentBurstFactor (for prefetch)\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: k=%d, VRatioPreY=%f\n", __func__, k, mode_lib->ms.VRatioPreY[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, VRatioPreC=%f\n", __func__, k, mode_lib->ms.VRatioPreC[k]); +#endif + CalculateUrgentBurstFactor( + &display_cfg->plane_descriptors[k], + mode_lib->ms.swath_width_luma_ub[k], + mode_lib->ms.swath_width_chroma_ub[k], + mode_lib->ms.SwathHeightY[k], + mode_lib->ms.SwathHeightC[k], + s->line_times[k], + mode_lib->ms.UrgLatency, + mode_lib->ms.VRatioPreY[k], + mode_lib->ms.VRatioPreC[k], + mode_lib->ms.BytePerPixelInDETY[k], + mode_lib->ms.BytePerPixelInDETC[k], + mode_lib->ms.DETBufferSizeY[k], + mode_lib->ms.DETBufferSizeC[k], + /* Output */ + &mode_lib->ms.UrgentBurstFactorLumaPre[k], + &mode_lib->ms.UrgentBurstFactorChromaPre[k], + &mode_lib->ms.NotEnoughUrgentLatencyHidingPre[k]); + } + + // Calculate urgent bandwidth required, both urg and non urg peak bandwidth + // assume flip bw is 0 at this point + for (k = 0; k < mode_lib->ms.num_active_planes; k++) + mode_lib->ms.final_flip_bw[k] = 0; + + calculate_peak_bandwidth_params->urg_vactive_bandwidth_required = mode_lib->ms.support.urg_vactive_bandwidth_required; + calculate_peak_bandwidth_params->urg_bandwidth_required = mode_lib->ms.support.urg_bandwidth_required; + calculate_peak_bandwidth_params->urg_bandwidth_required_qual = mode_lib->ms.support.urg_bandwidth_required_qual; + calculate_peak_bandwidth_params->non_urg_bandwidth_required = mode_lib->ms.support.non_urg_bandwidth_required; + calculate_peak_bandwidth_params->surface_avg_vactive_required_bw = mode_lib->ms.surface_avg_vactive_required_bw; + calculate_peak_bandwidth_params->surface_peak_required_bw = mode_lib->ms.surface_peak_required_bw; + + calculate_peak_bandwidth_params->display_cfg = display_cfg; + calculate_peak_bandwidth_params->inc_flip_bw = 0; + calculate_peak_bandwidth_params->num_active_planes = mode_lib->ms.num_active_planes; + calculate_peak_bandwidth_params->num_of_dpp = mode_lib->ms.NoOfDPP; + calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0; + calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1; + calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p0; + calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p1; + calculate_peak_bandwidth_params->mall_prefetch_sdp_overhead_factor = mode_lib->ms.mall_prefetch_sdp_overhead_factor; + calculate_peak_bandwidth_params->mall_prefetch_dram_overhead_factor = mode_lib->ms.mall_prefetch_dram_overhead_factor; + + calculate_peak_bandwidth_params->surface_read_bandwidth_l = mode_lib->ms.vactive_sw_bw_l; + calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->ms.vactive_sw_bw_c; + calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->ms.RequiredPrefetchPixelDataBWLuma; + calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->ms.RequiredPrefetchPixelDataBWChroma; + calculate_peak_bandwidth_params->prefetch_bandwidth_max = mode_lib->ms.RequiredPrefetchBWMax; + calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->ms.excess_vactive_fill_bw_l; + calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->ms.excess_vactive_fill_bw_c; + calculate_peak_bandwidth_params->cursor_bw = mode_lib->ms.cursor_bw; + calculate_peak_bandwidth_params->dpte_row_bw = mode_lib->ms.dpte_row_bw; + calculate_peak_bandwidth_params->meta_row_bw = mode_lib->ms.meta_row_bw; + calculate_peak_bandwidth_params->prefetch_cursor_bw = mode_lib->ms.prefetch_cursor_bw; + calculate_peak_bandwidth_params->prefetch_vmrow_bw = mode_lib->ms.prefetch_vmrow_bw; + calculate_peak_bandwidth_params->flip_bw = mode_lib->ms.final_flip_bw; + calculate_peak_bandwidth_params->urgent_burst_factor_l = mode_lib->ms.UrgentBurstFactorLuma; + calculate_peak_bandwidth_params->urgent_burst_factor_c = mode_lib->ms.UrgentBurstFactorChroma; + calculate_peak_bandwidth_params->urgent_burst_factor_cursor = mode_lib->ms.UrgentBurstFactorCursor; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_l = mode_lib->ms.UrgentBurstFactorLumaPre; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_c = mode_lib->ms.UrgentBurstFactorChromaPre; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_cursor = mode_lib->ms.UrgentBurstFactorCursorPre; + + calculate_peak_bandwidth_required( + &mode_lib->scratch, + calculate_peak_bandwidth_params); + + // Check urg peak bandwidth against available urg bw + // check at SDP and DRAM, for all soc states (SVP prefetch an Sys Active) + check_urgent_bandwidth_support( + &s->dummy_single[0], // double* frac_urg_bandwidth + &s->dummy_single[1], // double* frac_urg_bandwidth_mall + &mode_lib->ms.support.UrgVactiveBandwidthSupport, + &mode_lib->ms.support.PrefetchBandwidthSupported, + + mode_lib->soc.mall_allocated_for_dcn_mbytes, + mode_lib->ms.support.non_urg_bandwidth_required, + mode_lib->ms.support.urg_vactive_bandwidth_required, + mode_lib->ms.support.urg_bandwidth_required, + mode_lib->ms.support.urg_bandwidth_available); + + mode_lib->ms.support.PrefetchSupported &= mode_lib->ms.support.PrefetchBandwidthSupported; + DML_LOG_VERBOSE("DML::%s: PrefetchBandwidthSupported=%0d\n", __func__, mode_lib->ms.support.PrefetchBandwidthSupported); + + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + if (mode_lib->ms.NotEnoughUrgentLatencyHidingPre[k]) { + mode_lib->ms.support.PrefetchSupported = false; + DML_LOG_VERBOSE("DML::%s: k=%d, NotEnoughUrgentLatencyHidingPre=%d\n", __func__, k, mode_lib->ms.NotEnoughUrgentLatencyHidingPre[k]); + } + } + +#ifdef DML_GLOBAL_PREFETCH_CHECK + if (mode_lib->ms.support.PrefetchSupported && mode_lib->ms.num_active_planes > 1 && s->recalc_prefetch_done == 0) { + CheckGlobalPrefetchAdmissibility_params->num_active_planes = mode_lib->ms.num_active_planes; + CheckGlobalPrefetchAdmissibility_params->pixel_format = s->pixel_format; + CheckGlobalPrefetchAdmissibility_params->chunk_bytes_l = mode_lib->ip.pixel_chunk_size_kbytes * 1024; + CheckGlobalPrefetchAdmissibility_params->chunk_bytes_c = mode_lib->ip.pixel_chunk_size_kbytes * 1024; + CheckGlobalPrefetchAdmissibility_params->lb_source_lines_l = s->lb_source_lines_l; + CheckGlobalPrefetchAdmissibility_params->lb_source_lines_c = s->lb_source_lines_c; + CheckGlobalPrefetchAdmissibility_params->swath_height_l = mode_lib->ms.SwathHeightY; + CheckGlobalPrefetchAdmissibility_params->swath_height_c = mode_lib->ms.SwathHeightC; + CheckGlobalPrefetchAdmissibility_params->rob_buffer_size_kbytes = mode_lib->ip.rob_buffer_size_kbytes; + CheckGlobalPrefetchAdmissibility_params->compressed_buffer_size_kbytes = mode_lib->ms.CompressedBufferSizeInkByte; + CheckGlobalPrefetchAdmissibility_params->detile_buffer_size_bytes_l = mode_lib->ms.DETBufferSizeY; + CheckGlobalPrefetchAdmissibility_params->detile_buffer_size_bytes_c = mode_lib->ms.DETBufferSizeC; + CheckGlobalPrefetchAdmissibility_params->full_swath_bytes_l = s->full_swath_bytes_l; + CheckGlobalPrefetchAdmissibility_params->full_swath_bytes_c = s->full_swath_bytes_c; + CheckGlobalPrefetchAdmissibility_params->prefetch_sw_bytes = s->prefetch_sw_bytes; + CheckGlobalPrefetchAdmissibility_params->Tpre_rounded = s->Tpre_rounded; + CheckGlobalPrefetchAdmissibility_params->Tpre_oto = s->Tpre_oto; + CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps = mode_lib->ms.support.urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]; + CheckGlobalPrefetchAdmissibility_params->line_time = s->line_times; + CheckGlobalPrefetchAdmissibility_params->dst_y_prefetch = mode_lib->ms.dst_y_prefetch; + if (CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps < 10 * 1024) + CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps = 10 * 1024; + + CheckGlobalPrefetchAdmissibility_params->estimated_dcfclk_mhz = (CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps / (double) mode_lib->soc.return_bus_width_bytes) / + ((double)mode_lib->soc.qos_parameters.derate_table.system_active_urgent.dcfclk_derate_percent / 100.0); + + // if recalc_prefetch_schedule is set, recalculate the prefetch schedule with the new impacted_Tpre, prefetch should be possible + CheckGlobalPrefetchAdmissibility_params->recalc_prefetch_schedule = &s->recalc_prefetch_schedule; + CheckGlobalPrefetchAdmissibility_params->impacted_dst_y_pre = s->impacted_dst_y_pre; + mode_lib->ms.support.PrefetchSupported = CheckGlobalPrefetchAdmissibility(&mode_lib->scratch, CheckGlobalPrefetchAdmissibility_params); + s->recalc_prefetch_done = 1; + s->recalc_prefetch_schedule = 1; + } +#endif + } // prefetch schedule ok, do urg bw and flip schedule + } while (s->recalc_prefetch_schedule); + + // Flip Schedule + // Both prefetch schedule and BW okay + if (mode_lib->ms.support.PrefetchSupported == true) { + mode_lib->ms.BandwidthAvailableForImmediateFlip = + get_bandwidth_available_for_immediate_flip( + dml2_core_internal_soc_state_sys_active, + mode_lib->ms.support.urg_bandwidth_required_qual, // no flip + mode_lib->ms.support.urg_bandwidth_available); + + mode_lib->ms.TotImmediateFlipBytes = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + if (display_cfg->plane_descriptors[k].immediate_flip) { + s->per_pipe_flip_bytes[k] = get_pipe_flip_bytes( + s->HostVMInefficiencyFactor, + mode_lib->ms.vm_bytes[k], + mode_lib->ms.DPTEBytesPerRow[k], + mode_lib->ms.meta_row_bytes[k]); + } else { + s->per_pipe_flip_bytes[k] = 0; + } + mode_lib->ms.TotImmediateFlipBytes += s->per_pipe_flip_bytes[k] * mode_lib->ms.NoOfDPP[k]; + + } + + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + CalculateFlipSchedule( + &mode_lib->scratch, + display_cfg->plane_descriptors[k].immediate_flip, + 1, // use_lb_flip_bw + s->HostVMInefficiencyFactor, + s->Tvm_trips_flip[k], + s->Tr0_trips_flip[k], + s->Tvm_trips_flip_rounded[k], + s->Tr0_trips_flip_rounded[k], + display_cfg->gpuvm_enable, + mode_lib->ms.vm_bytes[k], + mode_lib->ms.DPTEBytesPerRow[k], + mode_lib->ms.BandwidthAvailableForImmediateFlip, + mode_lib->ms.TotImmediateFlipBytes, + display_cfg->plane_descriptors[k].pixel_format, + (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)), + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, + mode_lib->ms.Tno_bw_flip[k], + mode_lib->ms.dpte_row_height[k], + mode_lib->ms.dpte_row_height_chroma[k], + mode_lib->ms.use_one_row_for_frame_flip[k], + mode_lib->ip.max_flip_time_us, + mode_lib->ip.max_flip_time_lines, + s->per_pipe_flip_bytes[k], + mode_lib->ms.meta_row_bytes[k], + s->meta_row_height_luma[k], + s->meta_row_height_chroma[k], + mode_lib->ip.dcn_mrq_present && display_cfg->plane_descriptors[k].surface.dcc.enable, + + /* Output */ + &mode_lib->ms.dst_y_per_vm_flip[k], + &mode_lib->ms.dst_y_per_row_flip[k], + &mode_lib->ms.final_flip_bw[k], + &mode_lib->ms.ImmediateFlipSupportedForPipe[k]); + } + + calculate_peak_bandwidth_params->urg_vactive_bandwidth_required = s->dummy_bw; + calculate_peak_bandwidth_params->urg_bandwidth_required = mode_lib->ms.support.urg_bandwidth_required_flip; + calculate_peak_bandwidth_params->urg_bandwidth_required_qual = s->dummy_bw; + calculate_peak_bandwidth_params->non_urg_bandwidth_required = mode_lib->ms.support.non_urg_bandwidth_required_flip; + calculate_peak_bandwidth_params->surface_avg_vactive_required_bw = s->surface_dummy_bw; + calculate_peak_bandwidth_params->surface_peak_required_bw = mode_lib->ms.surface_peak_required_bw; + + calculate_peak_bandwidth_params->display_cfg = display_cfg; + calculate_peak_bandwidth_params->inc_flip_bw = 1; + calculate_peak_bandwidth_params->num_active_planes = mode_lib->ms.num_active_planes; + calculate_peak_bandwidth_params->num_of_dpp = mode_lib->ms.NoOfDPP; + calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0; + calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1; + calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p0; + calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p1; + calculate_peak_bandwidth_params->mall_prefetch_sdp_overhead_factor = mode_lib->ms.mall_prefetch_sdp_overhead_factor; + calculate_peak_bandwidth_params->mall_prefetch_dram_overhead_factor = mode_lib->ms.mall_prefetch_dram_overhead_factor; + + calculate_peak_bandwidth_params->surface_read_bandwidth_l = mode_lib->ms.vactive_sw_bw_l; + calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->ms.vactive_sw_bw_c; + calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->ms.RequiredPrefetchPixelDataBWLuma; + calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->ms.RequiredPrefetchPixelDataBWChroma; + calculate_peak_bandwidth_params->prefetch_bandwidth_max = mode_lib->ms.RequiredPrefetchBWMax; + calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->ms.excess_vactive_fill_bw_l; + calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->ms.excess_vactive_fill_bw_c; + calculate_peak_bandwidth_params->cursor_bw = mode_lib->ms.cursor_bw; + calculate_peak_bandwidth_params->dpte_row_bw = mode_lib->ms.dpte_row_bw; + calculate_peak_bandwidth_params->meta_row_bw = mode_lib->ms.meta_row_bw; + calculate_peak_bandwidth_params->prefetch_cursor_bw = mode_lib->ms.prefetch_cursor_bw; + calculate_peak_bandwidth_params->prefetch_vmrow_bw = mode_lib->ms.prefetch_vmrow_bw; + calculate_peak_bandwidth_params->flip_bw = mode_lib->ms.final_flip_bw; + calculate_peak_bandwidth_params->urgent_burst_factor_l = mode_lib->ms.UrgentBurstFactorLuma; + calculate_peak_bandwidth_params->urgent_burst_factor_c = mode_lib->ms.UrgentBurstFactorChroma; + calculate_peak_bandwidth_params->urgent_burst_factor_cursor = mode_lib->ms.UrgentBurstFactorCursor; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_l = mode_lib->ms.UrgentBurstFactorLumaPre; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_c = mode_lib->ms.UrgentBurstFactorChromaPre; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_cursor = mode_lib->ms.UrgentBurstFactorCursorPre; + + calculate_peak_bandwidth_required( + &mode_lib->scratch, + calculate_peak_bandwidth_params); + + calculate_immediate_flip_bandwidth_support( + &s->dummy_single[0], // double* frac_urg_bandwidth_flip + &mode_lib->ms.support.ImmediateFlipSupport, + + dml2_core_internal_soc_state_sys_active, + mode_lib->ms.support.urg_bandwidth_required_flip, + mode_lib->ms.support.non_urg_bandwidth_required_flip, + mode_lib->ms.support.urg_bandwidth_available); + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (display_cfg->plane_descriptors[k].immediate_flip == true && mode_lib->ms.ImmediateFlipSupportedForPipe[k] == false) + mode_lib->ms.support.ImmediateFlipSupport = false; + } + + } else { // if prefetch not support, assume iflip is not supported too + mode_lib->ms.support.ImmediateFlipSupport = false; + } + + s->mSOCParameters.UrgentLatency = mode_lib->ms.UrgLatency; + s->mSOCParameters.ExtraLatency = mode_lib->ms.ExtraLatency; + s->mSOCParameters.ExtraLatency_sr = mode_lib->ms.ExtraLatency_sr; + s->mSOCParameters.WritebackLatency = mode_lib->soc.qos_parameters.writeback.base_latency_us; + s->mSOCParameters.DRAMClockChangeLatency = mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us; + s->mSOCParameters.FCLKChangeLatency = mode_lib->soc.power_management_parameters.fclk_change_blackout_us; + s->mSOCParameters.SRExitTime = mode_lib->soc.power_management_parameters.stutter_exit_latency_us; + s->mSOCParameters.SREnterPlusExitTime = mode_lib->soc.power_management_parameters.stutter_enter_plus_exit_latency_us; + s->mSOCParameters.SRExitZ8Time = mode_lib->soc.power_management_parameters.z8_stutter_exit_latency_us; + s->mSOCParameters.SREnterPlusExitZ8Time = mode_lib->soc.power_management_parameters.z8_stutter_enter_plus_exit_latency_us; + s->mSOCParameters.USRRetrainingLatency = 0; + s->mSOCParameters.SMNLatency = 0; + s->mSOCParameters.g6_temp_read_blackout_us = get_g6_temp_read_blackout_us(&mode_lib->soc, (unsigned int)(mode_lib->ms.uclk_freq_mhz * 1000), mode_lib->ms.state_idx); + s->mSOCParameters.max_urgent_latency_us = get_max_urgent_latency_us(&mode_lib->soc.qos_parameters.qos_params.dcn4x, mode_lib->ms.uclk_freq_mhz, mode_lib->ms.FabricClock, mode_lib->ms.state_idx); + s->mSOCParameters.df_response_time_us = mode_lib->soc.qos_parameters.qos_params.dcn4x.df_qos_response_time_fclk_cycles / mode_lib->ms.FabricClock; + s->mSOCParameters.qos_type = mode_lib->soc.qos_parameters.qos_type; + + CalculateWatermarks_params->display_cfg = display_cfg; + CalculateWatermarks_params->USRRetrainingRequired = false; + CalculateWatermarks_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; + CalculateWatermarks_params->MaxLineBufferLines = mode_lib->ip.max_line_buffer_lines; + CalculateWatermarks_params->LineBufferSize = mode_lib->ip.line_buffer_size_bits; + CalculateWatermarks_params->WritebackInterfaceBufferSize = mode_lib->ip.writeback_interface_buffer_size_kbytes; + CalculateWatermarks_params->DCFCLK = mode_lib->ms.DCFCLK; + CalculateWatermarks_params->SynchronizeTimings = display_cfg->overrides.synchronize_timings; + CalculateWatermarks_params->SynchronizeDRRDisplaysForUCLKPStateChange = display_cfg->overrides.synchronize_ddr_displays_for_uclk_pstate_change; + CalculateWatermarks_params->dpte_group_bytes = mode_lib->ms.dpte_group_bytes; + CalculateWatermarks_params->mmSOCParameters = s->mSOCParameters; + CalculateWatermarks_params->WritebackChunkSize = mode_lib->ip.writeback_chunk_size_kbytes; + CalculateWatermarks_params->SOCCLK = mode_lib->ms.SOCCLK; + CalculateWatermarks_params->DCFClkDeepSleep = mode_lib->ms.dcfclk_deepsleep; + CalculateWatermarks_params->DETBufferSizeY = mode_lib->ms.DETBufferSizeY; + CalculateWatermarks_params->DETBufferSizeC = mode_lib->ms.DETBufferSizeC; + CalculateWatermarks_params->SwathHeightY = mode_lib->ms.SwathHeightY; + CalculateWatermarks_params->SwathHeightC = mode_lib->ms.SwathHeightC; + CalculateWatermarks_params->SwathWidthY = mode_lib->ms.SwathWidthY; + CalculateWatermarks_params->SwathWidthC = mode_lib->ms.SwathWidthC; + CalculateWatermarks_params->DPPPerSurface = mode_lib->ms.NoOfDPP; + CalculateWatermarks_params->BytePerPixelDETY = mode_lib->ms.BytePerPixelInDETY; + CalculateWatermarks_params->BytePerPixelDETC = mode_lib->ms.BytePerPixelInDETC; + CalculateWatermarks_params->DSTXAfterScaler = s->DSTXAfterScaler; + CalculateWatermarks_params->DSTYAfterScaler = s->DSTYAfterScaler; + CalculateWatermarks_params->UnboundedRequestEnabled = mode_lib->ms.UnboundedRequestEnabled; + CalculateWatermarks_params->CompressedBufferSizeInkByte = mode_lib->ms.CompressedBufferSizeInkByte; + CalculateWatermarks_params->meta_row_height_l = s->meta_row_height_luma; + CalculateWatermarks_params->meta_row_height_c = s->meta_row_height_chroma; + + // Output + CalculateWatermarks_params->Watermark = &mode_lib->ms.support.watermarks; // Watermarks *Watermark + CalculateWatermarks_params->DRAMClockChangeSupport = mode_lib->ms.support.DRAMClockChangeSupport; + CalculateWatermarks_params->global_dram_clock_change_supported = &mode_lib->ms.support.global_dram_clock_change_supported; + CalculateWatermarks_params->MaxActiveDRAMClockChangeLatencySupported = &s->dummy_single_array[0]; // double *MaxActiveDRAMClockChangeLatencySupported[] + CalculateWatermarks_params->SubViewportLinesNeededInMALL = mode_lib->ms.SubViewportLinesNeededInMALL; // unsigned int SubViewportLinesNeededInMALL[] + CalculateWatermarks_params->FCLKChangeSupport = mode_lib->ms.support.FCLKChangeSupport; + CalculateWatermarks_params->global_fclk_change_supported = &mode_lib->ms.support.global_fclk_change_supported; + CalculateWatermarks_params->MaxActiveFCLKChangeLatencySupported = &s->dummy_single[0]; // double *MaxActiveFCLKChangeLatencySupported + CalculateWatermarks_params->USRRetrainingSupport = &mode_lib->ms.support.USRRetrainingSupport; + CalculateWatermarks_params->g6_temp_read_support = &mode_lib->ms.support.g6_temp_read_support; + CalculateWatermarks_params->VActiveLatencyHidingMargin = mode_lib->ms.VActiveLatencyHidingMargin; + CalculateWatermarks_params->VActiveLatencyHidingUs = mode_lib->ms.VActiveLatencyHidingUs; + + CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport(&mode_lib->scratch, CalculateWatermarks_params); + + calculate_pstate_keepout_dst_lines(display_cfg, &mode_lib->ms.support.watermarks, s->dummy_integer_array[0]); + DML_LOG_VERBOSE("DML::%s: Done prefetch calculation\n", __func__); + +} + + static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out_params) { struct dml2_core_internal_display_mode_lib *mode_lib = in_out_params->mode_lib; const struct dml2_display_cfg *display_cfg = in_out_params->in_display_cfg; const struct dml2_mcg_min_clock_table *min_clk_table = in_out_params->min_clk_table; -#if defined(__DML_VBA_DEBUG__) - double old_ReadBandwidthLuma; - double old_ReadBandwidthChroma; -#endif double outstanding_latency_us = 0; - double min_return_bw_for_latency; struct dml2_core_calcs_mode_support_locals *s = &mode_lib->scratch.dml_core_mode_support_locals; - struct dml2_core_calcs_CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params *CalculateWatermarks_params = &mode_lib->scratch.CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params; struct dml2_core_calcs_CalculateVMRowAndSwath_params *CalculateVMRowAndSwath_params = &mode_lib->scratch.CalculateVMRowAndSwath_params; struct dml2_core_calcs_CalculateSwathAndDETConfiguration_params *CalculateSwathAndDETConfiguration_params = &mode_lib->scratch.CalculateSwathAndDETConfiguration_params; - struct dml2_core_calcs_CalculatePrefetchSchedule_params *CalculatePrefetchSchedule_params = &mode_lib->scratch.CalculatePrefetchSchedule_params; -#ifdef DML_GLOBAL_PREFETCH_CHECK - struct dml2_core_calcs_CheckGlobalPrefetchAdmissibility_params *CheckGlobalPrefetchAdmissibility_params = &mode_lib->scratch.CheckGlobalPrefetchAdmissibility_params; -#endif - struct dml2_core_calcs_calculate_tdlut_setting_params *calculate_tdlut_setting_params = &mode_lib->scratch.calculate_tdlut_setting_params; struct dml2_core_calcs_calculate_mcache_setting_params *calculate_mcache_setting_params = &mode_lib->scratch.calculate_mcache_setting_params; - struct dml2_core_calcs_calculate_peak_bandwidth_required_params *calculate_peak_bandwidth_params = &mode_lib->scratch.calculate_peak_bandwidth_params; struct dml2_core_calcs_calculate_bytes_to_fetch_required_to_hide_latency_params *calculate_bytes_to_fetch_required_to_hide_latency_params = &mode_lib->scratch.calculate_bytes_to_fetch_required_to_hide_latency_params; unsigned int k, m, n; @@ -7374,9 +7942,9 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.FabricClock = ((double)min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].min_fclk_khz / 1000); mode_lib->ms.MaxDCFCLK = (double)min_clk_table->max_clocks_khz.dcfclk / 1000; mode_lib->ms.MaxFabricClock = (double)min_clk_table->max_clocks_khz.fclk / 1000; - mode_lib->ms.max_dispclk_freq_mhz = (double)min_clk_table->max_clocks_khz.dispclk / 1000; + mode_lib->ms.max_dispclk_freq_mhz = (double)min_clk_table->max_ss_clocks_khz.dispclk / 1000; mode_lib->ms.max_dscclk_freq_mhz = (double)min_clk_table->max_clocks_khz.dscclk / 1000; - mode_lib->ms.max_dppclk_freq_mhz = (double)min_clk_table->max_clocks_khz.dppclk / 1000; + mode_lib->ms.max_dppclk_freq_mhz = (double)min_clk_table->max_ss_clocks_khz.dppclk / 1000; mode_lib->ms.uclk_freq_mhz = dram_bw_kbps_to_uclk_mhz(min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].pre_derate_dram_bw_kbps, &mode_lib->soc.clk_table.dram_config); mode_lib->ms.dram_bw_mbps = ((double)min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].pre_derate_dram_bw_kbps / 1000); mode_lib->ms.max_dram_bw_mbps = ((double)min_clk_table->dram_bw_table.entries[min_clk_table->dram_bw_table.num_entries - 1].pre_derate_dram_bw_kbps / 1000); @@ -7384,25 +7952,25 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.active_min_uclk_dpm_index = get_active_min_uclk_dpm_index((unsigned int) (mode_lib->ms.uclk_freq_mhz * 1000.0), &mode_lib->soc.clk_table); #if defined(__DML_VBA_DEBUG__) - dml2_printf("DML::%s: --- START --- \n", __func__); - dml2_printf("DML::%s: num_active_planes = %u\n", __func__, mode_lib->ms.num_active_planes); - dml2_printf("DML::%s: min_clk_index = %0d\n", __func__, in_out_params->min_clk_index); - dml2_printf("DML::%s: qos_param_index = %0d\n", __func__, mode_lib->ms.qos_param_index); - dml2_printf("DML::%s: SOCCLK = %f\n", __func__, mode_lib->ms.SOCCLK); - dml2_printf("DML::%s: dram_bw_mbps = %f\n", __func__, mode_lib->ms.dram_bw_mbps); - dml2_printf("DML::%s: uclk_freq_mhz = %f\n", __func__, mode_lib->ms.uclk_freq_mhz); - dml2_printf("DML::%s: DCFCLK = %f\n", __func__, mode_lib->ms.DCFCLK); - dml2_printf("DML::%s: FabricClock = %f\n", __func__, mode_lib->ms.FabricClock); - dml2_printf("DML::%s: MaxDCFCLK = %f\n", __func__, mode_lib->ms.MaxDCFCLK); - dml2_printf("DML::%s: max_dispclk_freq_mhz = %f\n", __func__, mode_lib->ms.max_dispclk_freq_mhz); - dml2_printf("DML::%s: max_dscclk_freq_mhz = %f\n", __func__, mode_lib->ms.max_dscclk_freq_mhz); - dml2_printf("DML::%s: max_dppclk_freq_mhz = %f\n", __func__, mode_lib->ms.max_dppclk_freq_mhz); - dml2_printf("DML::%s: MaxFabricClock = %f\n", __func__, mode_lib->ms.MaxFabricClock); - dml2_printf("DML::%s: ip.compressed_buffer_segment_size_in_kbytes = %u\n", __func__, mode_lib->ip.compressed_buffer_segment_size_in_kbytes); - dml2_printf("DML::%s: ip.dcn_mrq_present = %u\n", __func__, mode_lib->ip.dcn_mrq_present); + DML_LOG_VERBOSE("DML::%s: --- START --- \n", __func__); + DML_LOG_VERBOSE("DML::%s: num_active_planes = %u\n", __func__, mode_lib->ms.num_active_planes); + DML_LOG_VERBOSE("DML::%s: min_clk_index = %0d\n", __func__, in_out_params->min_clk_index); + DML_LOG_VERBOSE("DML::%s: qos_param_index = %0d\n", __func__, mode_lib->ms.qos_param_index); + DML_LOG_VERBOSE("DML::%s: SOCCLK = %f\n", __func__, mode_lib->ms.SOCCLK); + DML_LOG_VERBOSE("DML::%s: dram_bw_mbps = %f\n", __func__, mode_lib->ms.dram_bw_mbps); + DML_LOG_VERBOSE("DML::%s: uclk_freq_mhz = %f\n", __func__, mode_lib->ms.uclk_freq_mhz); + DML_LOG_VERBOSE("DML::%s: DCFCLK = %f\n", __func__, mode_lib->ms.DCFCLK); + DML_LOG_VERBOSE("DML::%s: FabricClock = %f\n", __func__, mode_lib->ms.FabricClock); + DML_LOG_VERBOSE("DML::%s: MaxDCFCLK = %f\n", __func__, mode_lib->ms.MaxDCFCLK); + DML_LOG_VERBOSE("DML::%s: max_dispclk_freq_mhz = %f\n", __func__, mode_lib->ms.max_dispclk_freq_mhz); + DML_LOG_VERBOSE("DML::%s: max_dscclk_freq_mhz = %f\n", __func__, mode_lib->ms.max_dscclk_freq_mhz); + DML_LOG_VERBOSE("DML::%s: max_dppclk_freq_mhz = %f\n", __func__, mode_lib->ms.max_dppclk_freq_mhz); + DML_LOG_VERBOSE("DML::%s: MaxFabricClock = %f\n", __func__, mode_lib->ms.MaxFabricClock); + DML_LOG_VERBOSE("DML::%s: ip.compressed_buffer_segment_size_in_kbytes = %u\n", __func__, mode_lib->ip.compressed_buffer_segment_size_in_kbytes); + DML_LOG_VERBOSE("DML::%s: ip.dcn_mrq_present = %u\n", __func__, mode_lib->ip.dcn_mrq_present); for (k = 0; k < mode_lib->ms.num_active_planes; k++) - dml2_printf("DML::%s: plane_%d: reserved_vblank_time_ns = %u\n", __func__, k, display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns); + DML_LOG_VERBOSE("DML::%s: plane_%d: reserved_vblank_time_ns = %lu\n", __func__, k, display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns); #endif CalculateMaxDETAndMinCompressedBufferSize( @@ -7504,12 +8072,10 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out display_cfg->plane_descriptors[k].cursor.cursor_bpp / 8.0 / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)); #ifdef __DML_VBA_DEBUG__ - old_ReadBandwidthLuma = mode_lib->ms.SwathWidthYSingleDPP[k] * math_ceil2(mode_lib->ms.BytePerPixelInDETY[k], 1.0) / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; - old_ReadBandwidthChroma = mode_lib->ms.SwathWidthYSingleDPP[k] / 2 * math_ceil2(mode_lib->ms.BytePerPixelInDETC[k], 2.0) / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio / 2.0; - dml2_printf("DML::%s: k=%u, old_ReadBandwidthLuma = %f\n", __func__, k, old_ReadBandwidthLuma); - dml2_printf("DML::%s: k=%u, old_ReadBandwidthChroma = %f\n", __func__, k, old_ReadBandwidthChroma); - dml2_printf("DML::%s: k=%u, vactive_sw_bw_l = %f\n", __func__, k, mode_lib->ms.vactive_sw_bw_l[k]); - dml2_printf("DML::%s: k=%u, vactive_sw_bw_c = %f\n", __func__, k, mode_lib->ms.vactive_sw_bw_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, old_ReadBandwidthLuma = %f\n", __func__, k, mode_lib->ms.SwathWidthYSingleDPP[k] * math_ceil2(mode_lib->ms.BytePerPixelInDETY[k], 1.0) / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u, old_ReadBandwidthChroma = %f\n", __func__, k, mode_lib->ms.SwathWidthYSingleDPP[k] / 2 * math_ceil2(mode_lib->ms.BytePerPixelInDETC[k], 2.0) / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio / 2.0); + DML_LOG_VERBOSE("DML::%s: k=%u, vactive_sw_bw_l = %f\n", __func__, k, mode_lib->ms.vactive_sw_bw_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, vactive_sw_bw_c = %f\n", __func__, k, mode_lib->ms.vactive_sw_bw_c[k]); #endif } @@ -7629,13 +8195,13 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.MaximumSwathWidthLuma[k] = math_min2(s->MaximumSwathWidthSupportLuma, mode_lib->ms.MaximumSwathWidthInLineBufferLuma); mode_lib->ms.MaximumSwathWidthChroma[k] = math_min2(s->MaximumSwathWidthSupportChroma, mode_lib->ms.MaximumSwathWidthInLineBufferChroma); - dml2_printf("DML::%s: k=%u MaximumSwathWidthLuma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthLuma[k]); - dml2_printf("DML::%s: k=%u MaximumSwathWidthSupportLuma=%u\n", __func__, k, s->MaximumSwathWidthSupportLuma); - dml2_printf("DML::%s: k=%u MaximumSwathWidthInLineBufferLuma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthInLineBufferLuma); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthLuma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthSupportLuma=%u\n", __func__, k, s->MaximumSwathWidthSupportLuma); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthInLineBufferLuma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthInLineBufferLuma); - dml2_printf("DML::%s: k=%u MaximumSwathWidthChroma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthChroma[k]); - dml2_printf("DML::%s: k=%u MaximumSwathWidthSupportChroma=%u\n", __func__, k, s->MaximumSwathWidthSupportChroma); - dml2_printf("DML::%s: k=%u MaximumSwathWidthInLineBufferChroma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthInLineBufferChroma); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthChroma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthSupportChroma=%u\n", __func__, k, s->MaximumSwathWidthSupportChroma); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthInLineBufferChroma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthInLineBufferChroma); } /* Cursor Support Check */ @@ -7672,11 +8238,11 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.support.AlignedCPitch[k] > display_cfg->plane_descriptors[k].surface.plane1.pitch) { mode_lib->ms.support.PitchSupport = false; #if defined(__DML_VBA_DEBUG__) - dml2_printf("DML::%s: k=%u AlignedYPitch = %d\n", __func__, k, mode_lib->ms.support.AlignedYPitch[k]); - dml2_printf("DML::%s: k=%u PitchY = %d\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane0.pitch); - dml2_printf("DML::%s: k=%u AlignedCPitch = %d\n", __func__, k, mode_lib->ms.support.AlignedCPitch[k]); - dml2_printf("DML::%s: k=%u PitchC = %d\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane1.pitch); - dml2_printf("DML::%s: k=%u PitchSupport = %d\n", __func__, k, mode_lib->ms.support.PitchSupport); + DML_LOG_VERBOSE("DML::%s: k=%u AlignedYPitch = %d\n", __func__, k, mode_lib->ms.support.AlignedYPitch[k]); + DML_LOG_VERBOSE("DML::%s: k=%u PitchY = %ld\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane0.pitch); + DML_LOG_VERBOSE("DML::%s: k=%u AlignedCPitch = %d\n", __func__, k, mode_lib->ms.support.AlignedCPitch[k]); + DML_LOG_VERBOSE("DML::%s: k=%u PitchC = %ld\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane1.pitch); + DML_LOG_VERBOSE("DML::%s: k=%u PitchSupport = %d\n", __func__, k, mode_lib->ms.support.PitchSupport); #endif } @@ -7708,11 +8274,11 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out display_cfg->plane_descriptors[k].composition.viewport.plane0.height > display_cfg->plane_descriptors[k].surface.plane0.height) { mode_lib->ms.support.ViewportExceedsSurface = true; #if defined(__DML_VBA_DEBUG__) - dml2_printf("DML::%s: k=%u ViewportWidth = %d\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.width); - dml2_printf("DML::%s: k=%u SurfaceWidthY = %d\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane0.width); - dml2_printf("DML::%s: k=%u ViewportHeight = %d\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.height); - dml2_printf("DML::%s: k=%u SurfaceHeightY = %d\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane0.height); - dml2_printf("DML::%s: k=%u ViewportExceedsSurface = %d\n", __func__, k, mode_lib->ms.support.ViewportExceedsSurface); + DML_LOG_VERBOSE("DML::%s: k=%u ViewportWidth = %ld\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.width); + DML_LOG_VERBOSE("DML::%s: k=%u SurfaceWidthY = %ld\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane0.width); + DML_LOG_VERBOSE("DML::%s: k=%u ViewportHeight = %ld\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.height); + DML_LOG_VERBOSE("DML::%s: k=%u SurfaceHeightY = %ld\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane0.height); + DML_LOG_VERBOSE("DML::%s: k=%u ViewportExceedsSurface = %d\n", __func__, k, mode_lib->ms.support.ViewportExceedsSurface); #endif } if (dml_is_420(display_cfg->plane_descriptors[k].pixel_format) || display_cfg->plane_descriptors[k].pixel_format == dml2_rgbe_alpha) { @@ -7894,8 +8460,8 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.TotalNumberOfActiveDPP = mode_lib->ms.TotalNumberOfActiveDPP + s->NumberOfDPPDSC; } #if defined(__DML_VBA_DEBUG__) - dml2_printf("DML::%s: k=%d RequiresDSC = %d\n", __func__, k, mode_lib->ms.RequiresDSC[k]); - dml2_printf("DML::%s: k=%d ODMMode = %d\n", __func__, k, mode_lib->ms.ODMMode[k]); + DML_LOG_VERBOSE("DML::%s: k=%d RequiresDSC = %d\n", __func__, k, mode_lib->ms.RequiresDSC[k]); + DML_LOG_VERBOSE("DML::%s: k=%d ODMMode = %d\n", __func__, k, mode_lib->ms.ODMMode[k]); #endif // ensure the number dsc slices is integer multiple based on ODM mode @@ -7911,9 +8477,9 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.support.DSCSlicesODMModeSupported = ((mode_lib->ms.support.NumberOfDSCSlices[k] % 4) == 0); #if defined(__DML_VBA_DEBUG__) if (!mode_lib->ms.support.DSCSlicesODMModeSupported) { - dml2_printf("DML::%s: k=%d Invalid dsc num_slices and ODM mode setting\n", __func__, k); - dml2_printf("DML::%s: k=%d num_slices = %d\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.overrides.num_slices); - dml2_printf("DML::%s: k=%d ODMMode = %d\n", __func__, k, mode_lib->ms.ODMMode[k]); + DML_LOG_VERBOSE("DML::%s: k=%d Invalid dsc num_slices and ODM mode setting\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: k=%d num_slices = %d\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.overrides.num_slices); + DML_LOG_VERBOSE("DML::%s: k=%d ODMMode = %d\n", __func__, k, mode_lib->ms.ODMMode[k]); } #endif } else { @@ -7958,7 +8524,7 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.MPCCombine[k] = false; mode_lib->ms.NoOfDPP[k] = 1; if (!mode_lib->ms.SingleDPPViewportSizeSupportPerSurface[k]) { - dml2_printf("WARNING: DML::%s: MPCC is override to disable but viewport is too large to be supported with single pipe!\n", __func__); + DML_LOG_VERBOSE("WARNING: DML::%s: MPCC is override to disable but viewport is too large to be supported with single pipe!\n", __func__); } } else { if ((mode_lib->ms.MinDPPCLKUsingSingleDPP[k] > mode_lib->ms.max_dppclk_freq_mhz) || !mode_lib->ms.SingleDPPViewportSizeSupportPerSurface[k]) { @@ -7968,7 +8534,7 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out } } #if defined(__DML_VBA_DEBUG__) - dml2_printf("DML::%s: k=%d, NoOfDPP = %d\n", __func__, k, mode_lib->ms.NoOfDPP[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, NoOfDPP = %d\n", __func__, k, mode_lib->ms.NoOfDPP[k]); #endif } @@ -8138,7 +8704,7 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.audio_sample_rate, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.audio_sample_layout); - if (mode_lib->ms.RequiredDTBCLK[k] > ((double)min_clk_table->max_clocks_khz.dtbclk / 1000)) { + if (mode_lib->ms.RequiredDTBCLK[k] > ((double)min_clk_table->max_ss_clocks_khz.dtbclk / 1000)) { mode_lib->ms.support.DTBCLKRequiredMoreThanSupported = true; } } else { @@ -8167,7 +8733,7 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out s->DSCFormatFactor = 1; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, RequiresDSC = %u\n", __func__, k, mode_lib->ms.RequiresDSC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, RequiresDSC = %u\n", __func__, k, mode_lib->ms.RequiresDSC[k]); #endif if (mode_lib->ms.RequiresDSC[k] == true) { s->PixelClockBackEndFactor = 3.0; @@ -8185,10 +8751,10 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, PixelClockBackEnd = %f\n", __func__, k, s->PixelClockBackEnd[k]); - dml2_printf("DML::%s: k=%u, required_dscclk_freq_mhz = %f\n", __func__, k, mode_lib->ms.required_dscclk_freq_mhz[k]); - dml2_printf("DML::%s: k=%u, DSCFormatFactor = %u\n", __func__, k, s->DSCFormatFactor); - dml2_printf("DML::%s: k=%u, DSCCLKRequiredMoreThanSupported = %u\n", __func__, k, mode_lib->ms.support.DSCCLKRequiredMoreThanSupported); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelClockBackEnd = %f\n", __func__, k, s->PixelClockBackEnd[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, required_dscclk_freq_mhz = %f\n", __func__, k, mode_lib->ms.required_dscclk_freq_mhz[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, DSCFormatFactor = %u\n", __func__, k, s->DSCFormatFactor); + DML_LOG_VERBOSE("DML::%s: k=%u, DSCCLKRequiredMoreThanSupported = %u\n", __func__, k, mode_lib->ms.support.DSCCLKRequiredMoreThanSupported); #endif } } @@ -8423,13 +8989,13 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.support.DCCMetaBufferSizeNotExceeded = false; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, PTEBufferSizeNotExceeded = %u\n", __func__, k, mode_lib->ms.PTEBufferSizeNotExceeded[k]); - dml2_printf("DML::%s: k=%u, DCCMetaBufferSizeNotExceeded = %u\n", __func__, k, mode_lib->ms.DCCMetaBufferSizeNotExceeded[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PTEBufferSizeNotExceeded = %u\n", __func__, k, mode_lib->ms.PTEBufferSizeNotExceeded[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, DCCMetaBufferSizeNotExceeded = %u\n", __func__, k, mode_lib->ms.DCCMetaBufferSizeNotExceeded[k]); #endif } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: PTEBufferSizeNotExceeded = %u\n", __func__, mode_lib->ms.support.PTEBufferSizeNotExceeded); - dml2_printf("DML::%s: DCCMetaBufferSizeNotExceeded = %u\n", __func__, mode_lib->ms.support.DCCMetaBufferSizeNotExceeded); + DML_LOG_VERBOSE("DML::%s: PTEBufferSizeNotExceeded = %u\n", __func__, mode_lib->ms.support.PTEBufferSizeNotExceeded); + DML_LOG_VERBOSE("DML::%s: DCCMetaBufferSizeNotExceeded = %u\n", __func__, mode_lib->ms.support.DCCMetaBufferSizeNotExceeded); #endif /* VActive bytes to fetch for UCLK P-State */ @@ -8502,7 +9068,7 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { double line_time_us = (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - bool cursor_not_enough_urgent_latency_hiding = 0; + bool cursor_not_enough_urgent_latency_hiding = false; if (display_cfg->plane_descriptors[k].cursor.num_cursors > 0) { calculate_cursor_req_attributes( @@ -8531,9 +9097,9 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.UrgentBurstFactorCursorPre[k] = mode_lib->ms.UrgentBurstFactorCursor[k]; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%d, Calling CalculateUrgentBurstFactor\n", __func__, k); - dml2_printf("DML::%s: k=%d, VRatio=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); - dml2_printf("DML::%s: k=%d, VRatioChroma=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio); + DML_LOG_VERBOSE("DML::%s: k=%d, Calling CalculateUrgentBurstFactor\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: k=%d, VRatio=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); + DML_LOG_VERBOSE("DML::%s: k=%d, VRatioChroma=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio); #endif CalculateUrgentBurstFactor( @@ -8605,7 +9171,7 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, MaximumVStartup = %u\n", __func__, k, s->MaximumVStartup[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, MaximumVStartup = %u\n", __func__, k, s->MaximumVStartup[k]); #endif /* Immediate Flip and MALL parameters */ @@ -8654,16 +9220,15 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out (s->SubViewportMALLPStateMethod && s->FullFrameMALLPStateMethod) || s->SubViewportMALLRefreshGreaterThan120Hz; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: SubViewportMALLPStateMethod = %u\n", __func__, s->SubViewportMALLPStateMethod); - dml2_printf("DML::%s: PhantomPipeMALLPStateMethod = %u\n", __func__, s->PhantomPipeMALLPStateMethod); - dml2_printf("DML::%s: FullFrameMALLPStateMethod = %u\n", __func__, s->FullFrameMALLPStateMethod); - dml2_printf("DML::%s: SubViewportMALLRefreshGreaterThan120Hz = %u\n", __func__, s->SubViewportMALLRefreshGreaterThan120Hz); - dml2_printf("DML::%s: InvalidCombinationOfMALLUseForPState = %u\n", __func__, mode_lib->ms.support.InvalidCombinationOfMALLUseForPState); - dml2_printf("DML::%s: in_out_params->min_clk_index = %u\n", __func__, in_out_params->min_clk_index); - dml2_printf("DML::%s: mode_lib->ms.DCFCLK = %f\n", __func__, mode_lib->ms.DCFCLK); - dml2_printf("DML::%s: mode_lib->ms.FabricClock = %f\n", __func__, mode_lib->ms.FabricClock); - dml2_printf("DML::%s: mode_lib->ms.uclk_freq_mhz = %f\n", __func__, mode_lib->ms.uclk_freq_mhz); - dml2_printf("DML::%s: urgent latency tolarance = %f\n", __func__, ((mode_lib->ip.rob_buffer_size_kbytes - mode_lib->ip.pixel_chunk_size_kbytes) * 1024 / (mode_lib->ms.DCFCLK * mode_lib->soc.return_bus_width_bytes))); + DML_LOG_VERBOSE("DML::%s: SubViewportMALLPStateMethod = %u\n", __func__, s->SubViewportMALLPStateMethod); + DML_LOG_VERBOSE("DML::%s: PhantomPipeMALLPStateMethod = %u\n", __func__, s->PhantomPipeMALLPStateMethod); + DML_LOG_VERBOSE("DML::%s: FullFrameMALLPStateMethod = %u\n", __func__, s->FullFrameMALLPStateMethod); + DML_LOG_VERBOSE("DML::%s: SubViewportMALLRefreshGreaterThan120Hz = %u\n", __func__, s->SubViewportMALLRefreshGreaterThan120Hz); + DML_LOG_VERBOSE("DML::%s: InvalidCombinationOfMALLUseForPState = %u\n", __func__, mode_lib->ms.support.InvalidCombinationOfMALLUseForPState); + DML_LOG_VERBOSE("DML::%s: in_out_params->min_clk_index = %u\n", __func__, in_out_params->min_clk_index); + DML_LOG_VERBOSE("DML::%s: mode_lib->ms.DCFCLK = %f\n", __func__, mode_lib->ms.DCFCLK); + DML_LOG_VERBOSE("DML::%s: mode_lib->ms.FabricClock = %f\n", __func__, mode_lib->ms.FabricClock); + DML_LOG_VERBOSE("DML::%s: mode_lib->ms.uclk_freq_mhz = %f\n", __func__, mode_lib->ms.uclk_freq_mhz); #endif mode_lib->ms.support.OutstandingRequestsSupport = true; @@ -8703,10 +9268,10 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: avg_urgent_latency_us = %f\n", __func__, mode_lib->ms.support.avg_urgent_latency_us); - dml2_printf("DML::%s: avg_non_urgent_latency_us = %f\n", __func__, mode_lib->ms.support.avg_non_urgent_latency_us); - dml2_printf("DML::%s: k=%d, request_size_bytes_luma = %d\n", __func__, k, mode_lib->ms.support.request_size_bytes_luma[k]); - dml2_printf("DML::%s: k=%d, outstanding_latency_us = %f (luma)\n", __func__, k, outstanding_latency_us); + DML_LOG_VERBOSE("DML::%s: avg_urgent_latency_us = %f\n", __func__, mode_lib->ms.support.avg_urgent_latency_us); + DML_LOG_VERBOSE("DML::%s: avg_non_urgent_latency_us = %f\n", __func__, mode_lib->ms.support.avg_non_urgent_latency_us); + DML_LOG_VERBOSE("DML::%s: k=%d, request_size_bytes_luma = %d\n", __func__, k, mode_lib->ms.support.request_size_bytes_luma[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, outstanding_latency_us = %f (luma)\n", __func__, k, outstanding_latency_us); #endif } @@ -8722,8 +9287,8 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.support.OutstandingRequestsUrgencyAvoidance = false; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%d, request_size_bytes_chroma = %d\n", __func__, k, mode_lib->ms.support.request_size_bytes_chroma[k]); - dml2_printf("DML::%s: k=%d, outstanding_latency_us = %f (chroma)\n", __func__, k, outstanding_latency_us); + DML_LOG_VERBOSE("DML::%s: k=%d, request_size_bytes_chroma = %d\n", __func__, k, mode_lib->ms.support.request_size_bytes_chroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, outstanding_latency_us = %f (chroma)\n", __func__, k, outstanding_latency_us); #endif } } @@ -8869,7 +9434,7 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { if (mode_lib->ms.NotEnoughUrgentLatencyHiding[k]) { mode_lib->ms.support.EnoughUrgentLatencyHidingSupport = false; - dml2_printf("DML::%s: k=%u NotEnoughUrgentLatencyHiding set\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: k=%u NotEnoughUrgentLatencyHiding set\n", __func__, k); } } @@ -8878,639 +9443,13 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out if (!mode_lib->ms.support.avg_bandwidth_support_ok[m][n] && (m == dml2_core_internal_soc_state_sys_active || mode_lib->soc.mall_allocated_for_dcn_mbytes > 0)) { mode_lib->ms.support.AvgBandwidthSupport = false; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: avg_bandwidth_support_ok[%s][%s] not ok\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n)); + DML_LOG_VERBOSE("DML::%s: avg_bandwidth_support_ok[%s][%s] not ok\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n)); #endif } } } - /* Prefetch Check */ - { - mode_lib->ms.TimeCalc = 24 / mode_lib->ms.dcfclk_deepsleep; - - calculate_hostvm_inefficiency_factor( - &s->HostVMInefficiencyFactor, - &s->HostVMInefficiencyFactorPrefetch, - - display_cfg->gpuvm_enable, - display_cfg->hostvm_enable, - mode_lib->ip.remote_iommu_outstanding_translations, - mode_lib->soc.max_outstanding_reqs, - mode_lib->ms.support.urg_bandwidth_available_pixel_and_vm[dml2_core_internal_soc_state_sys_active], - mode_lib->ms.support.urg_bandwidth_available_vm_only[dml2_core_internal_soc_state_sys_active]); - - mode_lib->ms.Total3dlutActive = 0; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut) - mode_lib->ms.Total3dlutActive = mode_lib->ms.Total3dlutActive + 1; - - // Calculate tdlut schedule related terms - calculate_tdlut_setting_params->dispclk_mhz = mode_lib->ms.RequiredDISPCLK; - calculate_tdlut_setting_params->setup_for_tdlut = display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut; - calculate_tdlut_setting_params->tdlut_width_mode = display_cfg->plane_descriptors[k].tdlut.tdlut_width_mode; - calculate_tdlut_setting_params->tdlut_addressing_mode = display_cfg->plane_descriptors[k].tdlut.tdlut_addressing_mode; - calculate_tdlut_setting_params->cursor_buffer_size = mode_lib->ip.cursor_buffer_size; - calculate_tdlut_setting_params->gpuvm_enable = display_cfg->gpuvm_enable; - calculate_tdlut_setting_params->gpuvm_page_size_kbytes = display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes; - calculate_tdlut_setting_params->tdlut_mpc_width_flag = display_cfg->plane_descriptors[k].tdlut.tdlut_mpc_width_flag; - calculate_tdlut_setting_params->is_gfx11 = dml_get_gfx_version(display_cfg->plane_descriptors[k].surface.tiling); - - // output - calculate_tdlut_setting_params->tdlut_pte_bytes_per_frame = &s->tdlut_pte_bytes_per_frame[k]; - calculate_tdlut_setting_params->tdlut_bytes_per_frame = &s->tdlut_bytes_per_frame[k]; - calculate_tdlut_setting_params->tdlut_groups_per_2row_ub = &s->tdlut_groups_per_2row_ub[k]; - calculate_tdlut_setting_params->tdlut_opt_time = &s->tdlut_opt_time[k]; - calculate_tdlut_setting_params->tdlut_drain_time = &s->tdlut_drain_time[k]; - calculate_tdlut_setting_params->tdlut_bytes_to_deliver = &s->tdlut_bytes_to_deliver[k]; - calculate_tdlut_setting_params->tdlut_bytes_per_group = &s->tdlut_bytes_per_group[k]; - - calculate_tdlut_setting(&mode_lib->scratch, calculate_tdlut_setting_params); - } - - min_return_bw_for_latency = mode_lib->ms.support.urg_bandwidth_available_min_latency[dml2_core_internal_soc_state_sys_active]; - - if (mode_lib->soc.qos_parameters.qos_type == dml2_qos_param_type_dcn3) - s->ReorderingBytes = (unsigned int)(mode_lib->soc.clk_table.dram_config.channel_count * math_max3(mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_pixel_only_bytes, - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_pixel_and_vm_bytes, - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_vm_only_bytes)); - - CalculateExtraLatency( - display_cfg, - mode_lib->ip.rob_buffer_size_kbytes, - mode_lib->soc.qos_parameters.qos_params.dcn32x.loaded_round_trip_latency_fclk_cycles, - s->ReorderingBytes, - mode_lib->ms.DCFCLK, - mode_lib->ms.FabricClock, - mode_lib->ip.pixel_chunk_size_kbytes, - min_return_bw_for_latency, - mode_lib->ms.num_active_planes, - mode_lib->ms.NoOfDPP, - mode_lib->ms.dpte_group_bytes, - s->tdlut_bytes_per_group, - s->HostVMInefficiencyFactor, - s->HostVMInefficiencyFactorPrefetch, - mode_lib->soc.hostvm_min_page_size_kbytes, - mode_lib->soc.qos_parameters.qos_type, - !(display_cfg->overrides.max_outstanding_when_urgent_expected_disable), - mode_lib->soc.max_outstanding_reqs, - mode_lib->ms.support.request_size_bytes_luma, - mode_lib->ms.support.request_size_bytes_chroma, - mode_lib->ip.meta_chunk_size_kbytes, - mode_lib->ip.dchub_arb_to_ret_delay, - mode_lib->ms.TripToMemory, - mode_lib->ip.hostvm_mode, - - // output - &mode_lib->ms.ExtraLatency, - &mode_lib->ms.ExtraLatency_sr, - &mode_lib->ms.ExtraLatencyPrefetch); - - for (k = 0; k < mode_lib->ms.num_active_planes; k++) - s->impacted_dst_y_pre[k] = 0; - - s->recalc_prefetch_schedule = 0; - s->recalc_prefetch_done = 0; - do { - mode_lib->ms.support.PrefetchSupported = true; - - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - s->line_times[k] = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - s->pixel_format[k] = display_cfg->plane_descriptors[k].pixel_format; - - s->lb_source_lines_l[k] = get_num_lb_source_lines(mode_lib->ip.max_line_buffer_lines, mode_lib->ip.line_buffer_size_bits, - mode_lib->ms.NoOfDPP[k], - display_cfg->plane_descriptors[k].composition.viewport.plane0.width, - display_cfg->plane_descriptors[k].composition.viewport.plane0.height, - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio, - display_cfg->plane_descriptors[k].composition.rotation_angle); - - s->lb_source_lines_c[k] = get_num_lb_source_lines(mode_lib->ip.max_line_buffer_lines, mode_lib->ip.line_buffer_size_bits, - mode_lib->ms.NoOfDPP[k], - display_cfg->plane_descriptors[k].composition.viewport.plane1.width, - display_cfg->plane_descriptors[k].composition.viewport.plane1.height, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio, - display_cfg->plane_descriptors[k].composition.rotation_angle); - - struct dml2_core_internal_DmlPipe *myPipe = &s->myPipe; - - mode_lib->ms.TWait[k] = CalculateTWait( - display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns, - mode_lib->ms.UrgLatency, - mode_lib->ms.TripToMemory, - !dml_is_phantom_pipe(&display_cfg->plane_descriptors[k]) && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.drr_config.enabled ? - get_g6_temp_read_blackout_us(&mode_lib->soc, (unsigned int)(mode_lib->ms.uclk_freq_mhz * 1000), in_out_params->min_clk_index) : 0.0); - - myPipe->Dppclk = mode_lib->ms.RequiredDPPCLK[k]; - myPipe->Dispclk = mode_lib->ms.RequiredDISPCLK; - myPipe->PixelClock = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - myPipe->DCFClkDeepSleep = mode_lib->ms.dcfclk_deepsleep; - myPipe->DPPPerSurface = mode_lib->ms.NoOfDPP[k]; - myPipe->ScalerEnabled = display_cfg->plane_descriptors[k].composition.scaler_info.enabled; - myPipe->VRatio = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; - myPipe->VRatioChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; - myPipe->VTaps = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps; - myPipe->VTapsChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps; - myPipe->RotationAngle = display_cfg->plane_descriptors[k].composition.rotation_angle; - myPipe->mirrored = display_cfg->plane_descriptors[k].composition.mirrored; - myPipe->BlockWidth256BytesY = mode_lib->ms.Read256BlockWidthY[k]; - myPipe->BlockHeight256BytesY = mode_lib->ms.Read256BlockHeightY[k]; - myPipe->BlockWidth256BytesC = mode_lib->ms.Read256BlockWidthC[k]; - myPipe->BlockHeight256BytesC = mode_lib->ms.Read256BlockHeightC[k]; - myPipe->InterlaceEnable = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced; - myPipe->NumberOfCursors = display_cfg->plane_descriptors[k].cursor.num_cursors; - myPipe->VBlank = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active; - myPipe->HTotal = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total; - myPipe->HActive = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active; - myPipe->DCCEnable = display_cfg->plane_descriptors[k].surface.dcc.enable; - myPipe->ODMMode = mode_lib->ms.ODMMode[k]; - myPipe->SourcePixelFormat = display_cfg->plane_descriptors[k].pixel_format; - myPipe->BytePerPixelY = mode_lib->ms.BytePerPixelY[k]; - myPipe->BytePerPixelC = mode_lib->ms.BytePerPixelC[k]; - myPipe->ProgressiveToInterlaceUnitInOPP = mode_lib->ip.ptoi_supported; - -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: Calling CalculatePrefetchSchedule for k=%u\n", __func__, k); - dml2_printf("DML::%s: MaximumVStartup = %u\n", __func__, s->MaximumVStartup[k]); -#endif - CalculatePrefetchSchedule_params->display_cfg = display_cfg; - CalculatePrefetchSchedule_params->HostVMInefficiencyFactor = s->HostVMInefficiencyFactorPrefetch; - CalculatePrefetchSchedule_params->myPipe = myPipe; - CalculatePrefetchSchedule_params->DSCDelay = mode_lib->ms.DSCDelay[k]; - CalculatePrefetchSchedule_params->DPPCLKDelaySubtotalPlusCNVCFormater = mode_lib->ip.dppclk_delay_subtotal + mode_lib->ip.dppclk_delay_cnvc_formatter; - CalculatePrefetchSchedule_params->DPPCLKDelaySCL = mode_lib->ip.dppclk_delay_scl; - CalculatePrefetchSchedule_params->DPPCLKDelaySCLLBOnly = mode_lib->ip.dppclk_delay_scl_lb_only; - CalculatePrefetchSchedule_params->DPPCLKDelayCNVCCursor = mode_lib->ip.dppclk_delay_cnvc_cursor; - CalculatePrefetchSchedule_params->DISPCLKDelaySubtotal = mode_lib->ip.dispclk_delay_subtotal; - CalculatePrefetchSchedule_params->DPP_RECOUT_WIDTH = (unsigned int)(mode_lib->ms.SwathWidthY[k] / display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio); - CalculatePrefetchSchedule_params->OutputFormat = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format; - CalculatePrefetchSchedule_params->MaxInterDCNTileRepeaters = mode_lib->ip.max_inter_dcn_tile_repeaters; - CalculatePrefetchSchedule_params->VStartup = s->MaximumVStartup[k]; - CalculatePrefetchSchedule_params->HostVMMinPageSize = mode_lib->soc.hostvm_min_page_size_kbytes; - CalculatePrefetchSchedule_params->DynamicMetadataEnable = display_cfg->plane_descriptors[k].dynamic_meta_data.enable; - CalculatePrefetchSchedule_params->DynamicMetadataVMEnabled = mode_lib->ip.dynamic_metadata_vm_enabled; - CalculatePrefetchSchedule_params->DynamicMetadataLinesBeforeActiveRequired = display_cfg->plane_descriptors[k].dynamic_meta_data.lines_before_active_required; - CalculatePrefetchSchedule_params->DynamicMetadataTransmittedBytes = display_cfg->plane_descriptors[k].dynamic_meta_data.transmitted_bytes; - CalculatePrefetchSchedule_params->UrgentLatency = mode_lib->ms.UrgLatency; - CalculatePrefetchSchedule_params->ExtraLatencyPrefetch = mode_lib->ms.ExtraLatencyPrefetch; - CalculatePrefetchSchedule_params->TCalc = mode_lib->ms.TimeCalc; - CalculatePrefetchSchedule_params->vm_bytes = mode_lib->ms.vm_bytes[k]; - CalculatePrefetchSchedule_params->PixelPTEBytesPerRow = mode_lib->ms.DPTEBytesPerRow[k]; - CalculatePrefetchSchedule_params->PrefetchSourceLinesY = mode_lib->ms.PrefetchLinesY[k]; - CalculatePrefetchSchedule_params->VInitPreFillY = mode_lib->ms.PrefillY[k]; - CalculatePrefetchSchedule_params->MaxNumSwathY = mode_lib->ms.MaxNumSwathY[k]; - CalculatePrefetchSchedule_params->PrefetchSourceLinesC = mode_lib->ms.PrefetchLinesC[k]; - CalculatePrefetchSchedule_params->VInitPreFillC = mode_lib->ms.PrefillC[k]; - CalculatePrefetchSchedule_params->MaxNumSwathC = mode_lib->ms.MaxNumSwathC[k]; - CalculatePrefetchSchedule_params->swath_width_luma_ub = mode_lib->ms.swath_width_luma_ub[k]; - CalculatePrefetchSchedule_params->swath_width_chroma_ub = mode_lib->ms.swath_width_chroma_ub[k]; - CalculatePrefetchSchedule_params->SwathHeightY = mode_lib->ms.SwathHeightY[k]; - CalculatePrefetchSchedule_params->SwathHeightC = mode_lib->ms.SwathHeightC[k]; - CalculatePrefetchSchedule_params->TWait = mode_lib->ms.TWait[k]; - CalculatePrefetchSchedule_params->Ttrip = mode_lib->ms.TripToMemory; - CalculatePrefetchSchedule_params->Turg = mode_lib->ms.UrgLatency; - CalculatePrefetchSchedule_params->setup_for_tdlut = display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut; - CalculatePrefetchSchedule_params->tdlut_pte_bytes_per_frame = s->tdlut_pte_bytes_per_frame[k]; - CalculatePrefetchSchedule_params->tdlut_bytes_per_frame = s->tdlut_bytes_per_frame[k]; - CalculatePrefetchSchedule_params->tdlut_opt_time = s->tdlut_opt_time[k]; - CalculatePrefetchSchedule_params->tdlut_drain_time = s->tdlut_drain_time[k]; - CalculatePrefetchSchedule_params->num_cursors = (display_cfg->plane_descriptors[k].cursor.cursor_width > 0); - CalculatePrefetchSchedule_params->cursor_bytes_per_chunk = s->cursor_bytes_per_chunk[k]; - CalculatePrefetchSchedule_params->cursor_bytes_per_line = s->cursor_bytes_per_line[k]; - CalculatePrefetchSchedule_params->dcc_enable = display_cfg->plane_descriptors[k].surface.dcc.enable; - CalculatePrefetchSchedule_params->mrq_present = mode_lib->ip.dcn_mrq_present; - CalculatePrefetchSchedule_params->meta_row_bytes = mode_lib->ms.meta_row_bytes[k]; - CalculatePrefetchSchedule_params->mall_prefetch_sdp_overhead_factor = mode_lib->ms.mall_prefetch_sdp_overhead_factor[k]; - CalculatePrefetchSchedule_params->impacted_dst_y_pre = s->impacted_dst_y_pre[k]; - CalculatePrefetchSchedule_params->vactive_sw_bw_l = mode_lib->ms.vactive_sw_bw_l[k]; - CalculatePrefetchSchedule_params->vactive_sw_bw_c = mode_lib->ms.vactive_sw_bw_c[k]; - - // output - CalculatePrefetchSchedule_params->DSTXAfterScaler = &s->DSTXAfterScaler[k]; - CalculatePrefetchSchedule_params->DSTYAfterScaler = &s->DSTYAfterScaler[k]; - CalculatePrefetchSchedule_params->dst_y_prefetch = &mode_lib->ms.dst_y_prefetch[k]; - CalculatePrefetchSchedule_params->dst_y_per_vm_vblank = &mode_lib->ms.LinesForVM[k]; - CalculatePrefetchSchedule_params->dst_y_per_row_vblank = &mode_lib->ms.LinesForDPTERow[k]; - CalculatePrefetchSchedule_params->VRatioPrefetchY = &mode_lib->ms.VRatioPreY[k]; - CalculatePrefetchSchedule_params->VRatioPrefetchC = &mode_lib->ms.VRatioPreC[k]; - CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWLuma = &mode_lib->ms.RequiredPrefetchPixelDataBWLuma[k]; // prefetch_sw_bw_l - CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWChroma = &mode_lib->ms.RequiredPrefetchPixelDataBWChroma[k]; // prefetch_sw_bw_c - CalculatePrefetchSchedule_params->RequiredPrefetchBWOTO = &mode_lib->ms.RequiredPrefetchBWOTO[k]; - CalculatePrefetchSchedule_params->NotEnoughTimeForDynamicMetadata = &mode_lib->ms.NoTimeForDynamicMetadata[k]; - CalculatePrefetchSchedule_params->Tno_bw = &mode_lib->ms.Tno_bw[k]; - CalculatePrefetchSchedule_params->Tno_bw_flip = &mode_lib->ms.Tno_bw_flip[k]; - CalculatePrefetchSchedule_params->prefetch_vmrow_bw = &mode_lib->ms.prefetch_vmrow_bw[k]; - CalculatePrefetchSchedule_params->Tdmdl_vm = &s->dummy_single[0]; - CalculatePrefetchSchedule_params->Tdmdl = &s->dummy_single[1]; - CalculatePrefetchSchedule_params->TSetup = &s->dummy_single[2]; - CalculatePrefetchSchedule_params->Tvm_trips = &s->Tvm_trips[k]; - CalculatePrefetchSchedule_params->Tr0_trips = &s->Tr0_trips[k]; - CalculatePrefetchSchedule_params->Tvm_trips_flip = &s->Tvm_trips_flip[k]; - CalculatePrefetchSchedule_params->Tr0_trips_flip = &s->Tr0_trips_flip[k]; - CalculatePrefetchSchedule_params->Tvm_trips_flip_rounded = &s->Tvm_trips_flip_rounded[k]; - CalculatePrefetchSchedule_params->Tr0_trips_flip_rounded = &s->Tr0_trips_flip_rounded[k]; - CalculatePrefetchSchedule_params->VUpdateOffsetPix = &s->dummy_integer[0]; - CalculatePrefetchSchedule_params->VUpdateWidthPix = &s->dummy_integer[1]; - CalculatePrefetchSchedule_params->VReadyOffsetPix = &s->dummy_integer[2]; - CalculatePrefetchSchedule_params->prefetch_cursor_bw = &mode_lib->ms.prefetch_cursor_bw[k]; - CalculatePrefetchSchedule_params->prefetch_sw_bytes = &s->prefetch_sw_bytes[k]; - CalculatePrefetchSchedule_params->Tpre_rounded = &s->Tpre_rounded[k]; - CalculatePrefetchSchedule_params->Tpre_oto = &s->Tpre_oto[k]; - CalculatePrefetchSchedule_params->prefetch_swath_time_us = &s->prefetch_swath_time_us[k]; - - mode_lib->ms.NoTimeForPrefetch[k] = CalculatePrefetchSchedule(&mode_lib->scratch, CalculatePrefetchSchedule_params); - - mode_lib->ms.support.PrefetchSupported &= !mode_lib->ms.NoTimeForPrefetch[k]; - dml2_printf("DML::%s: k=%d, dst_y_per_vm_vblank = %f\n", __func__, k, *CalculatePrefetchSchedule_params->dst_y_per_vm_vblank); - dml2_printf("DML::%s: k=%d, dst_y_per_row_vblank = %f\n", __func__, k, *CalculatePrefetchSchedule_params->dst_y_per_row_vblank); - } // for k num_planes - - CalculateDCFCLKDeepSleepTdlut( - display_cfg, - mode_lib->ms.num_active_planes, - mode_lib->ms.BytePerPixelY, - mode_lib->ms.BytePerPixelC, - mode_lib->ms.SwathWidthY, - mode_lib->ms.SwathWidthC, - mode_lib->ms.NoOfDPP, - mode_lib->ms.PSCL_FACTOR, - mode_lib->ms.PSCL_FACTOR_CHROMA, - mode_lib->ms.RequiredDPPCLK, - mode_lib->ms.vactive_sw_bw_l, - mode_lib->ms.vactive_sw_bw_c, - mode_lib->soc.return_bus_width_bytes, - mode_lib->ms.RequiredDISPCLK, - s->tdlut_bytes_to_deliver, - s->prefetch_swath_time_us, - - /* Output */ - &mode_lib->ms.dcfclk_deepsleep); - - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - if (mode_lib->ms.dst_y_prefetch[k] < 2.0 - || mode_lib->ms.LinesForVM[k] >= 32.0 - || mode_lib->ms.LinesForDPTERow[k] >= 16.0 - || mode_lib->ms.NoTimeForPrefetch[k] == true - || s->DSTYAfterScaler[k] > 8) { - mode_lib->ms.support.PrefetchSupported = false; - dml2_printf("DML::%s: k=%d, dst_y_prefetch=%f (should not be < 2)\n", __func__, k, mode_lib->ms.dst_y_prefetch[k]); - dml2_printf("DML::%s: k=%d, LinesForVM=%f (should not be >= 32)\n", __func__, k, mode_lib->ms.LinesForVM[k]); - dml2_printf("DML::%s: k=%d, LinesForDPTERow=%f (should not be >= 16)\n", __func__, k, mode_lib->ms.LinesForDPTERow[k]); - dml2_printf("DML::%s: k=%d, DSTYAfterScaler=%d (should be <= 8)\n", __func__, k, s->DSTYAfterScaler[k]); - dml2_printf("DML::%s: k=%d, NoTimeForPrefetch=%d\n", __func__, k, mode_lib->ms.NoTimeForPrefetch[k]); - } - } - - mode_lib->ms.support.DynamicMetadataSupported = true; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.NoTimeForDynamicMetadata[k] == true) { - mode_lib->ms.support.DynamicMetadataSupported = false; - } - } - - mode_lib->ms.support.VRatioInPrefetchSupported = true; - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - if (mode_lib->ms.VRatioPreY[k] > __DML2_CALCS_MAX_VRATIO_PRE__ || - mode_lib->ms.VRatioPreC[k] > __DML2_CALCS_MAX_VRATIO_PRE__) { - mode_lib->ms.support.VRatioInPrefetchSupported = false; - dml2_printf("DML::%s: k=%d VRatioPreY = %f (should be <= %f)\n", __func__, k, mode_lib->ms.VRatioPreY[k], __DML2_CALCS_MAX_VRATIO_PRE__); - dml2_printf("DML::%s: k=%d VRatioPreC = %f (should be <= %f)\n", __func__, k, mode_lib->ms.VRatioPreC[k], __DML2_CALCS_MAX_VRATIO_PRE__); - dml2_printf("DML::%s: VRatioInPrefetchSupported = %u\n", __func__, mode_lib->ms.support.VRatioInPrefetchSupported); - } - } - - mode_lib->ms.support.PrefetchSupported &= mode_lib->ms.support.VRatioInPrefetchSupported; - - // By default, do not recalc prefetch schedule - s->recalc_prefetch_schedule = 0; - - // Only do urg vs prefetch bandwidth check, flip schedule check, power saving feature support check IF the Prefetch Schedule Check is ok - if (mode_lib->ms.support.PrefetchSupported) { - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - // Calculate Urgent burst factor for prefetch -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%d, Calling CalculateUrgentBurstFactor (for prefetch)\n", __func__, k); - dml2_printf("DML::%s: k=%d, VRatioPreY=%f\n", __func__, k, mode_lib->ms.VRatioPreY[k]); - dml2_printf("DML::%s: k=%d, VRatioPreC=%f\n", __func__, k, mode_lib->ms.VRatioPreC[k]); -#endif - CalculateUrgentBurstFactor( - &display_cfg->plane_descriptors[k], - mode_lib->ms.swath_width_luma_ub[k], - mode_lib->ms.swath_width_chroma_ub[k], - mode_lib->ms.SwathHeightY[k], - mode_lib->ms.SwathHeightC[k], - s->line_times[k], - mode_lib->ms.UrgLatency, - mode_lib->ms.VRatioPreY[k], - mode_lib->ms.VRatioPreC[k], - mode_lib->ms.BytePerPixelInDETY[k], - mode_lib->ms.BytePerPixelInDETC[k], - mode_lib->ms.DETBufferSizeY[k], - mode_lib->ms.DETBufferSizeC[k], - /* Output */ - &mode_lib->ms.UrgentBurstFactorLumaPre[k], - &mode_lib->ms.UrgentBurstFactorChromaPre[k], - &mode_lib->ms.NotEnoughUrgentLatencyHidingPre[k]); - } - - // Calculate urgent bandwidth required, both urg and non urg peak bandwidth - // assume flip bw is 0 at this point - for (k = 0; k < mode_lib->ms.num_active_planes; k++) - mode_lib->ms.final_flip_bw[k] = 0; - - calculate_peak_bandwidth_params->urg_vactive_bandwidth_required = mode_lib->ms.support.urg_vactive_bandwidth_required; - calculate_peak_bandwidth_params->urg_bandwidth_required = mode_lib->ms.support.urg_bandwidth_required; - calculate_peak_bandwidth_params->urg_bandwidth_required_qual = mode_lib->ms.support.urg_bandwidth_required_qual; - calculate_peak_bandwidth_params->non_urg_bandwidth_required = mode_lib->ms.support.non_urg_bandwidth_required; - calculate_peak_bandwidth_params->surface_avg_vactive_required_bw = mode_lib->ms.surface_avg_vactive_required_bw; - calculate_peak_bandwidth_params->surface_peak_required_bw = mode_lib->ms.surface_peak_required_bw; - - calculate_peak_bandwidth_params->display_cfg = display_cfg; - calculate_peak_bandwidth_params->inc_flip_bw = 0; - calculate_peak_bandwidth_params->num_active_planes = mode_lib->ms.num_active_planes; - calculate_peak_bandwidth_params->num_of_dpp = mode_lib->ms.NoOfDPP; - calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0; - calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1; - calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p0; - calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p1; - calculate_peak_bandwidth_params->mall_prefetch_sdp_overhead_factor = mode_lib->ms.mall_prefetch_sdp_overhead_factor; - calculate_peak_bandwidth_params->mall_prefetch_dram_overhead_factor = mode_lib->ms.mall_prefetch_dram_overhead_factor; - - calculate_peak_bandwidth_params->surface_read_bandwidth_l = mode_lib->ms.vactive_sw_bw_l; - calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->ms.vactive_sw_bw_c; - calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->ms.RequiredPrefetchPixelDataBWLuma; - calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->ms.RequiredPrefetchPixelDataBWChroma; - calculate_peak_bandwidth_params->prefetch_bandwidth_oto = mode_lib->ms.RequiredPrefetchBWOTO; - calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->ms.excess_vactive_fill_bw_l; - calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->ms.excess_vactive_fill_bw_c; - calculate_peak_bandwidth_params->cursor_bw = mode_lib->ms.cursor_bw; - calculate_peak_bandwidth_params->dpte_row_bw = mode_lib->ms.dpte_row_bw; - calculate_peak_bandwidth_params->meta_row_bw = mode_lib->ms.meta_row_bw; - calculate_peak_bandwidth_params->prefetch_cursor_bw = mode_lib->ms.prefetch_cursor_bw; - calculate_peak_bandwidth_params->prefetch_vmrow_bw = mode_lib->ms.prefetch_vmrow_bw; - calculate_peak_bandwidth_params->flip_bw = mode_lib->ms.final_flip_bw; - calculate_peak_bandwidth_params->urgent_burst_factor_l = mode_lib->ms.UrgentBurstFactorLuma; - calculate_peak_bandwidth_params->urgent_burst_factor_c = mode_lib->ms.UrgentBurstFactorChroma; - calculate_peak_bandwidth_params->urgent_burst_factor_cursor = mode_lib->ms.UrgentBurstFactorCursor; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_l = mode_lib->ms.UrgentBurstFactorLumaPre; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_c = mode_lib->ms.UrgentBurstFactorChromaPre; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_cursor = mode_lib->ms.UrgentBurstFactorCursorPre; - - calculate_peak_bandwidth_required( - &mode_lib->scratch, - calculate_peak_bandwidth_params); - - // Check urg peak bandwidth against available urg bw - // check at SDP and DRAM, for all soc states (SVP prefetch an Sys Active) - check_urgent_bandwidth_support( - &s->dummy_single[0], // double* frac_urg_bandwidth - &s->dummy_single[1], // double* frac_urg_bandwidth_mall - &mode_lib->ms.support.UrgVactiveBandwidthSupport, - &mode_lib->ms.support.PrefetchBandwidthSupported, - - mode_lib->soc.mall_allocated_for_dcn_mbytes, - mode_lib->ms.support.non_urg_bandwidth_required, - mode_lib->ms.support.urg_vactive_bandwidth_required, - mode_lib->ms.support.urg_bandwidth_required, - mode_lib->ms.support.urg_bandwidth_available); - - mode_lib->ms.support.PrefetchSupported &= mode_lib->ms.support.PrefetchBandwidthSupported; - dml2_printf("DML::%s: PrefetchBandwidthSupported=%0d\n", __func__, mode_lib->ms.support.PrefetchBandwidthSupported); - - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - if (mode_lib->ms.NotEnoughUrgentLatencyHidingPre[k]) { - mode_lib->ms.support.PrefetchSupported = false; - dml2_printf("DML::%s: k=%d, NotEnoughUrgentLatencyHidingPre=%d\n", __func__, k, mode_lib->ms.NotEnoughUrgentLatencyHidingPre[k]); - } - } - -#ifdef DML_GLOBAL_PREFETCH_CHECK - if (mode_lib->ms.support.PrefetchSupported && mode_lib->ms.num_active_planes > 1 && s->recalc_prefetch_done == 0) { - CheckGlobalPrefetchAdmissibility_params->num_active_planes = mode_lib->ms.num_active_planes; - CheckGlobalPrefetchAdmissibility_params->pixel_format = s->pixel_format; - CheckGlobalPrefetchAdmissibility_params->chunk_bytes_l = mode_lib->ip.pixel_chunk_size_kbytes * 1024; - CheckGlobalPrefetchAdmissibility_params->chunk_bytes_c = mode_lib->ip.pixel_chunk_size_kbytes * 1024; - CheckGlobalPrefetchAdmissibility_params->lb_source_lines_l = s->lb_source_lines_l; - CheckGlobalPrefetchAdmissibility_params->lb_source_lines_c = s->lb_source_lines_c; - CheckGlobalPrefetchAdmissibility_params->swath_height_l = mode_lib->ms.SwathHeightY; - CheckGlobalPrefetchAdmissibility_params->swath_height_c = mode_lib->ms.SwathHeightC; - CheckGlobalPrefetchAdmissibility_params->rob_buffer_size_kbytes = mode_lib->ip.rob_buffer_size_kbytes; - CheckGlobalPrefetchAdmissibility_params->compressed_buffer_size_kbytes = mode_lib->ms.CompressedBufferSizeInkByte; - CheckGlobalPrefetchAdmissibility_params->detile_buffer_size_bytes_l = mode_lib->ms.DETBufferSizeY; - CheckGlobalPrefetchAdmissibility_params->detile_buffer_size_bytes_c = mode_lib->ms.DETBufferSizeC; - CheckGlobalPrefetchAdmissibility_params->full_swath_bytes_l = s->full_swath_bytes_l; - CheckGlobalPrefetchAdmissibility_params->full_swath_bytes_c = s->full_swath_bytes_c; - CheckGlobalPrefetchAdmissibility_params->prefetch_sw_bytes = s->prefetch_sw_bytes; - CheckGlobalPrefetchAdmissibility_params->Tpre_rounded = s->Tpre_rounded; - CheckGlobalPrefetchAdmissibility_params->Tpre_oto = s->Tpre_oto; - CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps = mode_lib->ms.support.urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]; - CheckGlobalPrefetchAdmissibility_params->line_time = s->line_times; - CheckGlobalPrefetchAdmissibility_params->dst_y_prefetch = mode_lib->ms.dst_y_prefetch; - if (CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps < 10 * 1024) - CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps = 10 * 1024; - - CheckGlobalPrefetchAdmissibility_params->estimated_dcfclk_mhz = (CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps / (double) mode_lib->soc.return_bus_width_bytes) / - ((double)mode_lib->soc.qos_parameters.derate_table.system_active_urgent.dcfclk_derate_percent / 100.0); - - // if recalc_prefetch_schedule is set, recalculate the prefetch schedule with the new impacted_Tpre, prefetch should be possible - CheckGlobalPrefetchAdmissibility_params->recalc_prefetch_schedule = &s->recalc_prefetch_schedule; - CheckGlobalPrefetchAdmissibility_params->impacted_dst_y_pre = s->impacted_dst_y_pre; - mode_lib->ms.support.PrefetchSupported = CheckGlobalPrefetchAdmissibility(&mode_lib->scratch, CheckGlobalPrefetchAdmissibility_params); - s->recalc_prefetch_done = 1; - s->recalc_prefetch_schedule = 1; - } -#endif - } // prefetch schedule ok, do urg bw and flip schedule - } while (s->recalc_prefetch_schedule); - - // Flip Schedule - // Both prefetch schedule and BW okay - if (mode_lib->ms.support.PrefetchSupported == true) { - mode_lib->ms.BandwidthAvailableForImmediateFlip = - get_bandwidth_available_for_immediate_flip( - dml2_core_internal_soc_state_sys_active, - mode_lib->ms.support.urg_bandwidth_required_qual, // no flip - mode_lib->ms.support.urg_bandwidth_available); - - mode_lib->ms.TotImmediateFlipBytes = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - if (display_cfg->plane_descriptors[k].immediate_flip) { - s->per_pipe_flip_bytes[k] = get_pipe_flip_bytes( - s->HostVMInefficiencyFactor, - mode_lib->ms.vm_bytes[k], - mode_lib->ms.DPTEBytesPerRow[k], - mode_lib->ms.meta_row_bytes[k]); - } else { - s->per_pipe_flip_bytes[k] = 0; - } - mode_lib->ms.TotImmediateFlipBytes += s->per_pipe_flip_bytes[k] * mode_lib->ms.NoOfDPP[k]; - - } - - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - CalculateFlipSchedule( - &mode_lib->scratch, - display_cfg->plane_descriptors[k].immediate_flip, - 1, // use_lb_flip_bw - s->HostVMInefficiencyFactor, - s->Tvm_trips_flip[k], - s->Tr0_trips_flip[k], - s->Tvm_trips_flip_rounded[k], - s->Tr0_trips_flip_rounded[k], - display_cfg->gpuvm_enable, - mode_lib->ms.vm_bytes[k], - mode_lib->ms.DPTEBytesPerRow[k], - mode_lib->ms.BandwidthAvailableForImmediateFlip, - mode_lib->ms.TotImmediateFlipBytes, - display_cfg->plane_descriptors[k].pixel_format, - (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)), - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, - mode_lib->ms.Tno_bw_flip[k], - mode_lib->ms.dpte_row_height[k], - mode_lib->ms.dpte_row_height_chroma[k], - mode_lib->ms.use_one_row_for_frame_flip[k], - mode_lib->ip.max_flip_time_us, - mode_lib->ip.max_flip_time_lines, - s->per_pipe_flip_bytes[k], - mode_lib->ms.meta_row_bytes[k], - s->meta_row_height_luma[k], - s->meta_row_height_chroma[k], - mode_lib->ip.dcn_mrq_present && display_cfg->plane_descriptors[k].surface.dcc.enable, - - /* Output */ - &mode_lib->ms.dst_y_per_vm_flip[k], - &mode_lib->ms.dst_y_per_row_flip[k], - &mode_lib->ms.final_flip_bw[k], - &mode_lib->ms.ImmediateFlipSupportedForPipe[k]); - } - - calculate_peak_bandwidth_params->urg_vactive_bandwidth_required = s->dummy_bw; - calculate_peak_bandwidth_params->urg_bandwidth_required = mode_lib->ms.support.urg_bandwidth_required_flip; - calculate_peak_bandwidth_params->urg_bandwidth_required_qual = s->dummy_bw; - calculate_peak_bandwidth_params->non_urg_bandwidth_required = mode_lib->ms.support.non_urg_bandwidth_required_flip; - calculate_peak_bandwidth_params->surface_avg_vactive_required_bw = s->surface_dummy_bw; - calculate_peak_bandwidth_params->surface_peak_required_bw = mode_lib->ms.surface_peak_required_bw; - - calculate_peak_bandwidth_params->display_cfg = display_cfg; - calculate_peak_bandwidth_params->inc_flip_bw = 1; - calculate_peak_bandwidth_params->num_active_planes = mode_lib->ms.num_active_planes; - calculate_peak_bandwidth_params->num_of_dpp = mode_lib->ms.NoOfDPP; - calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0; - calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1; - calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p0; - calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p1; - calculate_peak_bandwidth_params->mall_prefetch_sdp_overhead_factor = mode_lib->ms.mall_prefetch_sdp_overhead_factor; - calculate_peak_bandwidth_params->mall_prefetch_dram_overhead_factor = mode_lib->ms.mall_prefetch_dram_overhead_factor; - - calculate_peak_bandwidth_params->surface_read_bandwidth_l = mode_lib->ms.vactive_sw_bw_l; - calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->ms.vactive_sw_bw_c; - calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->ms.RequiredPrefetchPixelDataBWLuma; - calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->ms.RequiredPrefetchPixelDataBWChroma; - calculate_peak_bandwidth_params->prefetch_bandwidth_oto = mode_lib->ms.RequiredPrefetchBWOTO; - calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->ms.excess_vactive_fill_bw_l; - calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->ms.excess_vactive_fill_bw_c; - calculate_peak_bandwidth_params->cursor_bw = mode_lib->ms.cursor_bw; - calculate_peak_bandwidth_params->dpte_row_bw = mode_lib->ms.dpte_row_bw; - calculate_peak_bandwidth_params->meta_row_bw = mode_lib->ms.meta_row_bw; - calculate_peak_bandwidth_params->prefetch_cursor_bw = mode_lib->ms.prefetch_cursor_bw; - calculate_peak_bandwidth_params->prefetch_vmrow_bw = mode_lib->ms.prefetch_vmrow_bw; - calculate_peak_bandwidth_params->flip_bw = mode_lib->ms.final_flip_bw; - calculate_peak_bandwidth_params->urgent_burst_factor_l = mode_lib->ms.UrgentBurstFactorLuma; - calculate_peak_bandwidth_params->urgent_burst_factor_c = mode_lib->ms.UrgentBurstFactorChroma; - calculate_peak_bandwidth_params->urgent_burst_factor_cursor = mode_lib->ms.UrgentBurstFactorCursor; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_l = mode_lib->ms.UrgentBurstFactorLumaPre; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_c = mode_lib->ms.UrgentBurstFactorChromaPre; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_cursor = mode_lib->ms.UrgentBurstFactorCursorPre; - - calculate_peak_bandwidth_required( - &mode_lib->scratch, - calculate_peak_bandwidth_params); - - calculate_immediate_flip_bandwidth_support( - &s->dummy_single[0], // double* frac_urg_bandwidth_flip - &mode_lib->ms.support.ImmediateFlipSupport, - - dml2_core_internal_soc_state_sys_active, - mode_lib->ms.support.urg_bandwidth_required_flip, - mode_lib->ms.support.non_urg_bandwidth_required_flip, - mode_lib->ms.support.urg_bandwidth_available); - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (display_cfg->plane_descriptors[k].immediate_flip == true && mode_lib->ms.ImmediateFlipSupportedForPipe[k] == false) - mode_lib->ms.support.ImmediateFlipSupport = false; - } - - } else { // if prefetch not support, assume iflip is not supported too - mode_lib->ms.support.ImmediateFlipSupport = false; - } - - s->mSOCParameters.UrgentLatency = mode_lib->ms.UrgLatency; - s->mSOCParameters.ExtraLatency = mode_lib->ms.ExtraLatency; - s->mSOCParameters.ExtraLatency_sr = mode_lib->ms.ExtraLatency_sr; - s->mSOCParameters.WritebackLatency = mode_lib->soc.qos_parameters.writeback.base_latency_us; - s->mSOCParameters.DRAMClockChangeLatency = mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us; - s->mSOCParameters.FCLKChangeLatency = mode_lib->soc.power_management_parameters.fclk_change_blackout_us; - s->mSOCParameters.SRExitTime = mode_lib->soc.power_management_parameters.stutter_exit_latency_us; - s->mSOCParameters.SREnterPlusExitTime = mode_lib->soc.power_management_parameters.stutter_enter_plus_exit_latency_us; - s->mSOCParameters.SRExitZ8Time = mode_lib->soc.power_management_parameters.z8_stutter_exit_latency_us; - s->mSOCParameters.SREnterPlusExitZ8Time = mode_lib->soc.power_management_parameters.z8_stutter_enter_plus_exit_latency_us; - s->mSOCParameters.USRRetrainingLatency = 0; - s->mSOCParameters.SMNLatency = 0; - s->mSOCParameters.g6_temp_read_blackout_us = get_g6_temp_read_blackout_us(&mode_lib->soc, (unsigned int)(mode_lib->ms.uclk_freq_mhz * 1000), in_out_params->min_clk_index); - s->mSOCParameters.max_urgent_latency_us = get_max_urgent_latency_us(&mode_lib->soc.qos_parameters.qos_params.dcn4x, mode_lib->ms.uclk_freq_mhz, mode_lib->ms.FabricClock, in_out_params->min_clk_index); - s->mSOCParameters.df_response_time_us = mode_lib->soc.qos_parameters.qos_params.dcn4x.df_qos_response_time_fclk_cycles / mode_lib->ms.FabricClock; - s->mSOCParameters.qos_type = mode_lib->soc.qos_parameters.qos_type; - - CalculateWatermarks_params->display_cfg = display_cfg; - CalculateWatermarks_params->USRRetrainingRequired = false; - CalculateWatermarks_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; - CalculateWatermarks_params->MaxLineBufferLines = mode_lib->ip.max_line_buffer_lines; - CalculateWatermarks_params->LineBufferSize = mode_lib->ip.line_buffer_size_bits; - CalculateWatermarks_params->WritebackInterfaceBufferSize = mode_lib->ip.writeback_interface_buffer_size_kbytes; - CalculateWatermarks_params->DCFCLK = mode_lib->ms.DCFCLK; - CalculateWatermarks_params->SynchronizeTimings = display_cfg->overrides.synchronize_timings; - CalculateWatermarks_params->SynchronizeDRRDisplaysForUCLKPStateChange = display_cfg->overrides.synchronize_ddr_displays_for_uclk_pstate_change; - CalculateWatermarks_params->dpte_group_bytes = mode_lib->ms.dpte_group_bytes; - CalculateWatermarks_params->mmSOCParameters = s->mSOCParameters; - CalculateWatermarks_params->WritebackChunkSize = mode_lib->ip.writeback_chunk_size_kbytes; - CalculateWatermarks_params->SOCCLK = mode_lib->ms.SOCCLK; - CalculateWatermarks_params->DCFClkDeepSleep = mode_lib->ms.dcfclk_deepsleep; - CalculateWatermarks_params->DETBufferSizeY = mode_lib->ms.DETBufferSizeY; - CalculateWatermarks_params->DETBufferSizeC = mode_lib->ms.DETBufferSizeC; - CalculateWatermarks_params->SwathHeightY = mode_lib->ms.SwathHeightY; - CalculateWatermarks_params->SwathHeightC = mode_lib->ms.SwathHeightC; - CalculateWatermarks_params->SwathWidthY = mode_lib->ms.SwathWidthY; - CalculateWatermarks_params->SwathWidthC = mode_lib->ms.SwathWidthC; - CalculateWatermarks_params->DPPPerSurface = mode_lib->ms.NoOfDPP; - CalculateWatermarks_params->BytePerPixelDETY = mode_lib->ms.BytePerPixelInDETY; - CalculateWatermarks_params->BytePerPixelDETC = mode_lib->ms.BytePerPixelInDETC; - CalculateWatermarks_params->DSTXAfterScaler = s->DSTXAfterScaler; - CalculateWatermarks_params->DSTYAfterScaler = s->DSTYAfterScaler; - CalculateWatermarks_params->UnboundedRequestEnabled = mode_lib->ms.UnboundedRequestEnabled; - CalculateWatermarks_params->CompressedBufferSizeInkByte = mode_lib->ms.CompressedBufferSizeInkByte; - CalculateWatermarks_params->meta_row_height_l = s->meta_row_height_luma; - CalculateWatermarks_params->meta_row_height_c = s->meta_row_height_chroma; - - // Output - CalculateWatermarks_params->Watermark = &mode_lib->ms.support.watermarks; // Watermarks *Watermark - CalculateWatermarks_params->DRAMClockChangeSupport = mode_lib->ms.support.DRAMClockChangeSupport; - CalculateWatermarks_params->global_dram_clock_change_supported = &mode_lib->ms.support.global_dram_clock_change_supported; - CalculateWatermarks_params->MaxActiveDRAMClockChangeLatencySupported = &s->dummy_single_array[0]; // double *MaxActiveDRAMClockChangeLatencySupported[] - CalculateWatermarks_params->SubViewportLinesNeededInMALL = mode_lib->ms.SubViewportLinesNeededInMALL; // unsigned int SubViewportLinesNeededInMALL[] - CalculateWatermarks_params->FCLKChangeSupport = mode_lib->ms.support.FCLKChangeSupport; - CalculateWatermarks_params->global_fclk_change_supported = &mode_lib->ms.support.global_fclk_change_supported; - CalculateWatermarks_params->MaxActiveFCLKChangeLatencySupported = &s->dummy_single[0]; // double *MaxActiveFCLKChangeLatencySupported - CalculateWatermarks_params->USRRetrainingSupport = &mode_lib->ms.support.USRRetrainingSupport; - CalculateWatermarks_params->g6_temp_read_support = &mode_lib->ms.support.g6_temp_read_support; - CalculateWatermarks_params->VActiveLatencyHidingMargin = mode_lib->ms.VActiveLatencyHidingMargin; - CalculateWatermarks_params->VActiveLatencyHidingUs = mode_lib->ms.VActiveLatencyHidingUs; - - CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport(&mode_lib->scratch, CalculateWatermarks_params); - - calculate_pstate_keepout_dst_lines(display_cfg, &mode_lib->ms.support.watermarks, s->dummy_integer_array[0]); - } - dml2_printf("DML::%s: Done prefetch calculation\n", __func__); - // End of Prefetch Check + dml_core_ms_prefetch_check(mode_lib, display_cfg); mode_lib->ms.support.max_urgent_latency_us = s->mSOCParameters.max_urgent_latency_us; @@ -9546,8 +9485,8 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.dram_change_vactive_det_fill_delay_us); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: max_urgent_latency_us = %f\n", __func__, s->mSOCParameters.max_urgent_latency_us); - dml2_printf("DML::%s: ROBSupport = %u\n", __func__, mode_lib->ms.support.ROBSupport); + DML_LOG_VERBOSE("DML::%s: max_urgent_latency_us = %f\n", __func__, s->mSOCParameters.max_urgent_latency_us); + DML_LOG_VERBOSE("DML::%s: ROBSupport = %u\n", __func__, mode_lib->ms.support.ROBSupport); #endif /*Mode Support, Voltage State and SOC Configuration*/ @@ -9597,17 +9536,17 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out && !mode_lib->ms.support.ExceededMALLSize && mode_lib->ms.support.g6_temp_read_support && ((!display_cfg->hostvm_enable && !s->ImmediateFlipRequired) || mode_lib->ms.support.ImmediateFlipSupport)) { - dml2_printf("DML::%s: mode is supported\n", __func__); + DML_LOG_VERBOSE("DML::%s: mode is supported\n", __func__); mode_lib->ms.support.ModeSupport = true; } else { - dml2_printf("DML::%s: mode is NOT supported\n", __func__); + DML_LOG_VERBOSE("DML::%s: mode is NOT supported\n", __func__); mode_lib->ms.support.ModeSupport = false; } } // Since now the mode_support work on 1 particular power state, so there is only 1 state idx (index 0). - dml2_printf("DML::%s: ModeSupport = %u\n", __func__, mode_lib->ms.support.ModeSupport); - dml2_printf("DML::%s: ImmediateFlipSupport = %u\n", __func__, mode_lib->ms.support.ImmediateFlipSupport); + DML_LOG_VERBOSE("DML::%s: ModeSupport = %u\n", __func__, mode_lib->ms.support.ModeSupport); + DML_LOG_VERBOSE("DML::%s: ImmediateFlipSupport = %u\n", __func__, mode_lib->ms.support.ImmediateFlipSupport); for (k = 0; k < mode_lib->ms.num_active_planes; k++) { mode_lib->ms.support.MPCCombineEnable[k] = mode_lib->ms.MPCCombine[k]; @@ -9623,8 +9562,8 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.support.OutputRate[k] = mode_lib->ms.OutputRate[k]; #if defined(__DML_VBA_DEBUG__) - dml2_printf("DML::%s: k=%d, ODMMode = %u\n", __func__, k, mode_lib->ms.support.ODMMode[k]); - dml2_printf("DML::%s: k=%d, DSCEnabled = %u\n", __func__, k, mode_lib->ms.support.DSCEnabled[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, ODMMode = %u\n", __func__, k, mode_lib->ms.support.ODMMode[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, DSCEnabled = %u\n", __func__, k, mode_lib->ms.support.DSCEnabled[k]); #endif } @@ -9632,7 +9571,7 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out if (!mode_lib->ms.support.ModeSupport) dml2_print_mode_support_info(&mode_lib->ms.support, true); - dml2_printf("DML::%s: --- DONE --- \n", __func__); + DML_LOG_VERBOSE("DML::%s: --- DONE --- \n", __func__); #endif return mode_lib->ms.support.ModeSupport; @@ -9642,18 +9581,18 @@ unsigned int dml2_core_calcs_mode_support_ex(struct dml2_core_calcs_mode_support { unsigned int result; - dml2_printf("DML::%s: ------------- START ----------\n", __func__); + DML_LOG_VERBOSE("DML::%s: ------------- START ----------\n", __func__); result = dml_core_mode_support(in_out_params); if (result) *in_out_params->out_evaluation_info = in_out_params->mode_lib->ms.support; - dml2_printf("DML::%s: is_mode_support = %u (min_clk_index=%d)\n", __func__, result, in_out_params->min_clk_index); + DML_LOG_VERBOSE("DML::%s: is_mode_support = %u (min_clk_index=%d)\n", __func__, result, in_out_params->min_clk_index); for (unsigned int k = 0; k < in_out_params->in_display_cfg->num_planes; k++) - dml2_printf("DML::%s: plane_%d: reserved_vblank_time_ns = %u\n", __func__, k, in_out_params->in_display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns); + DML_LOG_VERBOSE("DML::%s: plane_%d: reserved_vblank_time_ns = %lu\n", __func__, k, in_out_params->in_display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns); - dml2_printf("DML::%s: ------------- DONE ----------\n", __func__); + DML_LOG_VERBOSE("DML::%s: ------------- DONE ----------\n", __func__); return result; } @@ -9687,19 +9626,19 @@ static void CalculatePixelDeliveryTimes( double pixel_clock_mhz = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u : HRatio = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio); - dml2_printf("DML::%s: k=%u : VRatio = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); - dml2_printf("DML::%s: k=%u : HRatioChroma = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio); - dml2_printf("DML::%s: k=%u : VRatioChroma = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio); - dml2_printf("DML::%s: k=%u : VRatioPrefetchY = %f\n", __func__, k, VRatioPrefetchY[k]); - dml2_printf("DML::%s: k=%u : VRatioPrefetchC = %f\n", __func__, k, VRatioPrefetchC[k]); - dml2_printf("DML::%s: k=%u : swath_width_luma_ub = %u\n", __func__, k, swath_width_luma_ub[k]); - dml2_printf("DML::%s: k=%u : swath_width_chroma_ub = %u\n", __func__, k, swath_width_chroma_ub[k]); - dml2_printf("DML::%s: k=%u : PSCL_THROUGHPUT = %f\n", __func__, k, PSCL_THROUGHPUT[k]); - dml2_printf("DML::%s: k=%u : PSCL_THROUGHPUT_CHROMA = %f\n", __func__, k, PSCL_THROUGHPUT_CHROMA[k]); - dml2_printf("DML::%s: k=%u : DPPPerSurface = %u\n", __func__, k, cfg_support_info->plane_support_info[k].dpps_used); - dml2_printf("DML::%s: k=%u : pixel_clock_mhz = %f\n", __func__, k, pixel_clock_mhz); - dml2_printf("DML::%s: k=%u : Dppclk = %f\n", __func__, k, Dppclk[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : HRatio = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u : VRatio = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u : HRatioChroma = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u : VRatioChroma = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u : VRatioPrefetchY = %f\n", __func__, k, VRatioPrefetchY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : VRatioPrefetchC = %f\n", __func__, k, VRatioPrefetchC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : swath_width_luma_ub = %u\n", __func__, k, swath_width_luma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : swath_width_chroma_ub = %u\n", __func__, k, swath_width_chroma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : PSCL_THROUGHPUT = %f\n", __func__, k, PSCL_THROUGHPUT[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : PSCL_THROUGHPUT_CHROMA = %f\n", __func__, k, PSCL_THROUGHPUT_CHROMA[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DPPPerSurface = %u\n", __func__, k, cfg_support_info->plane_support_info[k].dpps_used); + DML_LOG_VERBOSE("DML::%s: k=%u : pixel_clock_mhz = %f\n", __func__, k, pixel_clock_mhz); + DML_LOG_VERBOSE("DML::%s: k=%u : Dppclk = %f\n", __func__, k, Dppclk[k]); #endif if (display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio <= 1) { DisplayPipeLineDeliveryTimeLuma[k] = swath_width_luma_ub[k] * cfg_support_info->plane_support_info[k].dpps_used / display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio / pixel_clock_mhz; @@ -9733,10 +9672,10 @@ static void CalculatePixelDeliveryTimes( } } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u : DisplayPipeLineDeliveryTimeLuma = %f\n", __func__, k, DisplayPipeLineDeliveryTimeLuma[k]); - dml2_printf("DML::%s: k=%u : DisplayPipeLineDeliveryTimeLumaPrefetch = %f\n", __func__, k, DisplayPipeLineDeliveryTimeLumaPrefetch[k]); - dml2_printf("DML::%s: k=%u : DisplayPipeLineDeliveryTimeChroma = %f\n", __func__, k, DisplayPipeLineDeliveryTimeChroma[k]); - dml2_printf("DML::%s: k=%u : DisplayPipeLineDeliveryTimeChromaPrefetch = %f\n", __func__, k, DisplayPipeLineDeliveryTimeChromaPrefetch[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeLineDeliveryTimeLuma = %f\n", __func__, k, DisplayPipeLineDeliveryTimeLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeLineDeliveryTimeLumaPrefetch = %f\n", __func__, k, DisplayPipeLineDeliveryTimeLumaPrefetch[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeLineDeliveryTimeChroma = %f\n", __func__, k, DisplayPipeLineDeliveryTimeChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeLineDeliveryTimeChromaPrefetch = %f\n", __func__, k, DisplayPipeLineDeliveryTimeChromaPrefetch[k]); #endif } @@ -9752,12 +9691,12 @@ static void CalculatePixelDeliveryTimes( DisplayPipeRequestDeliveryTimeChromaPrefetch[k] = DisplayPipeLineDeliveryTimeChromaPrefetch[k] / req_per_swath_ub_c[k]; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeLuma = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeLuma[k]); - dml2_printf("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeLumaPrefetch = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeLumaPrefetch[k]); - dml2_printf("DML::%s: k=%u : req_per_swath_ub_l = %d\n", __func__, k, req_per_swath_ub_l[k]); - dml2_printf("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeChroma = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeChroma[k]); - dml2_printf("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeChromaPrefetch = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeChromaPrefetch[k]); - dml2_printf("DML::%s: k=%u : req_per_swath_ub_c = %d\n", __func__, k, req_per_swath_ub_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeLuma = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeLumaPrefetch = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeLumaPrefetch[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : req_per_swath_ub_l = %d\n", __func__, k, req_per_swath_ub_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeChroma = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeChromaPrefetch = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeChromaPrefetch[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : req_per_swath_ub_c = %d\n", __func__, k, req_per_swath_ub_c[k]); #endif } } @@ -9853,14 +9792,14 @@ static void CalculateMetaAndPTETimes(struct dml2_core_shared_CalculateMetaAndPTE } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%d, DST_Y_PER_META_ROW_NOM_L = %f\n", __func__, k, p->DST_Y_PER_META_ROW_NOM_L[k]); - dml2_printf("DML::%s: k=%d, DST_Y_PER_META_ROW_NOM_C = %f\n", __func__, k, p->DST_Y_PER_META_ROW_NOM_C[k]); - dml2_printf("DML::%s: k=%d, TimePerMetaChunkNominal = %f\n", __func__, k, p->TimePerMetaChunkNominal[k]); - dml2_printf("DML::%s: k=%d, TimePerMetaChunkVBlank = %f\n", __func__, k, p->TimePerMetaChunkVBlank[k]); - dml2_printf("DML::%s: k=%d, TimePerMetaChunkFlip = %f\n", __func__, k, p->TimePerMetaChunkFlip[k]); - dml2_printf("DML::%s: k=%d, TimePerChromaMetaChunkNominal = %f\n", __func__, k, p->TimePerChromaMetaChunkNominal[k]); - dml2_printf("DML::%s: k=%d, TimePerChromaMetaChunkVBlank = %f\n", __func__, k, p->TimePerChromaMetaChunkVBlank[k]); - dml2_printf("DML::%s: k=%d, TimePerChromaMetaChunkFlip = %f\n", __func__, k, p->TimePerChromaMetaChunkFlip[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, DST_Y_PER_META_ROW_NOM_L = %f\n", __func__, k, p->DST_Y_PER_META_ROW_NOM_L[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, DST_Y_PER_META_ROW_NOM_C = %f\n", __func__, k, p->DST_Y_PER_META_ROW_NOM_C[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, TimePerMetaChunkNominal = %f\n", __func__, k, p->TimePerMetaChunkNominal[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, TimePerMetaChunkVBlank = %f\n", __func__, k, p->TimePerMetaChunkVBlank[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, TimePerMetaChunkFlip = %f\n", __func__, k, p->TimePerMetaChunkFlip[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, TimePerChromaMetaChunkNominal = %f\n", __func__, k, p->TimePerChromaMetaChunkNominal[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, TimePerChromaMetaChunkVBlank = %f\n", __func__, k, p->TimePerChromaMetaChunkVBlank[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, TimePerChromaMetaChunkFlip = %f\n", __func__, k, p->TimePerChromaMetaChunkFlip[k]); #endif } @@ -9881,7 +9820,7 @@ static void CalculateMetaAndPTETimes(struct dml2_core_shared_CalculateMetaAndPTE else p->time_per_tdlut_group[k] = 0; - dml2_printf("DML::%s: k=%u, time_per_tdlut_group = %f\n", __func__, k, p->time_per_tdlut_group[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, time_per_tdlut_group = %f\n", __func__, k, p->time_per_tdlut_group[k]); if (p->display_cfg->gpuvm_enable == true) { if (!dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) { @@ -9897,14 +9836,14 @@ static void CalculateMetaAndPTETimes(struct dml2_core_shared_CalculateMetaAndPTE if (dpte_groups_per_row_luma_ub <= 2) { dpte_groups_per_row_luma_ub = dpte_groups_per_row_luma_ub + 1; } - dml2_printf("DML::%s: k=%u, use_one_row_for_frame = %u\n", __func__, k, p->use_one_row_for_frame[k]); - dml2_printf("DML::%s: k=%u, dpte_group_bytes = %u\n", __func__, k, p->dpte_group_bytes[k]); - dml2_printf("DML::%s: k=%u, PTERequestSizeY = %u\n", __func__, k, p->PTERequestSizeY[k]); - dml2_printf("DML::%s: k=%u, PixelPTEReqWidthY = %u\n", __func__, k, p->PixelPTEReqWidthY[k]); - dml2_printf("DML::%s: k=%u, PixelPTEReqHeightY = %u\n", __func__, k, p->PixelPTEReqHeightY[k]); - dml2_printf("DML::%s: k=%u, dpte_row_width_luma_ub = %u\n", __func__, k, p->dpte_row_width_luma_ub[k]); - dml2_printf("DML::%s: k=%u, dpte_group_width_luma = %u\n", __func__, k, dpte_group_width_luma); - dml2_printf("DML::%s: k=%u, dpte_groups_per_row_luma_ub = %u\n", __func__, k, dpte_groups_per_row_luma_ub); + DML_LOG_VERBOSE("DML::%s: k=%u, use_one_row_for_frame = %u\n", __func__, k, p->use_one_row_for_frame[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_group_bytes = %u\n", __func__, k, p->dpte_group_bytes[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PTERequestSizeY = %u\n", __func__, k, p->PTERequestSizeY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEReqWidthY = %u\n", __func__, k, p->PixelPTEReqWidthY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEReqHeightY = %u\n", __func__, k, p->PixelPTEReqHeightY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_width_luma_ub = %u\n", __func__, k, p->dpte_row_width_luma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_group_width_luma = %u\n", __func__, k, dpte_group_width_luma); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_groups_per_row_luma_ub = %u\n", __func__, k, dpte_groups_per_row_luma_ub); p->time_per_pte_group_nom_luma[k] = p->DST_Y_PER_PTE_ROW_NOM_L[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_luma_ub; p->time_per_pte_group_vblank_luma[k] = p->dst_y_per_row_vblank[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_luma_ub; @@ -9928,9 +9867,9 @@ static void CalculateMetaAndPTETimes(struct dml2_core_shared_CalculateMetaAndPTE if (dpte_groups_per_row_chroma_ub <= 2) { dpte_groups_per_row_chroma_ub = dpte_groups_per_row_chroma_ub + 1; } - dml2_printf("DML::%s: k=%u, dpte_row_width_chroma_ub = %u\n", __func__, k, p->dpte_row_width_chroma_ub[k]); - dml2_printf("DML::%s: k=%u, dpte_group_width_chroma = %u\n", __func__, k, dpte_group_width_chroma); - dml2_printf("DML::%s: k=%u, dpte_groups_per_row_chroma_ub = %u\n", __func__, k, dpte_groups_per_row_chroma_ub); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_width_chroma_ub = %u\n", __func__, k, p->dpte_row_width_chroma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_group_width_chroma = %u\n", __func__, k, dpte_group_width_chroma); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_groups_per_row_chroma_ub = %u\n", __func__, k, dpte_groups_per_row_chroma_ub); p->time_per_pte_group_nom_chroma[k] = p->DST_Y_PER_PTE_ROW_NOM_C[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_chroma_ub; p->time_per_pte_group_vblank_chroma[k] = p->dst_y_per_row_vblank[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_chroma_ub; @@ -9945,17 +9884,17 @@ static void CalculateMetaAndPTETimes(struct dml2_core_shared_CalculateMetaAndPTE p->time_per_pte_group_flip_chroma[k] = 0; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, dst_y_per_row_vblank = %f\n", __func__, k, p->dst_y_per_row_vblank[k]); - dml2_printf("DML::%s: k=%u, dst_y_per_row_flip = %f\n", __func__, k, p->dst_y_per_row_flip[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dst_y_per_row_vblank = %f\n", __func__, k, p->dst_y_per_row_vblank[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dst_y_per_row_flip = %f\n", __func__, k, p->dst_y_per_row_flip[k]); - dml2_printf("DML::%s: k=%u, DST_Y_PER_PTE_ROW_NOM_L = %f\n", __func__, k, p->DST_Y_PER_PTE_ROW_NOM_L[k]); - dml2_printf("DML::%s: k=%u, DST_Y_PER_PTE_ROW_NOM_C = %f\n", __func__, k, p->DST_Y_PER_PTE_ROW_NOM_C[k]); - dml2_printf("DML::%s: k=%u, time_per_pte_group_nom_luma = %f\n", __func__, k, p->time_per_pte_group_nom_luma[k]); - dml2_printf("DML::%s: k=%u, time_per_pte_group_vblank_luma = %f\n", __func__, k, p->time_per_pte_group_vblank_luma[k]); - dml2_printf("DML::%s: k=%u, time_per_pte_group_flip_luma = %f\n", __func__, k, p->time_per_pte_group_flip_luma[k]); - dml2_printf("DML::%s: k=%u, time_per_pte_group_nom_chroma = %f\n", __func__, k, p->time_per_pte_group_nom_chroma[k]); - dml2_printf("DML::%s: k=%u, time_per_pte_group_vblank_chroma = %f\n", __func__, k, p->time_per_pte_group_vblank_chroma[k]); - dml2_printf("DML::%s: k=%u, time_per_pte_group_flip_chroma = %f\n", __func__, k, p->time_per_pte_group_flip_chroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, DST_Y_PER_PTE_ROW_NOM_L = %f\n", __func__, k, p->DST_Y_PER_PTE_ROW_NOM_L[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, DST_Y_PER_PTE_ROW_NOM_C = %f\n", __func__, k, p->DST_Y_PER_PTE_ROW_NOM_C[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_nom_luma = %f\n", __func__, k, p->time_per_pte_group_nom_luma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_vblank_luma = %f\n", __func__, k, p->time_per_pte_group_vblank_luma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_flip_luma = %f\n", __func__, k, p->time_per_pte_group_flip_luma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_nom_chroma = %f\n", __func__, k, p->time_per_pte_group_nom_chroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_vblank_chroma = %f\n", __func__, k, p->time_per_pte_group_vblank_chroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_flip_chroma = %f\n", __func__, k, p->time_per_pte_group_flip_chroma[k]); #endif } } // CalculateMetaAndPTETimes @@ -9991,18 +9930,18 @@ static void CalculateVMGroupAndRequestTimes( double line_time; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); + DML_LOG_VERBOSE("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); #endif for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { double pixel_clock_mhz = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); bool dcc_mrq_enable = display_cfg->plane_descriptors[k].surface.dcc.enable && mrq_present; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, dcc_mrq_enable = %u\n", __func__, k, dcc_mrq_enable); - dml2_printf("DML::%s: k=%u, vm_group_bytes = %u\n", __func__, k, vm_group_bytes[k]); - dml2_printf("DML::%s: k=%u, dpde0_bytes_per_frame_ub_l = %u\n", __func__, k, dpde0_bytes_per_frame_ub_l[k]); - dml2_printf("DML::%s: k=%u, dpde0_bytes_per_frame_ub_c = %u\n", __func__, k, dpde0_bytes_per_frame_ub_c[k]); - dml2_printf("DML::%s: k=%d, meta_pte_bytes_per_frame_ub_l = %d\n", __func__, k, meta_pte_bytes_per_frame_ub_l[k]); - dml2_printf("DML::%s: k=%d, meta_pte_bytes_per_frame_ub_c = %d\n", __func__, k, meta_pte_bytes_per_frame_ub_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dcc_mrq_enable = %u\n", __func__, k, dcc_mrq_enable); + DML_LOG_VERBOSE("DML::%s: k=%u, vm_group_bytes = %u\n", __func__, k, vm_group_bytes[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpde0_bytes_per_frame_ub_l = %u\n", __func__, k, dpde0_bytes_per_frame_ub_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpde0_bytes_per_frame_ub_c = %u\n", __func__, k, dpde0_bytes_per_frame_ub_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, meta_pte_bytes_per_frame_ub_l = %d\n", __func__, k, meta_pte_bytes_per_frame_ub_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, meta_pte_bytes_per_frame_ub_c = %d\n", __func__, k, meta_pte_bytes_per_frame_ub_c[k]); #endif if (display_cfg->gpuvm_enable) { @@ -10071,13 +10010,13 @@ static void CalculateVMGroupAndRequestTimes( else TimePerVMRequestFlip[k] = 0.0; - dml2_printf("DML::%s: k=%u, dst_y_per_vm_vblank = %f\n", __func__, k, dst_y_per_vm_vblank[k]); - dml2_printf("DML::%s: k=%u, dst_y_per_vm_flip = %f\n", __func__, k, dst_y_per_vm_flip[k]); - dml2_printf("DML::%s: k=%u, line_time = %f\n", __func__, k, line_time); - dml2_printf("DML::%s: k=%u, num_group_per_lower_vm_stage_pref = %f\n", __func__, k, num_group_per_lower_vm_stage_pref); - dml2_printf("DML::%s: k=%u, num_group_per_lower_vm_stage_flip = %f\n", __func__, k, num_group_per_lower_vm_stage_flip); - dml2_printf("DML::%s: k=%u, num_req_per_lower_vm_stage_pref = %f\n", __func__, k, num_req_per_lower_vm_stage_pref); - dml2_printf("DML::%s: k=%u, num_req_per_lower_vm_stage_flip = %f\n", __func__, k, num_req_per_lower_vm_stage_flip); + DML_LOG_VERBOSE("DML::%s: k=%u, dst_y_per_vm_vblank = %f\n", __func__, k, dst_y_per_vm_vblank[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dst_y_per_vm_flip = %f\n", __func__, k, dst_y_per_vm_flip[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, line_time = %f\n", __func__, k, line_time); + DML_LOG_VERBOSE("DML::%s: k=%u, num_group_per_lower_vm_stage_pref = %d\n", __func__, k, num_group_per_lower_vm_stage_pref); + DML_LOG_VERBOSE("DML::%s: k=%u, num_group_per_lower_vm_stage_flip = %d\n", __func__, k, num_group_per_lower_vm_stage_flip); + DML_LOG_VERBOSE("DML::%s: k=%u, num_req_per_lower_vm_stage_pref = %d\n", __func__, k, num_req_per_lower_vm_stage_pref); + DML_LOG_VERBOSE("DML::%s: k=%u, num_req_per_lower_vm_stage_flip = %d\n", __func__, k, num_req_per_lower_vm_stage_flip); if (display_cfg->gpuvm_max_page_table_levels > 2) { TimePerVMGroupVBlank[k] = TimePerVMGroupVBlank[k] / 2; @@ -10094,10 +10033,10 @@ static void CalculateVMGroupAndRequestTimes( } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, TimePerVMGroupVBlank = %f\n", __func__, k, TimePerVMGroupVBlank[k]); - dml2_printf("DML::%s: k=%u, TimePerVMGroupFlip = %f\n", __func__, k, TimePerVMGroupFlip[k]); - dml2_printf("DML::%s: k=%u, TimePerVMRequestVBlank = %f\n", __func__, k, TimePerVMRequestVBlank[k]); - dml2_printf("DML::%s: k=%u, TimePerVMRequestFlip = %f\n", __func__, k, TimePerVMRequestFlip[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, TimePerVMGroupVBlank = %f\n", __func__, k, TimePerVMGroupVBlank[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, TimePerVMGroupFlip = %f\n", __func__, k, TimePerVMGroupFlip[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, TimePerVMRequestVBlank = %f\n", __func__, k, TimePerVMRequestVBlank[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, TimePerVMRequestFlip = %f\n", __func__, k, TimePerVMRequestFlip[k]); #endif } } @@ -10113,7 +10052,6 @@ static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratc unsigned int SingleVTotal = 0; bool SameTiming = true; bool FoundCriticalSurface = false; - double LastZ8StutterPeriod = 0; memset(l, 0, sizeof(struct dml2_core_calcs_CalculateStutterEfficiency_locals)); @@ -10127,9 +10065,9 @@ static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratc } l->TotalCompressedReadBandwidth = l->TotalCompressedReadBandwidth + p->ReadBandwidthSurfaceLuma[k] / math_min2(p->display_cfg->plane_descriptors[k].surface.dcc.informative.dcc_rate_plane0, l->MaximumEffectiveCompressionLuma); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, ReadBandwidthSurfaceLuma = %f\n", __func__, k, p->ReadBandwidthSurfaceLuma[k]); - dml2_printf("DML::%s: k=%u, NetDCCRateLuma = %f\n", __func__, k, p->display_cfg->plane_descriptors[k].surface.dcc.informative.dcc_rate_plane0); - dml2_printf("DML::%s: k=%u, MaximumEffectiveCompressionLuma = %f\n", __func__, k, l->MaximumEffectiveCompressionLuma); + DML_LOG_VERBOSE("DML::%s: k=%u, ReadBandwidthSurfaceLuma = %f\n", __func__, k, p->ReadBandwidthSurfaceLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, NetDCCRateLuma = %f\n", __func__, k, p->display_cfg->plane_descriptors[k].surface.dcc.informative.dcc_rate_plane0); + DML_LOG_VERBOSE("DML::%s: k=%u, MaximumEffectiveCompressionLuma = %f\n", __func__, k, l->MaximumEffectiveCompressionLuma); #endif l->TotalZeroSizeRequestReadBandwidth = l->TotalZeroSizeRequestReadBandwidth + p->ReadBandwidthSurfaceLuma[k] * p->display_cfg->plane_descriptors[k].surface.dcc.informative.fraction_of_zero_size_request_plane0; l->TotalZeroSizeCompressedReadBandwidth = l->TotalZeroSizeCompressedReadBandwidth + p->ReadBandwidthSurfaceLuma[k] * p->display_cfg->plane_descriptors[k].surface.dcc.informative.fraction_of_zero_size_request_plane0 / l->MaximumEffectiveCompressionLuma; @@ -10142,9 +10080,9 @@ static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratc } l->TotalCompressedReadBandwidth = l->TotalCompressedReadBandwidth + p->ReadBandwidthSurfaceChroma[k] / math_min2(p->display_cfg->plane_descriptors[k].surface.dcc.informative.dcc_rate_plane1, l->MaximumEffectiveCompressionChroma); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, ReadBandwidthSurfaceChroma = %f\n", __func__, k, p->ReadBandwidthSurfaceChroma[k]); - dml2_printf("DML::%s: k=%u, NetDCCRateChroma = %f\n", __func__, k, p->display_cfg->plane_descriptors[k].surface.dcc.informative.dcc_rate_plane1); - dml2_printf("DML::%s: k=%u, MaximumEffectiveCompressionChroma = %f\n", __func__, k, l->MaximumEffectiveCompressionChroma); + DML_LOG_VERBOSE("DML::%s: k=%u, ReadBandwidthSurfaceChroma = %f\n", __func__, k, p->ReadBandwidthSurfaceChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, NetDCCRateChroma = %f\n", __func__, k, p->display_cfg->plane_descriptors[k].surface.dcc.informative.dcc_rate_plane1); + DML_LOG_VERBOSE("DML::%s: k=%u, MaximumEffectiveCompressionChroma = %f\n", __func__, k, l->MaximumEffectiveCompressionChroma); #endif l->TotalZeroSizeRequestReadBandwidth = l->TotalZeroSizeRequestReadBandwidth + p->ReadBandwidthSurfaceChroma[k] * p->display_cfg->plane_descriptors[k].surface.dcc.informative.fraction_of_zero_size_request_plane1; l->TotalZeroSizeCompressedReadBandwidth = l->TotalZeroSizeCompressedReadBandwidth + p->ReadBandwidthSurfaceChroma[k] * p->display_cfg->plane_descriptors[k].surface.dcc.informative.fraction_of_zero_size_request_plane1 / l->MaximumEffectiveCompressionChroma; @@ -10160,19 +10098,19 @@ static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratc l->AverageDCCZeroSizeFraction = l->TotalZeroSizeRequestReadBandwidth / p->TotalDataReadBandwidth; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: UnboundedRequestEnabled = %u\n", __func__, p->UnboundedRequestEnabled); - dml2_printf("DML::%s: TotalCompressedReadBandwidth = %f\n", __func__, l->TotalCompressedReadBandwidth); - dml2_printf("DML::%s: TotalZeroSizeRequestReadBandwidth = %f\n", __func__, l->TotalZeroSizeRequestReadBandwidth); - dml2_printf("DML::%s: TotalZeroSizeCompressedReadBandwidth = %f\n", __func__, l->TotalZeroSizeCompressedReadBandwidth); - dml2_printf("DML::%s: MaximumEffectiveCompressionLuma = %f\n", __func__, l->MaximumEffectiveCompressionLuma); - dml2_printf("DML::%s: MaximumEffectiveCompressionChroma = %f\n", __func__, l->MaximumEffectiveCompressionChroma); - dml2_printf("DML::%s: AverageDCCCompressionRate = %f\n", __func__, l->AverageDCCCompressionRate); - dml2_printf("DML::%s: AverageDCCZeroSizeFraction = %f\n", __func__, l->AverageDCCZeroSizeFraction); + DML_LOG_VERBOSE("DML::%s: UnboundedRequestEnabled = %u\n", __func__, p->UnboundedRequestEnabled); + DML_LOG_VERBOSE("DML::%s: TotalCompressedReadBandwidth = %f\n", __func__, l->TotalCompressedReadBandwidth); + DML_LOG_VERBOSE("DML::%s: TotalZeroSizeRequestReadBandwidth = %f\n", __func__, l->TotalZeroSizeRequestReadBandwidth); + DML_LOG_VERBOSE("DML::%s: TotalZeroSizeCompressedReadBandwidth = %f\n", __func__, l->TotalZeroSizeCompressedReadBandwidth); + DML_LOG_VERBOSE("DML::%s: MaximumEffectiveCompressionLuma = %f\n", __func__, l->MaximumEffectiveCompressionLuma); + DML_LOG_VERBOSE("DML::%s: MaximumEffectiveCompressionChroma = %f\n", __func__, l->MaximumEffectiveCompressionChroma); + DML_LOG_VERBOSE("DML::%s: AverageDCCCompressionRate = %f\n", __func__, l->AverageDCCCompressionRate); + DML_LOG_VERBOSE("DML::%s: AverageDCCZeroSizeFraction = %f\n", __func__, l->AverageDCCZeroSizeFraction); - dml2_printf("DML::%s: CompbufReservedSpace64B = %u (%f kbytes)\n", __func__, p->CompbufReservedSpace64B, p->CompbufReservedSpace64B * 64 / 1024.0); - dml2_printf("DML::%s: CompbufReservedSpaceZs = %u\n", __func__, p->CompbufReservedSpaceZs); - dml2_printf("DML::%s: CompressedBufferSizeInkByte = %u kbytes\n", __func__, p->CompressedBufferSizeInkByte); - dml2_printf("DML::%s: ROBBufferSizeInKByte = %u kbytes\n", __func__, p->ROBBufferSizeInKByte); + DML_LOG_VERBOSE("DML::%s: CompbufReservedSpace64B = %u (%f kbytes)\n", __func__, p->CompbufReservedSpace64B, p->CompbufReservedSpace64B * 64 / 1024.0); + DML_LOG_VERBOSE("DML::%s: CompbufReservedSpaceZs = %u\n", __func__, p->CompbufReservedSpaceZs); + DML_LOG_VERBOSE("DML::%s: CompressedBufferSizeInkByte = %u kbytes\n", __func__, p->CompressedBufferSizeInkByte); + DML_LOG_VERBOSE("DML::%s: ROBBufferSizeInKByte = %u kbytes\n", __func__, p->ROBBufferSizeInKByte); #endif if (l->AverageDCCZeroSizeFraction == 1) { l->AverageZeroSizeCompressionRate = l->TotalZeroSizeRequestReadBandwidth / l->TotalZeroSizeCompressedReadBandwidth; @@ -10189,10 +10127,10 @@ static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratc #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: min 1 = %f\n", __func__, p->CompressedBufferSizeInkByte * 1024 * l->AverageDCCCompressionRate); - dml2_printf("DML::%s: min 2 = %f\n", __func__, p->MetaFIFOSizeInKEntries * 1024 * 64 / (l->AverageDCCZeroSizeFraction / l->AverageZeroSizeCompressionRate + 1 / l->AverageDCCCompressionRate)); - dml2_printf("DML::%s: min 3 = %d\n", __func__, (p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64)); - dml2_printf("DML::%s: min 4 = %f\n", __func__, (p->ZeroSizeBufferEntries - p->CompbufReservedSpaceZs) * 64 / (l->AverageDCCZeroSizeFraction / l->AverageZeroSizeCompressionRate)); + DML_LOG_VERBOSE("DML::%s: min 1 = %f\n", __func__, p->CompressedBufferSizeInkByte * 1024 * l->AverageDCCCompressionRate); + DML_LOG_VERBOSE("DML::%s: min 2 = %f\n", __func__, p->MetaFIFOSizeInKEntries * 1024 * 64 / (l->AverageDCCZeroSizeFraction / l->AverageZeroSizeCompressionRate + 1 / l->AverageDCCCompressionRate)); + DML_LOG_VERBOSE("DML::%s: min 3 = %d\n", __func__, (p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64)); + DML_LOG_VERBOSE("DML::%s: min 4 = %f\n", __func__, (p->ZeroSizeBufferEntries - p->CompbufReservedSpaceZs) * 64 / (l->AverageDCCZeroSizeFraction / l->AverageZeroSizeCompressionRate)); #endif } else { l->EffectiveCompressedBufferSize = math_min2((double)p->CompressedBufferSizeInkByte * 1024 * l->AverageDCCCompressionRate, @@ -10200,16 +10138,16 @@ static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratc ((double)p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64) * (p->rob_alloc_compressed ? l->AverageDCCCompressionRate : 1.0); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: min 1 = %f\n", __func__, p->CompressedBufferSizeInkByte * 1024 * l->AverageDCCCompressionRate); - dml2_printf("DML::%s: min 2 = %f\n", __func__, p->MetaFIFOSizeInKEntries * 1024 * 64 * l->AverageDCCCompressionRate); + DML_LOG_VERBOSE("DML::%s: min 1 = %f\n", __func__, p->CompressedBufferSizeInkByte * 1024 * l->AverageDCCCompressionRate); + DML_LOG_VERBOSE("DML::%s: min 2 = %f\n", __func__, p->MetaFIFOSizeInKEntries * 1024 * 64 * l->AverageDCCCompressionRate); #endif } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: MetaFIFOSizeInKEntries = %u\n", __func__, p->MetaFIFOSizeInKEntries); - dml2_printf("DML::%s: ZeroSizeBufferEntries = %u\n", __func__, p->ZeroSizeBufferEntries); - dml2_printf("DML::%s: AverageZeroSizeCompressionRate = %f\n", __func__, l->AverageZeroSizeCompressionRate); - dml2_printf("DML::%s: EffectiveCompressedBufferSize = %f (%f kbytes)\n", __func__, l->EffectiveCompressedBufferSize, l->EffectiveCompressedBufferSize / 1024.0); + DML_LOG_VERBOSE("DML::%s: MetaFIFOSizeInKEntries = %u\n", __func__, p->MetaFIFOSizeInKEntries); + DML_LOG_VERBOSE("DML::%s: ZeroSizeBufferEntries = %u\n", __func__, p->ZeroSizeBufferEntries); + DML_LOG_VERBOSE("DML::%s: AverageZeroSizeCompressionRate = %f\n", __func__, l->AverageZeroSizeCompressionRate); + DML_LOG_VERBOSE("DML::%s: EffectiveCompressedBufferSize = %f (%f kbytes)\n", __func__, l->EffectiveCompressedBufferSize, l->EffectiveCompressedBufferSize / 1024.0); #endif *p->StutterPeriod = 0; @@ -10220,15 +10158,15 @@ static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratc l->LinesInDETYRoundedDownToSwath = math_floor2(l->LinesInDETY, p->SwathHeightY[k]); l->DETBufferingTimeY = l->LinesInDETYRoundedDownToSwath * ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, DETBufferSizeY = %u (%u kbytes)\n", __func__, k, p->DETBufferSizeY[k], p->DETBufferSizeY[k] / 1024); - dml2_printf("DML::%s: k=%u, BytePerPixelDETY = %f\n", __func__, k, p->BytePerPixelDETY[k]); - dml2_printf("DML::%s: k=%u, SwathWidthY = %u\n", __func__, k, p->SwathWidthY[k]); - dml2_printf("DML::%s: k=%u, ReadBandwidthSurfaceLuma = %f\n", __func__, k, p->ReadBandwidthSurfaceLuma[k]); - dml2_printf("DML::%s: k=%u, TotalDataReadBandwidth = %f\n", __func__, k, p->TotalDataReadBandwidth); - dml2_printf("DML::%s: k=%u, LinesInDETY = %f\n", __func__, k, l->LinesInDETY); - dml2_printf("DML::%s: k=%u, LinesInDETYRoundedDownToSwath = %f\n", __func__, k, l->LinesInDETYRoundedDownToSwath); - dml2_printf("DML::%s: k=%u, VRatio = %f\n", __func__, k, p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); - dml2_printf("DML::%s: k=%u, DETBufferingTimeY = %f\n", __func__, k, l->DETBufferingTimeY); + DML_LOG_VERBOSE("DML::%s: k=%u, DETBufferSizeY = %u (%u kbytes)\n", __func__, k, p->DETBufferSizeY[k], p->DETBufferSizeY[k] / 1024); + DML_LOG_VERBOSE("DML::%s: k=%u, BytePerPixelDETY = %f\n", __func__, k, p->BytePerPixelDETY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, SwathWidthY = %u\n", __func__, k, p->SwathWidthY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, ReadBandwidthSurfaceLuma = %f\n", __func__, k, p->ReadBandwidthSurfaceLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, TotalDataReadBandwidth = %f\n", __func__, k, p->TotalDataReadBandwidth); + DML_LOG_VERBOSE("DML::%s: k=%u, LinesInDETY = %f\n", __func__, k, l->LinesInDETY); + DML_LOG_VERBOSE("DML::%s: k=%u, LinesInDETYRoundedDownToSwath = %f\n", __func__, k, l->LinesInDETYRoundedDownToSwath); + DML_LOG_VERBOSE("DML::%s: k=%u, VRatio = %f\n", __func__, k, p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u, DETBufferingTimeY = %f\n", __func__, k, l->DETBufferingTimeY); #endif if (!FoundCriticalSurface || l->DETBufferingTimeY < *p->StutterPeriod) { @@ -10248,17 +10186,17 @@ static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratc l->SinglePipeCriticalSurface = (p->DPPPerSurface[k] == 1); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, FoundCriticalSurface = %u\n", __func__, k, FoundCriticalSurface); - dml2_printf("DML::%s: k=%u, StutterPeriod = %f\n", __func__, k, *p->StutterPeriod); - dml2_printf("DML::%s: k=%u, MinTTUVBlankCriticalSurface = %f\n", __func__, k, l->MinTTUVBlankCriticalSurface); - dml2_printf("DML::%s: k=%u, FrameTimeCriticalSurface= %f\n", __func__, k, l->FrameTimeCriticalSurface); - dml2_printf("DML::%s: k=%u, VActiveTimeCriticalSurface = %f\n", __func__, k, l->VActiveTimeCriticalSurface); - dml2_printf("DML::%s: k=%u, BytePerPixelYCriticalSurface = %u\n", __func__, k, l->BytePerPixelYCriticalSurface); - dml2_printf("DML::%s: k=%u, SwathWidthYCriticalSurface = %f\n", __func__, k, l->SwathWidthYCriticalSurface); - dml2_printf("DML::%s: k=%u, SwathHeightYCriticalSurface = %f\n", __func__, k, l->SwathHeightYCriticalSurface); - dml2_printf("DML::%s: k=%u, BlockWidth256BytesYCriticalSurface = %u\n", __func__, k, l->BlockWidth256BytesYCriticalSurface); - dml2_printf("DML::%s: k=%u, SinglePlaneCriticalSurface = %u\n", __func__, k, l->SinglePlaneCriticalSurface); - dml2_printf("DML::%s: k=%u, SinglePipeCriticalSurface = %u\n", __func__, k, l->SinglePipeCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, FoundCriticalSurface = %u\n", __func__, k, FoundCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, StutterPeriod = %f\n", __func__, k, *p->StutterPeriod); + DML_LOG_VERBOSE("DML::%s: k=%u, MinTTUVBlankCriticalSurface = %f\n", __func__, k, l->MinTTUVBlankCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, FrameTimeCriticalSurface= %f\n", __func__, k, l->FrameTimeCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, VActiveTimeCriticalSurface = %f\n", __func__, k, l->VActiveTimeCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, BytePerPixelYCriticalSurface = %u\n", __func__, k, l->BytePerPixelYCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, SwathWidthYCriticalSurface = %f\n", __func__, k, l->SwathWidthYCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, SwathHeightYCriticalSurface = %f\n", __func__, k, l->SwathHeightYCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, BlockWidth256BytesYCriticalSurface = %u\n", __func__, k, l->BlockWidth256BytesYCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, SinglePlaneCriticalSurface = %u\n", __func__, k, l->SinglePlaneCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, SinglePipeCriticalSurface = %u\n", __func__, k, l->SinglePipeCriticalSurface); #endif } } @@ -10276,14 +10214,14 @@ static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratc l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer = math_min2(*p->StutterPeriod * p->TotalDataReadBandwidth, l->EffectiveCompressedBufferSize); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: AverageDCCCompressionRate = %f\n", __func__, l->AverageDCCCompressionRate); - dml2_printf("DML::%s: StutterPeriod*TotalDataReadBandwidth = %f (%f kbytes)\n", __func__, *p->StutterPeriod * p->TotalDataReadBandwidth, (*p->StutterPeriod * p->TotalDataReadBandwidth) / 1024.0); - dml2_printf("DML::%s: EffectiveCompressedBufferSize = %f (%f kbytes)\n", __func__, l->EffectiveCompressedBufferSize, l->EffectiveCompressedBufferSize / 1024.0); - dml2_printf("DML::%s: PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer = %f (%f kbytes)\n", __func__, l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer, l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer / 1024); - dml2_printf("DML::%s: ReturnBW = %f\n", __func__, p->ReturnBW); - dml2_printf("DML::%s: TotalDataReadBandwidth = %f\n", __func__, p->TotalDataReadBandwidth); - dml2_printf("DML::%s: TotalRowReadBandwidth = %f\n", __func__, l->TotalRowReadBandwidth); - dml2_printf("DML::%s: DCFCLK = %f\n", __func__, p->DCFCLK); + DML_LOG_VERBOSE("DML::%s: AverageDCCCompressionRate = %f\n", __func__, l->AverageDCCCompressionRate); + DML_LOG_VERBOSE("DML::%s: StutterPeriod*TotalDataReadBandwidth = %f (%f kbytes)\n", __func__, *p->StutterPeriod * p->TotalDataReadBandwidth, (*p->StutterPeriod * p->TotalDataReadBandwidth) / 1024.0); + DML_LOG_VERBOSE("DML::%s: EffectiveCompressedBufferSize = %f (%f kbytes)\n", __func__, l->EffectiveCompressedBufferSize, l->EffectiveCompressedBufferSize / 1024.0); + DML_LOG_VERBOSE("DML::%s: PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer = %f (%f kbytes)\n", __func__, l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer, l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer / 1024); + DML_LOG_VERBOSE("DML::%s: ReturnBW = %f\n", __func__, p->ReturnBW); + DML_LOG_VERBOSE("DML::%s: TotalDataReadBandwidth = %f\n", __func__, p->TotalDataReadBandwidth); + DML_LOG_VERBOSE("DML::%s: TotalRowReadBandwidth = %f\n", __func__, l->TotalRowReadBandwidth); + DML_LOG_VERBOSE("DML::%s: DCFCLK = %f\n", __func__, p->DCFCLK); #endif l->StutterBurstTime = l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer @@ -10292,10 +10230,10 @@ static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratc / math_min2(p->DCFCLK * 64, p->ReturnBW * (p->hw_debug5 ? 1 : l->AverageDCCCompressionRate)) + *p->StutterPeriod * l->TotalRowReadBandwidth / p->ReturnBW; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: Part 1 = %f\n", __func__, l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer / p->ReturnBW / (p->hw_debug5 ? 1 : l->AverageDCCCompressionRate)); - dml2_printf("DML::%s: Part 2 = %f\n", __func__, (*p->StutterPeriod * p->TotalDataReadBandwidth - l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer) / (p->DCFCLK * 64)); - dml2_printf("DML::%s: Part 3 = %f\n", __func__, *p->StutterPeriod * l->TotalRowReadBandwidth / p->ReturnBW); - dml2_printf("DML::%s: StutterBurstTime = %f\n", __func__, l->StutterBurstTime); + DML_LOG_VERBOSE("DML::%s: Part 1 = %f\n", __func__, l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer / p->ReturnBW / (p->hw_debug5 ? 1 : l->AverageDCCCompressionRate)); + DML_LOG_VERBOSE("DML::%s: Part 2 = %f\n", __func__, (*p->StutterPeriod * p->TotalDataReadBandwidth - l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer) / (p->DCFCLK * 64)); + DML_LOG_VERBOSE("DML::%s: Part 3 = %f\n", __func__, *p->StutterPeriod * l->TotalRowReadBandwidth / p->ReturnBW); + DML_LOG_VERBOSE("DML::%s: StutterBurstTime = %f\n", __func__, l->StutterBurstTime); #endif l->TotalActiveWriteback = 0; memset(l->stream_visited, 0, DML2_MAX_PLANES * sizeof(bool)); @@ -10324,9 +10262,9 @@ static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratc if (l->TotalActiveWriteback == 0) { #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: SRExitTime = %f\n", __func__, p->SRExitTime); - dml2_printf("DML::%s: SRExitZ8Time = %f\n", __func__, p->SRExitZ8Time); - dml2_printf("DML::%s: StutterPeriod = %f\n", __func__, *p->StutterPeriod); + DML_LOG_VERBOSE("DML::%s: SRExitTime = %f\n", __func__, p->SRExitTime); + DML_LOG_VERBOSE("DML::%s: SRExitZ8Time = %f\n", __func__, p->SRExitZ8Time); + DML_LOG_VERBOSE("DML::%s: StutterPeriod = %f\n", __func__, *p->StutterPeriod); #endif *p->StutterEfficiencyNotIncludingVBlank = math_max2(0., 1 - (p->SRExitTime + l->StutterBurstTime) / *p->StutterPeriod) * 100; *p->Z8StutterEfficiencyNotIncludingVBlank = math_max2(0., 1 - (p->SRExitZ8Time + l->StutterBurstTime) / *p->StutterPeriod) * 100; @@ -10339,11 +10277,11 @@ static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratc *p->Z8NumberOfStutterBurstsPerFrame = 0; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: VActiveTimeCriticalSurface = %f\n", __func__, l->VActiveTimeCriticalSurface); - dml2_printf("DML::%s: StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->StutterEfficiencyNotIncludingVBlank); - dml2_printf("DML::%s: Z8StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->Z8StutterEfficiencyNotIncludingVBlank); - dml2_printf("DML::%s: NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->NumberOfStutterBurstsPerFrame); - dml2_printf("DML::%s: Z8NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->Z8NumberOfStutterBurstsPerFrame); + DML_LOG_VERBOSE("DML::%s: VActiveTimeCriticalSurface = %f\n", __func__, l->VActiveTimeCriticalSurface); + DML_LOG_VERBOSE("DML::%s: StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->StutterEfficiencyNotIncludingVBlank); + DML_LOG_VERBOSE("DML::%s: Z8StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->Z8StutterEfficiencyNotIncludingVBlank); + DML_LOG_VERBOSE("DML::%s: NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->NumberOfStutterBurstsPerFrame); + DML_LOG_VERBOSE("DML::%s: Z8NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->Z8NumberOfStutterBurstsPerFrame); #endif if (*p->StutterEfficiencyNotIncludingVBlank > 0) { @@ -10358,7 +10296,7 @@ static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratc } if (*p->Z8StutterEfficiencyNotIncludingVBlank > 0) { - LastZ8StutterPeriod = l->VActiveTimeCriticalSurface - (*p->Z8NumberOfStutterBurstsPerFrame - 1) * *p->StutterPeriod; + //LastZ8StutterPeriod = l->VActiveTimeCriticalSurface - (*p->Z8NumberOfStutterBurstsPerFrame - 1) * *p->StutterPeriod; if (!((p->SynchronizeTimings || TotalNumberOfActiveOTG == 1) && SameTiming)) { *p->Z8StutterEfficiency = *p->Z8StutterEfficiencyNotIncludingVBlank; } else { @@ -10370,25 +10308,25 @@ static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratc } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: TotalNumberOfActiveOTG = %u\n", __func__, TotalNumberOfActiveOTG); - dml2_printf("DML::%s: SameTiming = %u\n", __func__, SameTiming); - dml2_printf("DML::%s: SynchronizeTimings = %u\n", __func__, p->SynchronizeTimings); - dml2_printf("DML::%s: LastZ8StutterPeriod = %f\n", __func__, LastZ8StutterPeriod); - dml2_printf("DML::%s: Z8StutterEnterPlusExitWatermark = %f\n", __func__, p->Z8StutterEnterPlusExitWatermark); - dml2_printf("DML::%s: StutterBurstTime = %f\n", __func__, l->StutterBurstTime); - dml2_printf("DML::%s: StutterPeriod = %f\n", __func__, *p->StutterPeriod); - dml2_printf("DML::%s: StutterEfficiency = %f\n", __func__, *p->StutterEfficiency); - dml2_printf("DML::%s: Z8StutterEfficiency = %f\n", __func__, *p->Z8StutterEfficiency); - dml2_printf("DML::%s: StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->StutterEfficiencyNotIncludingVBlank); - dml2_printf("DML::%s: Z8NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->Z8NumberOfStutterBurstsPerFrame); + DML_LOG_VERBOSE("DML::%s: TotalNumberOfActiveOTG = %u\n", __func__, TotalNumberOfActiveOTG); + DML_LOG_VERBOSE("DML::%s: SameTiming = %u\n", __func__, SameTiming); + DML_LOG_VERBOSE("DML::%s: SynchronizeTimings = %u\n", __func__, p->SynchronizeTimings); + DML_LOG_VERBOSE("DML::%s: LastZ8StutterPeriod = %f\n", __func__, *p->Z8StutterEfficiencyNotIncludingVBlank > 0 ? l->VActiveTimeCriticalSurface - (*p->Z8NumberOfStutterBurstsPerFrame - 1) * *p->StutterPeriod : 0); + DML_LOG_VERBOSE("DML::%s: Z8StutterEnterPlusExitWatermark = %f\n", __func__, p->Z8StutterEnterPlusExitWatermark); + DML_LOG_VERBOSE("DML::%s: StutterBurstTime = %f\n", __func__, l->StutterBurstTime); + DML_LOG_VERBOSE("DML::%s: StutterPeriod = %f\n", __func__, *p->StutterPeriod); + DML_LOG_VERBOSE("DML::%s: StutterEfficiency = %f\n", __func__, *p->StutterEfficiency); + DML_LOG_VERBOSE("DML::%s: Z8StutterEfficiency = %f\n", __func__, *p->Z8StutterEfficiency); + DML_LOG_VERBOSE("DML::%s: StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->StutterEfficiencyNotIncludingVBlank); + DML_LOG_VERBOSE("DML::%s: Z8NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->Z8NumberOfStutterBurstsPerFrame); #endif *p->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE = !(!p->UnboundedRequestEnabled && (p->NumberOfActiveSurfaces == 1) && l->SinglePlaneCriticalSurface && l->SinglePipeCriticalSurface); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: DETBufferSizeYCriticalSurface = %u\n", __func__, l->DETBufferSizeYCriticalSurface); - dml2_printf("DML::%s: PixelChunkSizeInKByte = %u\n", __func__, p->PixelChunkSizeInKByte); - dml2_printf("DML::%s: DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE = %u\n", __func__, *p->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE); + DML_LOG_VERBOSE("DML::%s: DETBufferSizeYCriticalSurface = %u\n", __func__, l->DETBufferSizeYCriticalSurface); + DML_LOG_VERBOSE("DML::%s: PixelChunkSizeInKByte = %u\n", __func__, p->PixelChunkSizeInKByte); + DML_LOG_VERBOSE("DML::%s: DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE = %u\n", __func__, *p->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE); #endif } @@ -10422,7 +10360,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex double max_uclk_mhz = 0; double min_return_latency_in_DCFCLK_cycles = 0; - dml2_printf("DML::%s: --- START --- \n", __func__); + DML_LOG_VERBOSE("DML::%s: --- START --- \n", __func__); memset(&mode_lib->scratch, 0, sizeof(struct dml2_core_internal_scratch)); memset(&mode_lib->mp, 0, sizeof(struct dml2_core_internal_mode_program)); @@ -10444,13 +10382,13 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex for (k = 0; k < s->num_active_planes; ++k) { unsigned int stream_index = display_cfg->plane_descriptors[k].stream_index; - DML2_ASSERT(cfg_support_info->stream_support_info[stream_index].odms_used <= 4); - DML2_ASSERT(cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 4 || + DML_ASSERT(cfg_support_info->stream_support_info[stream_index].odms_used <= 4); + DML_ASSERT(cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 4 || cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 2 || cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 1); if (cfg_support_info->stream_support_info[stream_index].odms_used > 1) - DML2_ASSERT(cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 1); + DML_ASSERT(cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 1); switch (cfg_support_info->stream_support_info[stream_index].odms_used) { case (4): @@ -10476,51 +10414,51 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex for (k = 0; k < s->num_active_planes; ++k) { mode_lib->mp.NoOfDPP[k] = cfg_support_info->plane_support_info[k].dpps_used; mode_lib->mp.Dppclk[k] = programming->plane_programming[k].min_clocks.dcn4x.dppclk_khz / 1000.0; - DML2_ASSERT(mode_lib->mp.Dppclk[k] > 0); + DML_ASSERT(mode_lib->mp.Dppclk[k] > 0); } for (k = 0; k < s->num_active_planes; ++k) { unsigned int stream_index = display_cfg->plane_descriptors[k].stream_index; mode_lib->mp.DSCCLK[k] = programming->stream_programming[stream_index].min_clocks.dcn4x.dscclk_khz / 1000.0; - dml2_printf("DML::%s: k=%d stream_index=%d, mode_lib->mp.DSCCLK = %f\n", __func__, k, stream_index, mode_lib->mp.DSCCLK[k]); + DML_LOG_VERBOSE("DML::%s: k=%d stream_index=%d, mode_lib->mp.DSCCLK = %f\n", __func__, k, stream_index, mode_lib->mp.DSCCLK[k]); } mode_lib->mp.Dispclk = programming->min_clocks.dcn4x.dispclk_khz / 1000.0; mode_lib->mp.DCFCLKDeepSleep = programming->min_clocks.dcn4x.deepsleep_dcfclk_khz / 1000.0; - DML2_ASSERT(mode_lib->mp.Dcfclk > 0); - DML2_ASSERT(mode_lib->mp.FabricClock > 0); - DML2_ASSERT(mode_lib->mp.dram_bw_mbps > 0); - DML2_ASSERT(mode_lib->mp.uclk_freq_mhz > 0); - DML2_ASSERT(mode_lib->mp.GlobalDPPCLK > 0); - DML2_ASSERT(mode_lib->mp.Dispclk > 0); - DML2_ASSERT(mode_lib->mp.DCFCLKDeepSleep > 0); - DML2_ASSERT(s->SOCCLK > 0); - -#ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: num_active_planes = %u\n", __func__, s->num_active_planes); - dml2_printf("DML::%s: num_active_pipes = %u\n", __func__, mode_lib->mp.num_active_pipes); - dml2_printf("DML::%s: Dcfclk = %f\n", __func__, mode_lib->mp.Dcfclk); - dml2_printf("DML::%s: FabricClock = %f\n", __func__, mode_lib->mp.FabricClock); - dml2_printf("DML::%s: dram_bw_mbps = %f\n", __func__, mode_lib->mp.dram_bw_mbps); - dml2_printf("DML::%s: uclk_freq_mhz = %f\n", __func__, mode_lib->mp.uclk_freq_mhz); - dml2_printf("DML::%s: Dispclk = %f\n", __func__, mode_lib->mp.Dispclk); + DML_ASSERT(mode_lib->mp.Dcfclk > 0); + DML_ASSERT(mode_lib->mp.FabricClock > 0); + DML_ASSERT(mode_lib->mp.dram_bw_mbps > 0); + DML_ASSERT(mode_lib->mp.uclk_freq_mhz > 0); + DML_ASSERT(mode_lib->mp.GlobalDPPCLK > 0); + DML_ASSERT(mode_lib->mp.Dispclk > 0); + DML_ASSERT(mode_lib->mp.DCFCLKDeepSleep > 0); + DML_ASSERT(s->SOCCLK > 0); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: num_active_planes = %u\n", __func__, s->num_active_planes); + DML_LOG_VERBOSE("DML::%s: num_active_pipes = %u\n", __func__, mode_lib->mp.num_active_pipes); + DML_LOG_VERBOSE("DML::%s: Dcfclk = %f\n", __func__, mode_lib->mp.Dcfclk); + DML_LOG_VERBOSE("DML::%s: FabricClock = %f\n", __func__, mode_lib->mp.FabricClock); + DML_LOG_VERBOSE("DML::%s: dram_bw_mbps = %f\n", __func__, mode_lib->mp.dram_bw_mbps); + DML_LOG_VERBOSE("DML::%s: uclk_freq_mhz = %f\n", __func__, mode_lib->mp.uclk_freq_mhz); + DML_LOG_VERBOSE("DML::%s: Dispclk = %f\n", __func__, mode_lib->mp.Dispclk); for (k = 0; k < s->num_active_planes; ++k) { - dml2_printf("DML::%s: Dppclk[%0d] = %f\n", __func__, k, mode_lib->mp.Dppclk[k]); - } - dml2_printf("DML::%s: GlobalDPPCLK = %f\n", __func__, mode_lib->mp.GlobalDPPCLK); - dml2_printf("DML::%s: DCFCLKDeepSleep = %f\n", __func__, mode_lib->mp.DCFCLKDeepSleep); - dml2_printf("DML::%s: SOCCLK = %f\n", __func__, s->SOCCLK); - dml2_printf("DML::%s: min_clk_index = %0d\n", __func__, in_out_params->min_clk_index); - dml2_printf("DML::%s: min_clk_table min_fclk_khz = %d\n", __func__, min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].min_fclk_khz); - dml2_printf("DML::%s: min_clk_table uclk_mhz = %f\n", __func__, dram_bw_kbps_to_uclk_mhz(min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].pre_derate_dram_bw_kbps, &mode_lib->soc.clk_table.dram_config)); + DML_LOG_VERBOSE("DML::%s: Dppclk[%0d] = %f\n", __func__, k, mode_lib->mp.Dppclk[k]); + } + DML_LOG_VERBOSE("DML::%s: GlobalDPPCLK = %f\n", __func__, mode_lib->mp.GlobalDPPCLK); + DML_LOG_VERBOSE("DML::%s: DCFCLKDeepSleep = %f\n", __func__, mode_lib->mp.DCFCLKDeepSleep); + DML_LOG_VERBOSE("DML::%s: SOCCLK = %f\n", __func__, s->SOCCLK); + DML_LOG_VERBOSE("DML::%s: min_clk_index = %0d\n", __func__, in_out_params->min_clk_index); + DML_LOG_VERBOSE("DML::%s: min_clk_table min_fclk_khz = %ld\n", __func__, min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].min_fclk_khz); + DML_LOG_VERBOSE("DML::%s: min_clk_table uclk_mhz = %f\n", __func__, dram_bw_kbps_to_uclk_mhz(min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].pre_derate_dram_bw_kbps, &mode_lib->soc.clk_table.dram_config)); for (k = 0; k < mode_lib->mp.num_active_pipes; ++k) { - dml2_printf("DML::%s: pipe=%d is in plane=%d\n", __func__, k, mode_lib->mp.pipe_plane[k]); - dml2_printf("DML::%s: Per-plane DPPPerSurface[%0d] = %d\n", __func__, k, mode_lib->mp.NoOfDPP[k]); + DML_LOG_VERBOSE("DML::%s: pipe=%d is in plane=%d\n", __func__, k, mode_lib->mp.pipe_plane[k]); + DML_LOG_VERBOSE("DML::%s: Per-plane DPPPerSurface[%0d] = %d\n", __func__, k, mode_lib->mp.NoOfDPP[k]); } for (k = 0; k < s->num_active_planes; k++) - dml2_printf("DML::%s: plane_%d: reserved_vblank_time_ns = %u\n", __func__, k, display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns); + DML_LOG_VERBOSE("DML::%s: plane_%d: reserved_vblank_time_ns = %lu\n", __func__, k, display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns); #endif CalculateMaxDETAndMinCompressedBufferSize( @@ -10617,8 +10555,8 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)); mode_lib->mp.vactive_sw_bw_l[k] = mode_lib->mp.SwathWidthSingleDPPY[k] * mode_lib->mp.BytePerPixelY[k] / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; mode_lib->mp.vactive_sw_bw_c[k] = mode_lib->mp.SwathWidthSingleDPPC[k] * mode_lib->mp.BytePerPixelC[k] / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; - dml2_printf("DML::%s: vactive_sw_bw_l[%i] = %fBps\n", __func__, k, mode_lib->mp.vactive_sw_bw_l[k]); - dml2_printf("DML::%s: vactive_sw_bw_c[%i] = %fBps\n", __func__, k, mode_lib->mp.vactive_sw_bw_c[k]); + DML_LOG_VERBOSE("DML::%s: vactive_sw_bw_l[%i] = %fBps\n", __func__, k, mode_lib->mp.vactive_sw_bw_l[k]); + DML_LOG_VERBOSE("DML::%s: vactive_sw_bw_c[%i] = %fBps\n", __func__, k, mode_lib->mp.vactive_sw_bw_c[k]); } CalculateSwathAndDETConfiguration_params->display_cfg = display_cfg; @@ -11097,7 +11035,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_max_transport_latency_margin); for (k = 0; k < s->num_active_planes; ++k) { - bool cursor_not_enough_urgent_latency_hiding = 0; + bool cursor_not_enough_urgent_latency_hiding = false; s->line_times[k] = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); @@ -11173,8 +11111,8 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex mode_lib->mp.WritebackDelay[k]); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); - dml2_printf("DML::%s: k=%u WritebackDelay = %f\n", __func__, k, mode_lib->mp.WritebackDelay[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); + DML_LOG_VERBOSE("DML::%s: k=%u WritebackDelay = %f\n", __func__, k, mode_lib->mp.WritebackDelay[k]); #endif } @@ -11183,7 +11121,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex s->immediate_flip_required = s->immediate_flip_required || display_cfg->plane_descriptors[k].immediate_flip; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: immediate_flip_required = %u\n", __func__, s->immediate_flip_required); + DML_LOG_VERBOSE("DML::%s: immediate_flip_required = %u\n", __func__, s->immediate_flip_required); #endif if (s->num_active_planes > 1) { @@ -11219,12 +11157,12 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex s->DestinationLineTimesForPrefetchLessThan2 = false; s->VRatioPrefetchMoreThanMax = false; - dml2_printf("DML::%s: Start one iteration of prefetch schedule evaluation\n", __func__); + DML_LOG_VERBOSE("DML::%s: Start one iteration of prefetch schedule evaluation\n", __func__); for (k = 0; k < s->num_active_planes; ++k) { struct dml2_core_internal_DmlPipe *myPipe = &s->myPipe; - dml2_printf("DML::%s: k=%d MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); + DML_LOG_VERBOSE("DML::%s: k=%d MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); mode_lib->mp.TWait[k] = CalculateTWait( display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns, mode_lib->mp.UrgentLatency, @@ -11261,7 +11199,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex myPipe->ProgressiveToInterlaceUnitInOPP = mode_lib->ip.ptoi_supported; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: Calling CalculatePrefetchSchedule for k=%u\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: Calling CalculatePrefetchSchedule for k=%u\n", __func__, k); #endif CalculatePrefetchSchedule_params->display_cfg = display_cfg; CalculatePrefetchSchedule_params->HostVMInefficiencyFactor = s->HostVMInefficiencyFactorPrefetch; @@ -11325,7 +11263,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex CalculatePrefetchSchedule_params->VRatioPrefetchC = &mode_lib->mp.VRatioPrefetchC[k]; CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWLuma = &mode_lib->mp.RequiredPrefetchPixelDataBWLuma[k]; CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWChroma = &mode_lib->mp.RequiredPrefetchPixelDataBWChroma[k]; - CalculatePrefetchSchedule_params->RequiredPrefetchBWOTO = &s->dummy_single_array[0][k]; + CalculatePrefetchSchedule_params->RequiredPrefetchBWMax = &s->dummy_single_array[0][k]; CalculatePrefetchSchedule_params->NotEnoughTimeForDynamicMetadata = &mode_lib->mp.NotEnoughTimeForDynamicMetadata[k]; CalculatePrefetchSchedule_params->Tno_bw = &mode_lib->mp.Tno_bw[k]; CalculatePrefetchSchedule_params->Tno_bw_flip = &mode_lib->mp.Tno_bw_flip[k]; @@ -11356,7 +11294,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex mode_lib->mp.impacted_prefetch_margin_us[k] = 0; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%0u NoTimeToPrefetch=%0d\n", __func__, k, mode_lib->mp.NoTimeToPrefetch[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u NoTimeToPrefetch=%0d\n", __func__, k, mode_lib->mp.NoTimeToPrefetch[k]); #endif mode_lib->mp.VStartupMin[k] = s->MaxVStartupLines[k]; } // for k @@ -11366,9 +11304,9 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex if (mode_lib->mp.NoTimeToPrefetch[k] == true || mode_lib->mp.NotEnoughTimeForDynamicMetadata[k] || mode_lib->mp.DSTYAfterScaler[k] > 8) { - dml2_printf("DML::%s: k=%u, NoTimeToPrefetch = %0d\n", __func__, k, mode_lib->mp.NoTimeToPrefetch[k]); - dml2_printf("DML::%s: k=%u, NotEnoughTimeForDynamicMetadata=%u\n", __func__, k, mode_lib->mp.NotEnoughTimeForDynamicMetadata[k]); - dml2_printf("DML::%s: k=%u, DSTYAfterScaler=%u (should be <= 0)\n", __func__, k, mode_lib->mp.DSTYAfterScaler[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, NoTimeToPrefetch = %0d\n", __func__, k, mode_lib->mp.NoTimeToPrefetch[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, NotEnoughTimeForDynamicMetadata=%u\n", __func__, k, mode_lib->mp.NotEnoughTimeForDynamicMetadata[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, DSTYAfterScaler=%u (should be <= 0)\n", __func__, k, mode_lib->mp.DSTYAfterScaler[k]); mode_lib->mp.PrefetchModeSupported = false; } if (mode_lib->mp.dst_y_prefetch[k] < 2) @@ -11377,24 +11315,24 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex if (mode_lib->mp.VRatioPrefetchY[k] > __DML2_CALCS_MAX_VRATIO_PRE__ || mode_lib->mp.VRatioPrefetchC[k] > __DML2_CALCS_MAX_VRATIO_PRE__) { s->VRatioPrefetchMoreThanMax = true; - dml2_printf("DML::%s: k=%d, VRatioPrefetchY=%f (should not be < %f)\n", __func__, k, mode_lib->mp.VRatioPrefetchY[k], __DML2_CALCS_MAX_VRATIO_PRE__); - dml2_printf("DML::%s: k=%d, VRatioPrefetchC=%f (should not be < %f)\n", __func__, k, mode_lib->mp.VRatioPrefetchC[k], __DML2_CALCS_MAX_VRATIO_PRE__); - dml2_printf("DML::%s: VRatioPrefetchMoreThanMax = %u\n", __func__, s->VRatioPrefetchMoreThanMax); + DML_LOG_VERBOSE("DML::%s: k=%d, VRatioPrefetchY=%f (should not be < %f)\n", __func__, k, mode_lib->mp.VRatioPrefetchY[k], __DML2_CALCS_MAX_VRATIO_PRE__); + DML_LOG_VERBOSE("DML::%s: k=%d, VRatioPrefetchC=%f (should not be < %f)\n", __func__, k, mode_lib->mp.VRatioPrefetchC[k], __DML2_CALCS_MAX_VRATIO_PRE__); + DML_LOG_VERBOSE("DML::%s: VRatioPrefetchMoreThanMax = %u\n", __func__, s->VRatioPrefetchMoreThanMax); } if (mode_lib->mp.NotEnoughUrgentLatencyHiding[k]) { - dml2_printf("DML::%s: k=%u, NotEnoughUrgentLatencyHiding = %u\n", __func__, k, mode_lib->mp.NotEnoughUrgentLatencyHiding[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, NotEnoughUrgentLatencyHiding = %u\n", __func__, k, mode_lib->mp.NotEnoughUrgentLatencyHiding[k]); mode_lib->mp.PrefetchModeSupported = false; } } if (s->VRatioPrefetchMoreThanMax == true || s->DestinationLineTimesForPrefetchLessThan2 == true) { - dml2_printf("DML::%s: VRatioPrefetchMoreThanMax = %u\n", __func__, s->VRatioPrefetchMoreThanMax); - dml2_printf("DML::%s: DestinationLineTimesForPrefetchLessThan2 = %u\n", __func__, s->DestinationLineTimesForPrefetchLessThan2); + DML_LOG_VERBOSE("DML::%s: VRatioPrefetchMoreThanMax = %u\n", __func__, s->VRatioPrefetchMoreThanMax); + DML_LOG_VERBOSE("DML::%s: DestinationLineTimesForPrefetchLessThan2 = %u\n", __func__, s->DestinationLineTimesForPrefetchLessThan2); mode_lib->mp.PrefetchModeSupported = false; } - dml2_printf("DML::%s: Prefetch schedule is %sOK at vstartup = %u\n", __func__, + DML_LOG_VERBOSE("DML::%s: Prefetch schedule is %sOK at vstartup = %u\n", __func__, mode_lib->mp.PrefetchModeSupported ? "" : "NOT ", CalculatePrefetchSchedule_params->VStartup); // Prefetch schedule OK, now check prefetch bw @@ -11422,24 +11360,24 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex &mode_lib->mp.NotEnoughUrgentLatencyHidingPre[k]); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%0u DPPPerSurface=%u\n", __func__, k, mode_lib->mp.NoOfDPP[k]); - dml2_printf("DML::%s: k=%0u UrgentBurstFactorLuma=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorLuma[k]); - dml2_printf("DML::%s: k=%0u UrgentBurstFactorChroma=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorChroma[k]); - dml2_printf("DML::%s: k=%0u UrgentBurstFactorLumaPre=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorLumaPre[k]); - dml2_printf("DML::%s: k=%0u UrgentBurstFactorChromaPre=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorChromaPre[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u DPPPerSurface=%u\n", __func__, k, mode_lib->mp.NoOfDPP[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u UrgentBurstFactorLuma=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u UrgentBurstFactorChroma=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u UrgentBurstFactorLumaPre=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorLumaPre[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u UrgentBurstFactorChromaPre=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorChromaPre[k]); - dml2_printf("DML::%s: k=%0u VRatioPrefetchY=%f\n", __func__, k, mode_lib->mp.VRatioPrefetchY[k]); - dml2_printf("DML::%s: k=%0u VRatioY=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); + DML_LOG_VERBOSE("DML::%s: k=%0u VRatioPrefetchY=%f\n", __func__, k, mode_lib->mp.VRatioPrefetchY[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u VRatioY=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); - dml2_printf("DML::%s: k=%0u prefetch_vmrow_bw=%f\n", __func__, k, mode_lib->mp.prefetch_vmrow_bw[k]); - dml2_printf("DML::%s: k=%0u vactive_sw_bw_l=%f\n", __func__, k, mode_lib->mp.vactive_sw_bw_l[k]); - dml2_printf("DML::%s: k=%0u vactive_sw_bw_c=%f\n", __func__, k, mode_lib->mp.vactive_sw_bw_c[k]); - dml2_printf("DML::%s: k=%0u cursor_bw=%f\n", __func__, k, mode_lib->mp.cursor_bw[k]); - dml2_printf("DML::%s: k=%0u dpte_row_bw=%f\n", __func__, k, mode_lib->mp.dpte_row_bw[k]); - dml2_printf("DML::%s: k=%0u meta_row_bw=%f\n", __func__, k, mode_lib->mp.meta_row_bw[k]); - dml2_printf("DML::%s: k=%0u RequiredPrefetchPixelDataBWLuma=%f\n", __func__, k, mode_lib->mp.RequiredPrefetchPixelDataBWLuma[k]); - dml2_printf("DML::%s: k=%0u RequiredPrefetchPixelDataBWChroma=%f\n", __func__, k, mode_lib->mp.RequiredPrefetchPixelDataBWChroma[k]); - dml2_printf("DML::%s: k=%0u prefetch_cursor_bw=%f\n", __func__, k, mode_lib->mp.prefetch_cursor_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u prefetch_vmrow_bw=%f\n", __func__, k, mode_lib->mp.prefetch_vmrow_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u vactive_sw_bw_l=%f\n", __func__, k, mode_lib->mp.vactive_sw_bw_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u vactive_sw_bw_c=%f\n", __func__, k, mode_lib->mp.vactive_sw_bw_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u cursor_bw=%f\n", __func__, k, mode_lib->mp.cursor_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u dpte_row_bw=%f\n", __func__, k, mode_lib->mp.dpte_row_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u meta_row_bw=%f\n", __func__, k, mode_lib->mp.meta_row_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u RequiredPrefetchPixelDataBWLuma=%f\n", __func__, k, mode_lib->mp.RequiredPrefetchPixelDataBWLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u RequiredPrefetchPixelDataBWChroma=%f\n", __func__, k, mode_lib->mp.RequiredPrefetchPixelDataBWChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u prefetch_cursor_bw=%f\n", __func__, k, mode_lib->mp.prefetch_cursor_bw[k]); #endif } @@ -11468,7 +11406,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->mp.vactive_sw_bw_c; calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->mp.RequiredPrefetchPixelDataBWLuma; calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->mp.RequiredPrefetchPixelDataBWChroma; - calculate_peak_bandwidth_params->prefetch_bandwidth_oto = s->dummy_single_array[0]; + calculate_peak_bandwidth_params->prefetch_bandwidth_max = s->dummy_single_array[0]; calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->mp.excess_vactive_fill_bw_l; calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->mp.excess_vactive_fill_bw_c; calculate_peak_bandwidth_params->cursor_bw = mode_lib->mp.cursor_bw; @@ -11503,11 +11441,11 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex mode_lib->mp.urg_bandwidth_available); if (!mode_lib->mp.PrefetchModeSupported) - dml2_printf("DML::%s: Bandwidth not sufficient for prefetch!\n", __func__); + DML_LOG_VERBOSE("DML::%s: Bandwidth not sufficient for prefetch!\n", __func__); for (k = 0; k < s->num_active_planes; ++k) { if (mode_lib->mp.NotEnoughUrgentLatencyHidingPre[k]) { - dml2_printf("DML::%s: k=%u, NotEnoughUrgentLatencyHidingPre = %u\n", __func__, k, mode_lib->mp.NotEnoughUrgentLatencyHidingPre[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, NotEnoughUrgentLatencyHidingPre = %u\n", __func__, k, mode_lib->mp.NotEnoughUrgentLatencyHidingPre[k]); mode_lib->mp.PrefetchModeSupported = false; } } @@ -11533,12 +11471,12 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex } mode_lib->mp.TotImmediateFlipBytes += s->per_pipe_flip_bytes[k] * mode_lib->mp.NoOfDPP[k]; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k = %u\n", __func__, k); - dml2_printf("DML::%s: DPPPerSurface = %u\n", __func__, mode_lib->mp.NoOfDPP[k]); - dml2_printf("DML::%s: vm_bytes = %u\n", __func__, mode_lib->mp.vm_bytes[k]); - dml2_printf("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, mode_lib->mp.PixelPTEBytesPerRow[k]); - dml2_printf("DML::%s: meta_row_bytes = %u\n", __func__, mode_lib->mp.meta_row_bytes[k]); - dml2_printf("DML::%s: TotImmediateFlipBytes = %u\n", __func__, mode_lib->mp.TotImmediateFlipBytes); + DML_LOG_VERBOSE("DML::%s: k = %u\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: DPPPerSurface = %u\n", __func__, mode_lib->mp.NoOfDPP[k]); + DML_LOG_VERBOSE("DML::%s: vm_bytes = %u\n", __func__, mode_lib->mp.vm_bytes[k]); + DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, mode_lib->mp.PixelPTEBytesPerRow[k]); + DML_LOG_VERBOSE("DML::%s: meta_row_bytes = %u\n", __func__, mode_lib->mp.meta_row_bytes[k]); + DML_LOG_VERBOSE("DML::%s: TotImmediateFlipBytes = %u\n", __func__, mode_lib->mp.TotImmediateFlipBytes); #endif } for (k = 0; k < s->num_active_planes; ++k) { @@ -11608,7 +11546,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex calculate_peak_bandwidth_params->meta_row_bw = mode_lib->mp.meta_row_bw; calculate_peak_bandwidth_params->prefetch_cursor_bw = mode_lib->mp.prefetch_cursor_bw; calculate_peak_bandwidth_params->prefetch_vmrow_bw = mode_lib->mp.prefetch_vmrow_bw; - calculate_peak_bandwidth_params->prefetch_bandwidth_oto = s->dummy_single_array[0]; + calculate_peak_bandwidth_params->prefetch_bandwidth_max = s->dummy_single_array[0]; calculate_peak_bandwidth_params->flip_bw = mode_lib->mp.final_flip_bw; calculate_peak_bandwidth_params->urgent_burst_factor_l = mode_lib->mp.UrgentBurstFactorLuma; calculate_peak_bandwidth_params->urgent_burst_factor_c = mode_lib->mp.UrgentBurstFactorChroma; @@ -11631,13 +11569,13 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex mode_lib->mp.urg_bandwidth_available); if (!mode_lib->mp.ImmediateFlipSupported) - dml2_printf("DML::%s: Bandwidth not sufficient for flip!", __func__); + DML_LOG_VERBOSE("DML::%s: Bandwidth not sufficient for flip!", __func__); for (k = 0; k < s->num_active_planes; ++k) { if (display_cfg->plane_descriptors[k].immediate_flip && mode_lib->mp.ImmediateFlipSupportedForPipe[k] == false) { mode_lib->mp.ImmediateFlipSupported = false; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: Pipe %0d not supporting iflip!\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: Pipe %0d not supporting iflip!\n", __func__, k); #endif } } @@ -11650,28 +11588,28 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex mode_lib->mp.PrefetchAndImmediateFlipSupported = (mode_lib->mp.PrefetchModeSupported == true && (!must_support_iflip || mode_lib->mp.ImmediateFlipSupported)); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: PrefetchModeSupported = %u\n", __func__, mode_lib->mp.PrefetchModeSupported); + DML_LOG_VERBOSE("DML::%s: PrefetchModeSupported = %u\n", __func__, mode_lib->mp.PrefetchModeSupported); for (k = 0; k < s->num_active_planes; ++k) - dml2_printf("DML::%s: immediate_flip_required[%u] = %u\n", __func__, k, display_cfg->plane_descriptors[k].immediate_flip); - dml2_printf("DML::%s: HostVMEnable = %u\n", __func__, display_cfg->hostvm_enable); - dml2_printf("DML::%s: ImmediateFlipSupported = %u\n", __func__, mode_lib->mp.ImmediateFlipSupported); - dml2_printf("DML::%s: PrefetchAndImmediateFlipSupported = %u\n", __func__, mode_lib->mp.PrefetchAndImmediateFlipSupported); + DML_LOG_VERBOSE("DML::%s: immediate_flip_required[%u] = %u\n", __func__, k, display_cfg->plane_descriptors[k].immediate_flip); + DML_LOG_VERBOSE("DML::%s: HostVMEnable = %u\n", __func__, display_cfg->hostvm_enable); + DML_LOG_VERBOSE("DML::%s: ImmediateFlipSupported = %u\n", __func__, mode_lib->mp.ImmediateFlipSupported); + DML_LOG_VERBOSE("DML::%s: PrefetchAndImmediateFlipSupported = %u\n", __func__, mode_lib->mp.PrefetchAndImmediateFlipSupported); #endif - dml2_printf("DML::%s: Done one iteration: k=%d, MaxVStartupLines=%u\n", __func__, k, s->MaxVStartupLines[k]); + DML_LOG_VERBOSE("DML::%s: Done one iteration: k=%d, MaxVStartupLines=%u\n", __func__, k, s->MaxVStartupLines[k]); } for (k = 0; k < s->num_active_planes; ++k) - dml2_printf("DML::%s: k=%d MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); + DML_LOG_VERBOSE("DML::%s: k=%d MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); if (!mode_lib->mp.PrefetchAndImmediateFlipSupported) { - dml2_printf("DML::%s: Bad, Prefetch and flip scheduling solution NOT found!\n", __func__); + DML_LOG_VERBOSE("DML::%s: Bad, Prefetch and flip scheduling solution NOT found!\n", __func__); } else { - dml2_printf("DML::%s: Good, Prefetch and flip scheduling solution found\n", __func__); + DML_LOG_VERBOSE("DML::%s: Good, Prefetch and flip scheduling solution found\n", __func__); // DCC Configuration for (k = 0; k < s->num_active_planes; ++k) { #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: Calculate DCC configuration for surface k=%u\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: Calculate DCC configuration for surface k=%u\n", __func__, k); #endif CalculateDCCConfiguration( display_cfg->plane_descriptors[k].surface.dcc.enable, @@ -11780,8 +11718,8 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex calculate_pstate_keepout_dst_lines(display_cfg, &mode_lib->mp.Watermark, mode_lib->mp.pstate_keepout_dst_lines); - dml2_printf("DML::%s: DEBUG stream_index = %0d\n", __func__, display_cfg->plane_descriptors[0].stream_index); - dml2_printf("DML::%s: DEBUG PixelClock = %d kHz\n", __func__, (display_cfg->stream_descriptors[display_cfg->plane_descriptors[0].stream_index].timing.pixel_clock_khz)); + DML_LOG_VERBOSE("DML::%s: DEBUG stream_index = %0d\n", __func__, display_cfg->plane_descriptors[0].stream_index); + DML_LOG_VERBOSE("DML::%s: DEBUG PixelClock = %ld kHz\n", __func__, (display_cfg->stream_descriptors[display_cfg->plane_descriptors[0].stream_index].timing.pixel_clock_khz)); //Display Pipeline Delivery Time in Prefetch, Groups CalculatePixelDeliveryTimes( @@ -11893,15 +11831,15 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex mode_lib->mp.MinTTUVBlank[k] = mode_lib->mp.TCalc + mode_lib->mp.MinTTUVBlank[k]; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, MinTTUVBlank = %f (before vstartup margin)\n", __func__, k, mode_lib->mp.MinTTUVBlank[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, MinTTUVBlank = %f (before vstartup margin)\n", __func__, k, mode_lib->mp.MinTTUVBlank[k]); #endif s->Tvstartup_margin = (s->MaxVStartupLines[k] - mode_lib->mp.VStartupMin[k]) * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); mode_lib->mp.MinTTUVBlank[k] = mode_lib->mp.MinTTUVBlank[k] + s->Tvstartup_margin; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, Tvstartup_margin = %f\n", __func__, k, s->Tvstartup_margin); - dml2_printf("DML::%s: k=%u, MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); - dml2_printf("DML::%s: k=%u, MinTTUVBlank = %f\n", __func__, k, mode_lib->mp.MinTTUVBlank[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, Tvstartup_margin = %f\n", __func__, k, s->Tvstartup_margin); + DML_LOG_VERBOSE("DML::%s: k=%u, MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, MinTTUVBlank = %f\n", __func__, k, mode_lib->mp.MinTTUVBlank[k]); #endif mode_lib->mp.Tdmdl[k] = mode_lib->mp.Tdmdl[k] + s->Tvstartup_margin; @@ -11920,9 +11858,9 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex s->blank_lines_remaining = (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active) - mode_lib->mp.VStartup[k]; if (s->blank_lines_remaining < 0) { - dml2_printf("ERROR: Vstartup is larger than vblank!?\n"); + DML_LOG_VERBOSE("ERROR: Vstartup is larger than vblank!?\n"); s->blank_lines_remaining = 0; - DML2_ASSERT(0); + DML_ASSERT(0); } mode_lib->mp.MIN_DST_Y_NEXT_START[k] = s->dlg_vblank_start + s->blank_lines_remaining + s->LSetup; @@ -11936,23 +11874,23 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC[k] = false; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, VStartup = %u (max)\n", __func__, k, mode_lib->mp.VStartup[k]); - dml2_printf("DML::%s: k=%u, VStartupMin = %u (max)\n", __func__, k, mode_lib->mp.VStartupMin[k]); - dml2_printf("DML::%s: k=%u, VUpdateOffsetPix = %u\n", __func__, k, mode_lib->mp.VUpdateOffsetPix[k]); - dml2_printf("DML::%s: k=%u, VUpdateWidthPix = %u\n", __func__, k, mode_lib->mp.VUpdateWidthPix[k]); - dml2_printf("DML::%s: k=%u, VReadyOffsetPix = %u\n", __func__, k, mode_lib->mp.VReadyOffsetPix[k]); - dml2_printf("DML::%s: k=%u, HTotal = %u\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total); - dml2_printf("DML::%s: k=%u, VTotal = %u\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total); - dml2_printf("DML::%s: k=%u, VActive = %u\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active); - dml2_printf("DML::%s: k=%u, VFrontPorch = %u\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_front_porch); - dml2_printf("DML::%s: k=%u, TSetup = %f\n", __func__, k, mode_lib->mp.TSetup[k]); - dml2_printf("DML::%s: k=%u, MIN_DST_Y_NEXT_START = %f\n", __func__, k, mode_lib->mp.MIN_DST_Y_NEXT_START[k]); - dml2_printf("DML::%s: k=%u, VREADY_AT_OR_AFTER_VSYNC = %u\n", __func__, k, mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, VStartup = %u (max)\n", __func__, k, mode_lib->mp.VStartup[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, VStartupMin = %u (max)\n", __func__, k, mode_lib->mp.VStartupMin[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, VUpdateOffsetPix = %u\n", __func__, k, mode_lib->mp.VUpdateOffsetPix[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, VUpdateWidthPix = %u\n", __func__, k, mode_lib->mp.VUpdateWidthPix[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, VReadyOffsetPix = %u\n", __func__, k, mode_lib->mp.VReadyOffsetPix[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, HTotal = %lu\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total); + DML_LOG_VERBOSE("DML::%s: k=%u, VTotal = %lu\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total); + DML_LOG_VERBOSE("DML::%s: k=%u, VActive = %lu\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active); + DML_LOG_VERBOSE("DML::%s: k=%u, VFrontPorch = %lu\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_front_porch); + DML_LOG_VERBOSE("DML::%s: k=%u, TSetup = %f\n", __func__, k, mode_lib->mp.TSetup[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, MIN_DST_Y_NEXT_START = %f\n", __func__, k, mode_lib->mp.MIN_DST_Y_NEXT_START[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, VREADY_AT_OR_AFTER_VSYNC = %u\n", __func__, k, mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC[k]); #endif } //Maximum Bandwidth Used - s->TotalWRBandwidth = 0; + mode_lib->mp.TotalWRBandwidth = 0; for (k = 0; k < display_cfg->num_streams; ++k) { s->WRBandwidth = 0; if (display_cfg->stream_descriptors[k].writeback.active_writebacks_per_stream > 0) { @@ -11961,7 +11899,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex (display_cfg->stream_descriptors[k].timing.h_total * display_cfg->stream_descriptors[k].writeback.writeback_stream[0].input_height / ((double)display_cfg->stream_descriptors[k].timing.pixel_clock_khz / 1000)) * (display_cfg->stream_descriptors[k].writeback.writeback_stream[0].pixel_format == dml2_444_32 ? 4.0 : 8.0); - s->TotalWRBandwidth = s->TotalWRBandwidth + s->WRBandwidth; + mode_lib->mp.TotalWRBandwidth = mode_lib->mp.TotalWRBandwidth + s->WRBandwidth; } } @@ -11969,9 +11907,9 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex for (k = 0; k < s->num_active_planes; ++k) { mode_lib->mp.TotalDataReadBandwidth = mode_lib->mp.TotalDataReadBandwidth + mode_lib->mp.vactive_sw_bw_l[k] + mode_lib->mp.vactive_sw_bw_c[k]; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%u, TotalDataReadBandwidth = %f\n", __func__, k, mode_lib->mp.TotalDataReadBandwidth); - dml2_printf("DML::%s: k=%u, vactive_sw_bw_l = %f\n", __func__, k, mode_lib->mp.vactive_sw_bw_l[k]); - dml2_printf("DML::%s: k=%u, vactive_sw_bw_c = %f\n", __func__, k, mode_lib->mp.vactive_sw_bw_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, TotalDataReadBandwidth = %f\n", __func__, k, mode_lib->mp.TotalDataReadBandwidth); + DML_LOG_VERBOSE("DML::%s: k=%u, vactive_sw_bw_l = %f\n", __func__, k, mode_lib->mp.vactive_sw_bw_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, vactive_sw_bw_c = %f\n", __func__, k, mode_lib->mp.vactive_sw_bw_c[k]); #endif } @@ -12051,28 +11989,28 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex min_return_latency_in_DCFCLK_cycles = (min_return_uclk_cycles / max_uclk_mhz + min_return_fclk_cycles / max_fclk_mhz) * hard_minimum_dcfclk_mhz; mode_lib->mp.min_return_latency_in_dcfclk = (unsigned int)min_return_latency_in_DCFCLK_cycles; mode_lib->mp.dcfclk_deep_sleep_hysteresis = (unsigned int)math_max2(32, (double)mode_lib->ip.pixel_chunk_size_kbytes * 1024 * 3 / 4 / 64 - min_return_latency_in_DCFCLK_cycles); - DML2_ASSERT(mode_lib->mp.dcfclk_deep_sleep_hysteresis < 256); + DML_ASSERT(mode_lib->mp.dcfclk_deep_sleep_hysteresis < 256); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: max_fclk_mhz = %f\n", __func__, max_fclk_mhz); - dml2_printf("DML::%s: max_uclk_mhz = %f\n", __func__, max_uclk_mhz); - dml2_printf("DML::%s: hard_minimum_dcfclk_mhz = %f\n", __func__, hard_minimum_dcfclk_mhz); - dml2_printf("DML::%s: min_return_uclk_cycles = %d\n", __func__, min_return_uclk_cycles); - dml2_printf("DML::%s: min_return_fclk_cycles = %d\n", __func__, min_return_fclk_cycles); - dml2_printf("DML::%s: min_return_latency_in_DCFCLK_cycles = %f\n", __func__, min_return_latency_in_DCFCLK_cycles); - dml2_printf("DML::%s: dcfclk_deep_sleep_hysteresis = %d \n", __func__, mode_lib->mp.dcfclk_deep_sleep_hysteresis); - dml2_printf("DML::%s: --- END --- \n", __func__); + DML_LOG_VERBOSE("DML::%s: max_fclk_mhz = %f\n", __func__, max_fclk_mhz); + DML_LOG_VERBOSE("DML::%s: max_uclk_mhz = %f\n", __func__, max_uclk_mhz); + DML_LOG_VERBOSE("DML::%s: hard_minimum_dcfclk_mhz = %f\n", __func__, hard_minimum_dcfclk_mhz); + DML_LOG_VERBOSE("DML::%s: min_return_uclk_cycles = %ld\n", __func__, min_return_uclk_cycles); + DML_LOG_VERBOSE("DML::%s: min_return_fclk_cycles = %ld\n", __func__, min_return_fclk_cycles); + DML_LOG_VERBOSE("DML::%s: min_return_latency_in_DCFCLK_cycles = %f\n", __func__, min_return_latency_in_DCFCLK_cycles); + DML_LOG_VERBOSE("DML::%s: dcfclk_deep_sleep_hysteresis = %d \n", __func__, mode_lib->mp.dcfclk_deep_sleep_hysteresis); + DML_LOG_VERBOSE("DML::%s: --- END --- \n", __func__); #endif return (in_out_params->mode_lib->mp.PrefetchAndImmediateFlipSupported); } bool dml2_core_calcs_mode_programming_ex(struct dml2_core_calcs_mode_programming_ex *in_out_params) { - dml2_printf("DML::%s: ------------- START ----------\n", __func__); + DML_LOG_VERBOSE("DML::%s: ------------- START ----------\n", __func__); bool result = dml_core_mode_programming(in_out_params); - dml2_printf("DML::%s: result = %0d\n", __func__, result); - dml2_printf("DML::%s: ------------- DONE ----------\n", __func__); + DML_LOG_VERBOSE("DML::%s: result = %0d\n", __func__, result); + DML_LOG_VERBOSE("DML::%s: ------------- DONE ----------\n", __func__); return result; } @@ -12130,16 +12068,16 @@ void dml2_core_calcs_get_dpte_row_height( unsigned int MacroTileHeight = is_plane1 ? MacroTileHeightC : MacroTileHeightY; unsigned int PTEBufferSizeInRequests = is_plane1 ? mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma : mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML: %s: is_plane1 = %u\n", __func__, is_plane1); - dml2_printf("DML: %s: BytePerPixel = %u\n", __func__, BytePerPixel); - dml2_printf("DML: %s: BlockHeight256Bytes = %u\n", __func__, BlockHeight256Bytes); - dml2_printf("DML: %s: BlockWidth256Bytes = %u\n", __func__, BlockWidth256Bytes); - dml2_printf("DML: %s: MacroTileWidth = %u\n", __func__, MacroTileWidth); - dml2_printf("DML: %s: MacroTileHeight = %u\n", __func__, MacroTileHeight); - dml2_printf("DML: %s: PTEBufferSizeInRequests = %u\n", __func__, PTEBufferSizeInRequests); - dml2_printf("DML: %s: dpte_buffer_size_in_pte_reqs_luma = %u\n", __func__, mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma); - dml2_printf("DML: %s: dpte_buffer_size_in_pte_reqs_chroma = %u\n", __func__, mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma); - dml2_printf("DML: %s: GPUVMMinPageSizeKBytes = %u\n", __func__, GPUVMMinPageSizeKBytes); + DML_LOG_VERBOSE("DML: %s: is_plane1 = %u\n", __func__, is_plane1); + DML_LOG_VERBOSE("DML: %s: BytePerPixel = %u\n", __func__, BytePerPixel); + DML_LOG_VERBOSE("DML: %s: BlockHeight256Bytes = %u\n", __func__, BlockHeight256Bytes); + DML_LOG_VERBOSE("DML: %s: BlockWidth256Bytes = %u\n", __func__, BlockWidth256Bytes); + DML_LOG_VERBOSE("DML: %s: MacroTileWidth = %u\n", __func__, MacroTileWidth); + DML_LOG_VERBOSE("DML: %s: MacroTileHeight = %u\n", __func__, MacroTileHeight); + DML_LOG_VERBOSE("DML: %s: PTEBufferSizeInRequests = %u\n", __func__, PTEBufferSizeInRequests); + DML_LOG_VERBOSE("DML: %s: dpte_buffer_size_in_pte_reqs_luma = %u\n", __func__, mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma); + DML_LOG_VERBOSE("DML: %s: dpte_buffer_size_in_pte_reqs_chroma = %u\n", __func__, mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma); + DML_LOG_VERBOSE("DML: %s: GPUVMMinPageSizeKBytes = %u\n", __func__, GPUVMMinPageSizeKBytes); #endif unsigned int dummy_integer[21]; @@ -12193,16 +12131,16 @@ void dml2_core_calcs_get_dpte_row_height( CalculateVMAndRowBytes(&mode_lib->scratch.calculate_vm_and_row_bytes_params); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML: %s: dpte_row_height = %u\n", __func__, *dpte_row_height); + DML_LOG_VERBOSE("DML: %s: dpte_row_height = %u\n", __func__, *dpte_row_height); #endif } static bool is_dual_plane(enum dml2_source_format_class source_format) { - bool ret_val = 0; + bool ret_val = false; if ((source_format == dml2_420_12) || (source_format == dml2_420_8) || (source_format == dml2_420_10) || (source_format == dml2_rgbe_alpha)) - ret_val = 1; + ret_val = true; return ret_val; } @@ -12220,6 +12158,8 @@ static void rq_dlg_get_wm_regs(const struct dml2_display_cfg *display_cfg, const wm_regs->fclk_pstate = (int unsigned)(mode_lib->mp.Watermark.FCLKChangeWatermark * refclk_freq_in_mhz); wm_regs->sr_enter = (int unsigned)(mode_lib->mp.Watermark.StutterEnterPlusExitWatermark * refclk_freq_in_mhz); wm_regs->sr_exit = (int unsigned)(mode_lib->mp.Watermark.StutterExitWatermark * refclk_freq_in_mhz); + wm_regs->sr_enter_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterEnterPlusExitWatermark * refclk_freq_in_mhz); + wm_regs->sr_exit_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterExitWatermark * refclk_freq_in_mhz); wm_regs->temp_read_or_ppt = (int unsigned)(mode_lib->mp.Watermark.temp_read_or_ppt_watermark_us * refclk_freq_in_mhz); wm_regs->uclk_pstate = (int unsigned)(mode_lib->mp.Watermark.DRAMClockChangeWatermark * refclk_freq_in_mhz); wm_regs->urgent = (int unsigned)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); @@ -12246,11 +12186,11 @@ void dml2_core_calcs_cursor_dlg_reg(struct dml2_cursor_dlg_regs *cursor_dlg_regs cursor_dlg_regs->dst_x_offset = (unsigned int) ((dst_x_offset > 0) ? dst_x_offset : 0); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML_DLG::%s: cursor_x_position=%d\n", __func__, p->cursor_x_position); - dml2_printf("DML_DLG::%s: dlg_refclk_mhz=%f\n", __func__, p->dlg_refclk_mhz); - dml2_printf("DML_DLG::%s: pixel_rate_mhz=%f\n", __func__, p->pixel_rate_mhz); - dml2_printf("DML_DLG::%s: dst_x_offset=%d\n", __func__, dst_x_offset); - dml2_printf("DML_DLG::%s: dst_x_offset=%d (reg)\n", __func__, cursor_dlg_regs->dst_x_offset); + DML_LOG_VERBOSE("DML_DLG::%s: cursor_x_position=%d\n", __func__, p->cursor_x_position); + DML_LOG_VERBOSE("DML_DLG::%s: dlg_refclk_mhz=%f\n", __func__, p->dlg_refclk_mhz); + DML_LOG_VERBOSE("DML_DLG::%s: pixel_rate_mhz=%f\n", __func__, p->pixel_rate_mhz); + DML_LOG_VERBOSE("DML_DLG::%s: dst_x_offset=%d\n", __func__, dst_x_offset); + DML_LOG_VERBOSE("DML_DLG::%s: dst_x_offset=%d (reg)\n", __func__, cursor_dlg_regs->dst_x_offset); #endif cursor_dlg_regs->chunk_hdl_adjust = 3; @@ -12286,7 +12226,7 @@ static void rq_dlg_get_rq_reg(struct dml2_display_rq_regs *rq_regs, double stored_swath_c_bytes; bool is_phantom_pipe; - dml2_printf("DML_DLG::%s: Calculation for pipe[%d] start\n", __func__, pipe_idx); + DML_LOG_VERBOSE("DML_DLG::%s: Calculation for pipe[%d] start\n", __func__, pipe_idx); pixel_chunk_bytes = (unsigned int)(mode_lib->ip.pixel_chunk_size_kbytes * 1024); min_pixel_chunk_bytes = (unsigned int)(mode_lib->ip.min_pixel_chunk_size_bytes); @@ -12329,19 +12269,19 @@ static void rq_dlg_get_rq_reg(struct dml2_display_rq_regs *rq_regs, if (sw_mode == dml2_sw_linear && display_cfg->gpuvm_enable) { unsigned int p0_pte_row_height_linear = (unsigned int)(dml_get_dpte_row_height_linear_l(mode_lib, pipe_idx)); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML_DLG: %s: p0_pte_row_height_linear = %u\n", __func__, p0_pte_row_height_linear); + DML_LOG_VERBOSE("DML_DLG: %s: p0_pte_row_height_linear = %u\n", __func__, p0_pte_row_height_linear); #endif - DML2_ASSERT(p0_pte_row_height_linear >= 8); + DML_ASSERT(p0_pte_row_height_linear >= 8); rq_regs->rq_regs_l.pte_row_height_linear = math_log2_approx(p0_pte_row_height_linear) - 3; if (dual_plane) { unsigned int p1_pte_row_height_linear = (unsigned int)(dml_get_dpte_row_height_linear_c(mode_lib, pipe_idx)); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML_DLG: %s: p1_pte_row_height_linear = %u\n", __func__, p1_pte_row_height_linear); + DML_LOG_VERBOSE("DML_DLG: %s: p1_pte_row_height_linear = %u\n", __func__, p1_pte_row_height_linear); #endif if (sw_mode == dml2_sw_linear) { - DML2_ASSERT(p1_pte_row_height_linear >= 8); + DML_ASSERT(p1_pte_row_height_linear >= 8); } rq_regs->rq_regs_c.pte_row_height_linear = math_log2_approx(p1_pte_row_height_linear) - 3; } @@ -12375,12 +12315,12 @@ static void rq_dlg_get_rq_reg(struct dml2_display_rq_regs *rq_regs, if (stored_swath_l_bytes / stored_swath_c_bytes <= 1.5) { detile_buf_plane1_addr = (unsigned int)(detile_buf_size_in_bytes / 2.0 / 1024.0); // half to chroma #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML_DLG: %s: detile_buf_plane1_addr = %d (1/2 to chroma)\n", __func__, detile_buf_plane1_addr); + DML_LOG_VERBOSE("DML_DLG: %s: detile_buf_plane1_addr = %d (1/2 to chroma)\n", __func__, detile_buf_plane1_addr); #endif } else { detile_buf_plane1_addr = (unsigned int)(dml_round_to_multiple((unsigned int)((2.0 * detile_buf_size_in_bytes) / 3.0), 1024, 0) / 1024.0); // 2/3 to luma #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML_DLG: %s: detile_buf_plane1_addr = %d (1/3 chroma)\n", __func__, detile_buf_plane1_addr); + DML_LOG_VERBOSE("DML_DLG: %s: detile_buf_plane1_addr = %d (1/3 chroma)\n", __func__, detile_buf_plane1_addr); #endif } } @@ -12388,15 +12328,15 @@ static void rq_dlg_get_rq_reg(struct dml2_display_rq_regs *rq_regs, rq_regs->plane1_base_address = detile_buf_plane1_addr; #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML_DLG: %s: is_phantom_pipe = %d\n", __func__, is_phantom_pipe); - dml2_printf("DML_DLG: %s: stored_swath_l_bytes = %f\n", __func__, stored_swath_l_bytes); - dml2_printf("DML_DLG: %s: stored_swath_c_bytes = %f\n", __func__, stored_swath_c_bytes); - dml2_printf("DML_DLG: %s: detile_buf_size_in_bytes = %d\n", __func__, detile_buf_size_in_bytes); - dml2_printf("DML_DLG: %s: detile_buf_plane1_addr = %d\n", __func__, detile_buf_plane1_addr); - dml2_printf("DML_DLG: %s: plane1_base_address = %d\n", __func__, rq_regs->plane1_base_address); + DML_LOG_VERBOSE("DML_DLG: %s: is_phantom_pipe = %d\n", __func__, is_phantom_pipe); + DML_LOG_VERBOSE("DML_DLG: %s: stored_swath_l_bytes = %f\n", __func__, stored_swath_l_bytes); + DML_LOG_VERBOSE("DML_DLG: %s: stored_swath_c_bytes = %f\n", __func__, stored_swath_c_bytes); + DML_LOG_VERBOSE("DML_DLG: %s: detile_buf_size_in_bytes = %d\n", __func__, detile_buf_size_in_bytes); + DML_LOG_VERBOSE("DML_DLG: %s: detile_buf_plane1_addr = %d\n", __func__, detile_buf_plane1_addr); + DML_LOG_VERBOSE("DML_DLG: %s: plane1_base_address = %d\n", __func__, rq_regs->plane1_base_address); #endif - //dml2_printf_rq_regs_st(rq_regs); - dml2_printf("DML_DLG::%s: Calculation for pipe[%d] done\n", __func__, pipe_idx); + //DML_LOG_VERBOSE_rq_regs_st(rq_regs); + DML_LOG_VERBOSE("DML_DLG::%s: Calculation for pipe[%d] done\n", __func__, pipe_idx); } static void rq_dlg_get_dlg_reg( @@ -12411,10 +12351,10 @@ static void rq_dlg_get_dlg_reg( memset(l, 0, sizeof(struct dml2_core_shared_rq_dlg_get_dlg_reg_locals)); - dml2_printf("DML_DLG::%s: Calculation for pipe_idx=%d\n", __func__, pipe_idx); + DML_LOG_VERBOSE("DML_DLG::%s: Calculation for pipe_idx=%d\n", __func__, pipe_idx); l->plane_idx = dml_get_plane_idx(mode_lib, pipe_idx); - DML2_ASSERT(l->plane_idx < DML2_MAX_PLANES); + DML_ASSERT(l->plane_idx < DML2_MAX_PLANES); l->source_format = dml2_444_8; l->odm_mode = dml2_odm_mode_bypass; @@ -12444,18 +12384,18 @@ static void rq_dlg_get_dlg_reg( l->pclk_freq_in_mhz = (double)l->timing->pixel_clock_khz / 1000; l->ref_freq_to_pix_freq = l->refclk_freq_in_mhz / l->pclk_freq_in_mhz; - dml2_printf("DML_DLG::%s: plane_idx = %d\n", __func__, l->plane_idx); - dml2_printf("DML_DLG: %s: htotal = %d\n", __func__, l->htotal); - dml2_printf("DML_DLG: %s: refclk_freq_in_mhz = %3.2f\n", __func__, l->refclk_freq_in_mhz); - dml2_printf("DML_DLG: %s: dlg_ref_clk_mhz = %3.2f\n", __func__, display_cfg->overrides.hw.dlg_ref_clk_mhz); - dml2_printf("DML_DLG: %s: soc.refclk_mhz = %3.2f\n", __func__, mode_lib->soc.dchub_refclk_mhz); - dml2_printf("DML_DLG: %s: pclk_freq_in_mhz = %3.2f\n", __func__, l->pclk_freq_in_mhz); - dml2_printf("DML_DLG: %s: ref_freq_to_pix_freq = %3.2f\n", __func__, l->ref_freq_to_pix_freq); - dml2_printf("DML_DLG: %s: interlaced = %d\n", __func__, l->interlaced); + DML_LOG_VERBOSE("DML_DLG::%s: plane_idx = %d\n", __func__, l->plane_idx); + DML_LOG_VERBOSE("DML_DLG: %s: htotal = %d\n", __func__, l->htotal); + DML_LOG_VERBOSE("DML_DLG: %s: refclk_freq_in_mhz = %3.2f\n", __func__, l->refclk_freq_in_mhz); + DML_LOG_VERBOSE("DML_DLG: %s: dlg_ref_clk_mhz = %3.2f\n", __func__, display_cfg->overrides.hw.dlg_ref_clk_mhz); + DML_LOG_VERBOSE("DML_DLG: %s: soc.refclk_mhz = %d\n", __func__, mode_lib->soc.dchub_refclk_mhz); + DML_LOG_VERBOSE("DML_DLG: %s: pclk_freq_in_mhz = %3.2f\n", __func__, l->pclk_freq_in_mhz); + DML_LOG_VERBOSE("DML_DLG: %s: ref_freq_to_pix_freq = %3.2f\n", __func__, l->ref_freq_to_pix_freq); + DML_LOG_VERBOSE("DML_DLG: %s: interlaced = %d\n", __func__, l->interlaced); - DML2_ASSERT(l->refclk_freq_in_mhz != 0); - DML2_ASSERT(l->pclk_freq_in_mhz != 0); - DML2_ASSERT(l->ref_freq_to_pix_freq < 4.0); + DML_ASSERT(l->refclk_freq_in_mhz != 0); + DML_ASSERT(l->pclk_freq_in_mhz != 0); + DML_ASSERT(l->ref_freq_to_pix_freq < 4.0); // Need to figure out which side of odm combine we're in // Assume the pipe instance under the same plane is in order @@ -12484,14 +12424,14 @@ static void rq_dlg_get_dlg_reg( l->pipe_idx_in_combine = pipe_idx - l->first_pipe_idx_in_plane; // DML assumes the pipes in the same plane will have continuous indexing (i.e. plane 0 use pipe 0, 1, and plane 1 uses pipe 2, 3, etc.) disp_dlg_regs->refcyc_h_blank_end = (unsigned int)(((double)l->hblank_end + (double)l->pipe_idx_in_combine * (double)l->hactive / (double)l->odm_combine_factor) * l->ref_freq_to_pix_freq); - dml2_printf("DML_DLG: %s: pipe_idx = %d\n", __func__, pipe_idx); - dml2_printf("DML_DLG: %s: first_pipe_idx_in_plane = %d\n", __func__, l->first_pipe_idx_in_plane); - dml2_printf("DML_DLG: %s: pipe_idx_in_combine = %d\n", __func__, l->pipe_idx_in_combine); - dml2_printf("DML_DLG: %s: odm_combine_factor = %d\n", __func__, l->odm_combine_factor); + DML_LOG_VERBOSE("DML_DLG: %s: pipe_idx = %d\n", __func__, pipe_idx); + DML_LOG_VERBOSE("DML_DLG: %s: first_pipe_idx_in_plane = %d\n", __func__, l->first_pipe_idx_in_plane); + DML_LOG_VERBOSE("DML_DLG: %s: pipe_idx_in_combine = %d\n", __func__, l->pipe_idx_in_combine); + DML_LOG_VERBOSE("DML_DLG: %s: odm_combine_factor = %d\n", __func__, l->odm_combine_factor); } - dml2_printf("DML_DLG: %s: refcyc_h_blank_end = %d\n", __func__, disp_dlg_regs->refcyc_h_blank_end); + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_h_blank_end = %d\n", __func__, disp_dlg_regs->refcyc_h_blank_end); - DML2_ASSERT(disp_dlg_regs->refcyc_h_blank_end < (unsigned int)math_pow(2, 13)); + DML_ASSERT(disp_dlg_regs->refcyc_h_blank_end < (unsigned int)math_pow(2, 13)); disp_dlg_regs->ref_freq_to_pix_freq = (unsigned int)(l->ref_freq_to_pix_freq * math_pow(2, 19)); disp_dlg_regs->refcyc_per_htotal = (unsigned int)(l->ref_freq_to_pix_freq * (double)l->htotal * math_pow(2, 8)); @@ -12500,20 +12440,20 @@ static void rq_dlg_get_dlg_reg( l->min_ttu_vblank = mode_lib->mp.MinTTUVBlank[mode_lib->mp.pipe_plane[pipe_idx]]; l->min_dst_y_next_start = (unsigned int)(mode_lib->mp.MIN_DST_Y_NEXT_START[mode_lib->mp.pipe_plane[pipe_idx]]); - dml2_printf("DML_DLG: %s: min_ttu_vblank (us) = %3.2f\n", __func__, l->min_ttu_vblank); - dml2_printf("DML_DLG: %s: min_dst_y_next_start = %d\n", __func__, l->min_dst_y_next_start); - dml2_printf("DML_DLG: %s: ref_freq_to_pix_freq = %3.2f\n", __func__, l->ref_freq_to_pix_freq); + DML_LOG_VERBOSE("DML_DLG: %s: min_ttu_vblank (us) = %3.2f\n", __func__, l->min_ttu_vblank); + DML_LOG_VERBOSE("DML_DLG: %s: min_dst_y_next_start = %d\n", __func__, l->min_dst_y_next_start); + DML_LOG_VERBOSE("DML_DLG: %s: ref_freq_to_pix_freq = %3.2f\n", __func__, l->ref_freq_to_pix_freq); l->vready_after_vcount0 = (unsigned int)(mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC[mode_lib->mp.pipe_plane[pipe_idx]]); disp_dlg_regs->vready_after_vcount0 = l->vready_after_vcount0; - dml2_printf("DML_DLG: %s: vready_after_vcount0 = %d\n", __func__, disp_dlg_regs->vready_after_vcount0); + DML_LOG_VERBOSE("DML_DLG: %s: vready_after_vcount0 = %d\n", __func__, disp_dlg_regs->vready_after_vcount0); l->dst_x_after_scaler = (unsigned int)(mode_lib->mp.DSTXAfterScaler[mode_lib->mp.pipe_plane[pipe_idx]]); l->dst_y_after_scaler = (unsigned int)(mode_lib->mp.DSTYAfterScaler[mode_lib->mp.pipe_plane[pipe_idx]]); - dml2_printf("DML_DLG: %s: dst_x_after_scaler = %d\n", __func__, l->dst_x_after_scaler); - dml2_printf("DML_DLG: %s: dst_y_after_scaler = %d\n", __func__, l->dst_y_after_scaler); + DML_LOG_VERBOSE("DML_DLG: %s: dst_x_after_scaler = %d\n", __func__, l->dst_x_after_scaler); + DML_LOG_VERBOSE("DML_DLG: %s: dst_y_after_scaler = %d\n", __func__, l->dst_y_after_scaler); l->dst_y_prefetch = mode_lib->mp.dst_y_prefetch[mode_lib->mp.pipe_plane[pipe_idx]]; l->dst_y_per_vm_vblank = mode_lib->mp.dst_y_per_vm_vblank[mode_lib->mp.pipe_plane[pipe_idx]]; @@ -12521,28 +12461,28 @@ static void rq_dlg_get_dlg_reg( l->dst_y_per_vm_flip = mode_lib->mp.dst_y_per_vm_flip[mode_lib->mp.pipe_plane[pipe_idx]]; l->dst_y_per_row_flip = mode_lib->mp.dst_y_per_row_flip[mode_lib->mp.pipe_plane[pipe_idx]]; - dml2_printf("DML_DLG: %s: dst_y_prefetch (after rnd) = %3.2f\n", __func__, l->dst_y_prefetch); - dml2_printf("DML_DLG: %s: dst_y_per_vm_flip = %3.2f\n", __func__, l->dst_y_per_vm_flip); - dml2_printf("DML_DLG: %s: dst_y_per_row_flip = %3.2f\n", __func__, l->dst_y_per_row_flip); - dml2_printf("DML_DLG: %s: dst_y_per_vm_vblank = %3.2f\n", __func__, l->dst_y_per_vm_vblank); - dml2_printf("DML_DLG: %s: dst_y_per_row_vblank = %3.2f\n", __func__, l->dst_y_per_row_vblank); + DML_LOG_VERBOSE("DML_DLG: %s: dst_y_prefetch (after rnd) = %3.2f\n", __func__, l->dst_y_prefetch); + DML_LOG_VERBOSE("DML_DLG: %s: dst_y_per_vm_flip = %3.2f\n", __func__, l->dst_y_per_vm_flip); + DML_LOG_VERBOSE("DML_DLG: %s: dst_y_per_row_flip = %3.2f\n", __func__, l->dst_y_per_row_flip); + DML_LOG_VERBOSE("DML_DLG: %s: dst_y_per_vm_vblank = %3.2f\n", __func__, l->dst_y_per_vm_vblank); + DML_LOG_VERBOSE("DML_DLG: %s: dst_y_per_row_vblank = %3.2f\n", __func__, l->dst_y_per_row_vblank); if (l->dst_y_prefetch > 0 && l->dst_y_per_vm_vblank > 0 && l->dst_y_per_row_vblank > 0) { - DML2_ASSERT(l->dst_y_prefetch > (l->dst_y_per_vm_vblank + l->dst_y_per_row_vblank)); + DML_ASSERT(l->dst_y_prefetch > (l->dst_y_per_vm_vblank + l->dst_y_per_row_vblank)); } l->vratio_pre_l = mode_lib->mp.VRatioPrefetchY[mode_lib->mp.pipe_plane[pipe_idx]]; l->vratio_pre_c = mode_lib->mp.VRatioPrefetchC[mode_lib->mp.pipe_plane[pipe_idx]]; - dml2_printf("DML_DLG: %s: vratio_pre_l = %3.2f\n", __func__, l->vratio_pre_l); - dml2_printf("DML_DLG: %s: vratio_pre_c = %3.2f\n", __func__, l->vratio_pre_c); + DML_LOG_VERBOSE("DML_DLG: %s: vratio_pre_l = %3.2f\n", __func__, l->vratio_pre_l); + DML_LOG_VERBOSE("DML_DLG: %s: vratio_pre_c = %3.2f\n", __func__, l->vratio_pre_c); // Active l->refcyc_per_line_delivery_pre_l = mode_lib->mp.DisplayPipeLineDeliveryTimeLumaPrefetch[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; l->refcyc_per_line_delivery_l = mode_lib->mp.DisplayPipeLineDeliveryTimeLuma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - dml2_printf("DML_DLG: %s: refcyc_per_line_delivery_pre_l = %3.2f\n", __func__, l->refcyc_per_line_delivery_pre_l); - dml2_printf("DML_DLG: %s: refcyc_per_line_delivery_l = %3.2f\n", __func__, l->refcyc_per_line_delivery_l); + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_line_delivery_pre_l = %3.2f\n", __func__, l->refcyc_per_line_delivery_pre_l); + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_line_delivery_l = %3.2f\n", __func__, l->refcyc_per_line_delivery_l); l->refcyc_per_line_delivery_pre_c = 0.0; l->refcyc_per_line_delivery_c = 0.0; @@ -12551,8 +12491,8 @@ static void rq_dlg_get_dlg_reg( l->refcyc_per_line_delivery_pre_c = mode_lib->mp.DisplayPipeLineDeliveryTimeChromaPrefetch[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; l->refcyc_per_line_delivery_c = mode_lib->mp.DisplayPipeLineDeliveryTimeChroma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - dml2_printf("DML_DLG: %s: refcyc_per_line_delivery_pre_c = %3.2f\n", __func__, l->refcyc_per_line_delivery_pre_c); - dml2_printf("DML_DLG: %s: refcyc_per_line_delivery_c = %3.2f\n", __func__, l->refcyc_per_line_delivery_c); + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_line_delivery_pre_c = %3.2f\n", __func__, l->refcyc_per_line_delivery_pre_c); + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_line_delivery_c = %3.2f\n", __func__, l->refcyc_per_line_delivery_c); } disp_dlg_regs->refcyc_per_vm_dmdata = (unsigned int)(mode_lib->mp.Tdmdl_vm[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz); @@ -12561,8 +12501,8 @@ static void rq_dlg_get_dlg_reg( l->refcyc_per_req_delivery_pre_l = mode_lib->mp.DisplayPipeRequestDeliveryTimeLumaPrefetch[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; l->refcyc_per_req_delivery_l = mode_lib->mp.DisplayPipeRequestDeliveryTimeLuma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - dml2_printf("DML_DLG: %s: refcyc_per_req_delivery_pre_l = %3.2f\n", __func__, l->refcyc_per_req_delivery_pre_l); - dml2_printf("DML_DLG: %s: refcyc_per_req_delivery_l = %3.2f\n", __func__, l->refcyc_per_req_delivery_l); + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_req_delivery_pre_l = %3.2f\n", __func__, l->refcyc_per_req_delivery_pre_l); + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_req_delivery_l = %3.2f\n", __func__, l->refcyc_per_req_delivery_l); l->refcyc_per_req_delivery_pre_c = 0.0; l->refcyc_per_req_delivery_c = 0.0; @@ -12570,16 +12510,16 @@ static void rq_dlg_get_dlg_reg( l->refcyc_per_req_delivery_pre_c = mode_lib->mp.DisplayPipeRequestDeliveryTimeChromaPrefetch[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; l->refcyc_per_req_delivery_c = mode_lib->mp.DisplayPipeRequestDeliveryTimeChroma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - dml2_printf("DML_DLG: %s: refcyc_per_req_delivery_pre_c = %3.2f\n", __func__, l->refcyc_per_req_delivery_pre_c); - dml2_printf("DML_DLG: %s: refcyc_per_req_delivery_c = %3.2f\n", __func__, l->refcyc_per_req_delivery_c); + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_req_delivery_pre_c = %3.2f\n", __func__, l->refcyc_per_req_delivery_pre_c); + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_req_delivery_c = %3.2f\n", __func__, l->refcyc_per_req_delivery_c); } // TTU - Cursor - DML2_ASSERT(display_cfg->plane_descriptors[l->plane_idx].cursor.num_cursors <= 1); + DML_ASSERT(display_cfg->plane_descriptors[l->plane_idx].cursor.num_cursors <= 1); // Assign to register structures disp_dlg_regs->min_dst_y_next_start = (unsigned int)((double)l->min_dst_y_next_start * math_pow(2, 2)); - DML2_ASSERT(disp_dlg_regs->min_dst_y_next_start < (unsigned int)math_pow(2, 18)); + DML_ASSERT(disp_dlg_regs->min_dst_y_next_start < (unsigned int)math_pow(2, 18)); disp_dlg_regs->dst_y_after_scaler = l->dst_y_after_scaler; // in terms of line disp_dlg_regs->refcyc_x_after_scaler = (unsigned int)((double)l->dst_x_after_scaler * l->ref_freq_to_pix_freq); // in terms of refclk @@ -12592,10 +12532,10 @@ static void rq_dlg_get_dlg_reg( disp_dlg_regs->vratio_prefetch = (unsigned int)(l->vratio_pre_l * math_pow(2, 19)); disp_dlg_regs->vratio_prefetch_c = (unsigned int)(l->vratio_pre_c * math_pow(2, 19)); - dml2_printf("DML_DLG: %s: disp_dlg_regs->dst_y_per_vm_vblank = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_vm_vblank); - dml2_printf("DML_DLG: %s: disp_dlg_regs->dst_y_per_row_vblank = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_row_vblank); - dml2_printf("DML_DLG: %s: disp_dlg_regs->dst_y_per_vm_flip = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_vm_flip); - dml2_printf("DML_DLG: %s: disp_dlg_regs->dst_y_per_row_flip = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_row_flip); + DML_LOG_VERBOSE("DML_DLG: %s: disp_dlg_regs->dst_y_per_vm_vblank = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_vm_vblank); + DML_LOG_VERBOSE("DML_DLG: %s: disp_dlg_regs->dst_y_per_row_vblank = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_row_vblank); + DML_LOG_VERBOSE("DML_DLG: %s: disp_dlg_regs->dst_y_per_vm_flip = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_vm_flip); + DML_LOG_VERBOSE("DML_DLG: %s: disp_dlg_regs->dst_y_per_row_flip = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_row_flip); disp_dlg_regs->refcyc_per_vm_group_vblank = (unsigned int)(mode_lib->mp.TimePerVMGroupVBlank[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz); disp_dlg_regs->refcyc_per_vm_group_flip = (unsigned int)(mode_lib->mp.TimePerVMGroupFlip[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz); @@ -12662,11 +12602,11 @@ static void rq_dlg_get_dlg_reg( disp_ttu_regs->qos_ramp_disable_c = 0; disp_ttu_regs->min_ttu_vblank = (unsigned int)(l->min_ttu_vblank * l->refclk_freq_in_mhz); - // CHECK for HW registers' range, DML2_ASSERT or clamp - DML2_ASSERT(l->refcyc_per_req_delivery_pre_l < math_pow(2, 13)); - DML2_ASSERT(l->refcyc_per_req_delivery_l < math_pow(2, 13)); - DML2_ASSERT(l->refcyc_per_req_delivery_pre_c < math_pow(2, 13)); - DML2_ASSERT(l->refcyc_per_req_delivery_c < math_pow(2, 13)); + // CHECK for HW registers' range, DML_ASSERT or clamp + DML_ASSERT(l->refcyc_per_req_delivery_pre_l < math_pow(2, 13)); + DML_ASSERT(l->refcyc_per_req_delivery_l < math_pow(2, 13)); + DML_ASSERT(l->refcyc_per_req_delivery_pre_c < math_pow(2, 13)); + DML_ASSERT(l->refcyc_per_req_delivery_c < math_pow(2, 13)); if (disp_dlg_regs->refcyc_per_vm_group_vblank >= (unsigned int)math_pow(2, 23)) disp_dlg_regs->refcyc_per_vm_group_vblank = (unsigned int)(math_pow(2, 23) - 1); @@ -12680,16 +12620,16 @@ static void rq_dlg_get_dlg_reg( disp_dlg_regs->refcyc_per_vm_req_flip = (unsigned int)(math_pow(2, 23) - 1); - DML2_ASSERT(disp_dlg_regs->dst_y_after_scaler < (unsigned int)8); - DML2_ASSERT(disp_dlg_regs->refcyc_x_after_scaler < (unsigned int)math_pow(2, 13)); + DML_ASSERT(disp_dlg_regs->dst_y_after_scaler < (unsigned int)8); + DML_ASSERT(disp_dlg_regs->refcyc_x_after_scaler < (unsigned int)math_pow(2, 13)); if (disp_dlg_regs->dst_y_per_pte_row_nom_l >= (unsigned int)math_pow(2, 17)) { - dml2_printf("DML_DLG: %s: Warning DST_Y_PER_PTE_ROW_NOM_L %u > register max U15.2 %u, clamp to max\n", __func__, disp_dlg_regs->dst_y_per_pte_row_nom_l, (unsigned int)math_pow(2, 17) - 1); + DML_LOG_VERBOSE("DML_DLG: %s: Warning DST_Y_PER_PTE_ROW_NOM_L %u > register max U15.2 %u, clamp to max\n", __func__, disp_dlg_regs->dst_y_per_pte_row_nom_l, (unsigned int)math_pow(2, 17) - 1); l->dst_y_per_pte_row_nom_l = (unsigned int)math_pow(2, 17) - 1; } if (l->dual_plane) { if (disp_dlg_regs->dst_y_per_pte_row_nom_c >= (unsigned int)math_pow(2, 17)) { - dml2_printf("DML_DLG: %s: Warning DST_Y_PER_PTE_ROW_NOM_C %u > register max U15.2 %u, clamp to max\n", __func__, disp_dlg_regs->dst_y_per_pte_row_nom_c, (unsigned int)math_pow(2, 17) - 1); + DML_LOG_VERBOSE("DML_DLG: %s: Warning DST_Y_PER_PTE_ROW_NOM_C %u > register max U15.2 %u, clamp to max\n", __func__, disp_dlg_regs->dst_y_per_pte_row_nom_c, (unsigned int)math_pow(2, 17) - 1); l->dst_y_per_pte_row_nom_c = (unsigned int)math_pow(2, 17) - 1; } } @@ -12700,20 +12640,20 @@ static void rq_dlg_get_dlg_reg( if (disp_dlg_regs->refcyc_per_pte_group_nom_c >= (unsigned int)math_pow(2, 23)) disp_dlg_regs->refcyc_per_pte_group_nom_c = (unsigned int)(math_pow(2, 23) - 1); } - DML2_ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_l < (unsigned int)math_pow(2, 13)); + DML_ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_l < (unsigned int)math_pow(2, 13)); if (l->dual_plane) { - DML2_ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_c < (unsigned int)math_pow(2, 13)); + DML_ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_c < (unsigned int)math_pow(2, 13)); } - DML2_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_l < (unsigned int)math_pow(2, 13)); - DML2_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_l < (unsigned int)math_pow(2, 13)); - DML2_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_c < (unsigned int)math_pow(2, 13)); - DML2_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_c < (unsigned int)math_pow(2, 13)); - DML2_ASSERT(disp_ttu_regs->qos_level_low_wm < (unsigned int)math_pow(2, 14)); - DML2_ASSERT(disp_ttu_regs->qos_level_high_wm < (unsigned int)math_pow(2, 14)); - DML2_ASSERT(disp_ttu_regs->min_ttu_vblank < (unsigned int)math_pow(2, 24)); + DML_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_l < (unsigned int)math_pow(2, 13)); + DML_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_l < (unsigned int)math_pow(2, 13)); + DML_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_c < (unsigned int)math_pow(2, 13)); + DML_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_c < (unsigned int)math_pow(2, 13)); + DML_ASSERT(disp_ttu_regs->qos_level_low_wm < (unsigned int)math_pow(2, 14)); + DML_ASSERT(disp_ttu_regs->qos_level_high_wm < (unsigned int)math_pow(2, 14)); + DML_ASSERT(disp_ttu_regs->min_ttu_vblank < (unsigned int)math_pow(2, 24)); - dml2_printf("DML_DLG::%s: Calculation for pipe[%d] done\n", __func__, pipe_idx); + DML_LOG_VERBOSE("DML_DLG::%s: Calculation for pipe[%d] done\n", __func__, pipe_idx); } } @@ -12736,11 +12676,11 @@ static void rq_dlg_get_arb_params(const struct dml2_display_cfg *display_cfg, co arb_param->pstate_stall_threshold = (unsigned int)(mode_lib->ip_caps.fams2.max_allow_delay_us * refclk_freq_in_mhz); #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: max_req_outstanding = %d\n", __func__, arb_param->max_req_outstanding); - dml2_printf("DML::%s: sdpif_request_rate_limit = %d\n", __func__, arb_param->sdpif_request_rate_limit); - dml2_printf("DML::%s: compbuf_reserved_space_kbytes = %d\n", __func__, arb_param->compbuf_reserved_space_kbytes); - dml2_printf("DML::%s: allow_sdpif_rate_limit_when_cstate_req = %d\n", __func__, arb_param->allow_sdpif_rate_limit_when_cstate_req); - dml2_printf("DML::%s: dcfclk_deep_sleep_hysteresis = %d\n", __func__, arb_param->dcfclk_deep_sleep_hysteresis); + DML_LOG_VERBOSE("DML::%s: max_req_outstanding = %d\n", __func__, arb_param->max_req_outstanding); + DML_LOG_VERBOSE("DML::%s: sdpif_request_rate_limit = %d\n", __func__, arb_param->sdpif_request_rate_limit); + DML_LOG_VERBOSE("DML::%s: compbuf_reserved_space_kbytes = %d\n", __func__, arb_param->compbuf_reserved_space_kbytes); + DML_LOG_VERBOSE("DML::%s: allow_sdpif_rate_limit_when_cstate_req = %d\n", __func__, arb_param->allow_sdpif_rate_limit_when_cstate_req); + DML_LOG_VERBOSE("DML::%s: dcfclk_deep_sleep_hysteresis = %d\n", __func__, arb_param->dcfclk_deep_sleep_hysteresis); #endif } @@ -13013,10 +12953,10 @@ void dml2_core_calcs_get_stream_support_info(const struct dml2_display_cfg *disp out->vblank_reserved_time_us = display_cfg->plane_descriptors[plane_index].overrides.reserved_vblank_time_ns / 1000; #if defined(__DML_VBA_DEBUG__) - dml2_printf("DML::%s: subvp_fw_processing_delay_us = %d\n", __func__, mode_lib->ip.subvp_fw_processing_delay_us); - dml2_printf("DML::%s: subvp_pstate_allow_width_us = %d\n", __func__, mode_lib->ip.subvp_pstate_allow_width_us); - dml2_printf("DML::%s: subvp_swath_height_margin_lines = %d\n", __func__, mode_lib->ip.subvp_swath_height_margin_lines); - dml2_printf("DML::%s: vblank_reserved_time_us = %f\n", __func__, out->vblank_reserved_time_us); + DML_LOG_VERBOSE("DML::%s: subvp_fw_processing_delay_us = %d\n", __func__, mode_lib->ip.subvp_fw_processing_delay_us); + DML_LOG_VERBOSE("DML::%s: subvp_pstate_allow_width_us = %d\n", __func__, mode_lib->ip.subvp_pstate_allow_width_us); + DML_LOG_VERBOSE("DML::%s: subvp_swath_height_margin_lines = %d\n", __func__, mode_lib->ip.subvp_swath_height_margin_lines); + DML_LOG_VERBOSE("DML::%s: vblank_reserved_time_us = %u\n", __func__, out->vblank_reserved_time_us); #endif } @@ -13129,6 +13069,10 @@ void dml2_core_calcs_get_informative(const struct dml2_core_internal_display_mod out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_10x4; else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_12x4) out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_12x4; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_16x4) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_16x4; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_20x4) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_20x4; out->informative.mode_support_info.AlignedYPitch[k] = mode_lib->ms.support.AlignedYPitch[k]; out->informative.mode_support_info.AlignedCPitch[k] = mode_lib->ms.support.AlignedCPitch[k]; @@ -13313,7 +13257,7 @@ void dml2_core_calcs_get_informative(const struct dml2_core_internal_display_mod out->informative.misc.DisplayPipeLineDeliveryTimeLumaPrefetch[k] = mode_lib->mp.DisplayPipeLineDeliveryTimeLumaPrefetch[k]; out->informative.misc.DisplayPipeLineDeliveryTimeChromaPrefetch[k] = mode_lib->mp.DisplayPipeLineDeliveryTimeChromaPrefetch[k]; - out->informative.misc.WritebackRequiredBandwidth = mode_lib->scratch.dml_core_mode_programming_locals.TotalWRBandwidth / 1000.0; + out->informative.misc.WritebackRequiredBandwidth = mode_lib->mp.TotalWRBandwidth / 1000.0; out->informative.misc.WritebackAllowDRAMClockChangeEndPosition[k] = mode_lib->mp.WritebackAllowDRAMClockChangeEndPosition[k]; out->informative.misc.WritebackAllowFCLKChangeEndPosition[k] = mode_lib->mp.WritebackAllowFCLKChangeEndPosition[k]; out->informative.misc.DSCCLK_calculated[k] = mode_lib->mp.DSCCLK[k]; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h index 4e502f0a6d207..28687565ac222 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h @@ -102,6 +102,7 @@ struct dml2_core_internal_DmlPipe { double DCFClkDeepSleep; unsigned int DPPPerSurface; bool ScalerEnabled; + bool UPSPEnabled; enum dml2_rotation_angle RotationAngle; bool mirrored; unsigned int ViewportHeight; @@ -186,7 +187,9 @@ enum dml2_core_internal_output_type_rate { dml2_core_internal_output_rate_hdmi_rate_6x4 = 9, dml2_core_internal_output_rate_hdmi_rate_8x4 = 10, dml2_core_internal_output_rate_hdmi_rate_10x4 = 11, - dml2_core_internal_output_rate_hdmi_rate_12x4 = 12 + dml2_core_internal_output_rate_hdmi_rate_12x4 = 12, + dml2_core_internal_output_rate_hdmi_rate_16x4 = 13, + dml2_core_internal_output_rate_hdmi_rate_20x4 = 14 }; struct dml2_core_internal_watermarks { @@ -260,12 +263,14 @@ struct dml2_core_internal_mode_support_info { bool AvgBandwidthSupport; bool UrgVactiveBandwidthSupport; bool EnoughUrgentLatencyHidingSupport; + bool PrefetchScheduleSupported; bool PrefetchSupported; bool PrefetchBandwidthSupported; bool DynamicMetadataSupported; bool VRatioInPrefetchSupported; bool DISPCLK_DPPCLK_Support; bool TotalAvailablePipesSupport; + bool ODMSupport; bool ModeSupport; bool ViewportSizeSupport; @@ -314,9 +319,7 @@ struct dml2_core_internal_mode_support_info { double non_urg_bandwidth_required[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max]; // same as urg_bandwidth, except not scaled by urg burst factor double non_urg_bandwidth_required_flip[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max]; - bool avg_bandwidth_support_ok[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max]; - double max_urgent_latency_us; double max_non_urgent_latency_us; double avg_non_urgent_latency_us; @@ -329,6 +332,8 @@ struct dml2_core_internal_mode_support_info { bool temp_read_or_ppt_support; struct dml2_core_internal_watermarks watermarks; + bool dcfclk_support; + bool qos_bandwidth_support; }; struct dml2_core_internal_mode_support { @@ -350,9 +355,11 @@ struct dml2_core_internal_mode_support { double SOCCLK; /// <brief Basically just the clock freq at the min (or given) state double DCFCLK; /// <brief Basically just the clock freq at the min (or given) state and max combine setting double GlobalDPPCLK; /// <brief the Max DPPCLK freq out of all pipes + double GlobalDTBCLK; /// <brief the Max DTBCLK freq out of all pipes double uclk_freq_mhz; double dram_bw_mbps; double max_dram_bw_mbps; + double min_available_urgent_bandwidth_MBps; /// <brief Minimum guaranteed available urgent return bandwidth in MBps double MaxFabricClock; /// <brief Basically just the clock freq at the min (or given) state double MaxDCFCLK; /// <brief Basically just the clock freq at the min (or given) state and max combine setting @@ -394,9 +401,13 @@ struct dml2_core_internal_mode_support { double TWait[DML2_MAX_PLANES]; bool UnboundedRequestEnabled; + unsigned int compbuf_reserved_space_64b; + bool hw_debug5; unsigned int CompressedBufferSizeInkByte; double VRatioPreY[DML2_MAX_PLANES]; double VRatioPreC[DML2_MAX_PLANES]; + unsigned int req_per_swath_ub_l[DML2_MAX_PLANES]; + unsigned int req_per_swath_ub_c[DML2_MAX_PLANES]; unsigned int swath_width_luma_ub[DML2_MAX_PLANES]; unsigned int swath_width_chroma_ub[DML2_MAX_PLANES]; unsigned int RequiredSlots[DML2_MAX_PLANES]; @@ -417,8 +428,8 @@ struct dml2_core_internal_mode_support { double dst_y_prefetch[DML2_MAX_PLANES]; double LinesForVM[DML2_MAX_PLANES]; double LinesForDPTERow[DML2_MAX_PLANES]; - double SwathWidthYSingleDPP[DML2_MAX_PLANES]; - double SwathWidthCSingleDPP[DML2_MAX_PLANES]; + unsigned int SwathWidthYSingleDPP[DML2_MAX_PLANES]; + unsigned int SwathWidthCSingleDPP[DML2_MAX_PLANES]; unsigned int BytePerPixelY[DML2_MAX_PLANES]; unsigned int BytePerPixelC[DML2_MAX_PLANES]; double BytePerPixelInDETY[DML2_MAX_PLANES]; @@ -469,13 +480,58 @@ struct dml2_core_internal_mode_support { double mall_prefetch_sdp_overhead_factor[DML2_MAX_PLANES]; // overhead to the imall or phantom pipe double mall_prefetch_dram_overhead_factor[DML2_MAX_PLANES]; + bool is_using_mall_for_ss[DML2_MAX_PLANES]; + unsigned int meta_row_width_chroma[DML2_MAX_PLANES]; + unsigned int PixelPTEReqHeightC[DML2_MAX_PLANES]; + bool PTE_BUFFER_MODE[DML2_MAX_PLANES]; + unsigned int meta_req_height_chroma[DML2_MAX_PLANES]; + unsigned int meta_pte_bytes_per_frame_ub_c[DML2_MAX_PLANES]; + unsigned int dpde0_bytes_per_frame_ub_c[DML2_MAX_PLANES]; + unsigned int dpte_row_width_luma_ub[DML2_MAX_PLANES]; + unsigned int meta_req_width[DML2_MAX_PLANES]; + unsigned int meta_row_width[DML2_MAX_PLANES]; + unsigned int PixelPTEReqWidthY[DML2_MAX_PLANES]; + unsigned int dpte_row_height_linear[DML2_MAX_PLANES]; + unsigned int PTERequestSizeY[DML2_MAX_PLANES]; + unsigned int dpte_row_width_chroma_ub[DML2_MAX_PLANES]; + unsigned int PixelPTEReqWidthC[DML2_MAX_PLANES]; + unsigned int meta_pte_bytes_per_frame_ub_l[DML2_MAX_PLANES]; + unsigned int dpte_row_height_linear_chroma[DML2_MAX_PLANES]; + unsigned int PTERequestSizeC[DML2_MAX_PLANES]; + unsigned int meta_req_height[DML2_MAX_PLANES]; + unsigned int dpde0_bytes_per_frame_ub_l[DML2_MAX_PLANES]; + unsigned int meta_req_width_chroma[DML2_MAX_PLANES]; + unsigned int PixelPTEReqHeightY[DML2_MAX_PLANES]; + unsigned int BIGK_FRAGMENT_SIZE[DML2_MAX_PLANES]; + unsigned int vm_group_bytes[DML2_MAX_PLANES]; + unsigned int VReadyOffsetPix[DML2_MAX_PLANES]; + unsigned int VUpdateOffsetPix[DML2_MAX_PLANES]; + unsigned int VUpdateWidthPix[DML2_MAX_PLANES]; + double TSetup[DML2_MAX_PLANES]; + double Tdmdl_vm_raw[DML2_MAX_PLANES]; + double Tdmdl_raw[DML2_MAX_PLANES]; + unsigned int VStartupMin[DML2_MAX_PLANES]; /// <brief Minimum vstartup to meet the prefetch schedule (i.e. the prefetch solution can be found at this vstartup time); not the actual global sync vstartup pos. + double MaxActiveDRAMClockChangeLatencySupported[DML2_MAX_PLANES]; + double MaxActiveFCLKChangeLatencySupported; + // Backend bool RequiresDSC[DML2_MAX_PLANES]; bool RequiresFEC[DML2_MAX_PLANES]; double OutputBpp[DML2_MAX_PLANES]; + double DesiredOutputBpp[DML2_MAX_PLANES]; + double PixelClockBackEnd[DML2_MAX_PLANES]; unsigned int DSCDelay[DML2_MAX_PLANES]; enum dml2_core_internal_output_type OutputType[DML2_MAX_PLANES]; enum dml2_core_internal_output_type_rate OutputRate[DML2_MAX_PLANES]; + bool TotalAvailablePipesSupportNoDSC; + bool TotalAvailablePipesSupportDSC; + unsigned int NumberOfDPPNoDSC; + unsigned int NumberOfDPPDSC; + enum dml2_odm_mode ODMModeNoDSC; + enum dml2_odm_mode ODMModeDSC; + double RequiredDISPCLKPerSurfaceNoDSC; + double RequiredDISPCLKPerSurfaceDSC; + unsigned int EstimatedNumberOfDSCSlices[DML2_MAX_PLANES]; // Bandwidth Related Info double BandwidthAvailableForImmediateFlip; @@ -484,8 +540,14 @@ struct dml2_core_internal_mode_support { double WriteBandwidth[DML2_MAX_PLANES][DML2_MAX_WRITEBACK]; double RequiredPrefetchPixelDataBWLuma[DML2_MAX_PLANES]; double RequiredPrefetchPixelDataBWChroma[DML2_MAX_PLANES]; - /* oto bw should also be considered when calculating peak urgent bw to avoid situations oto/equ mismatches between ms and mp */ - double RequiredPrefetchBWOTO[DML2_MAX_PLANES]; + /* Max bandwidth calculated from prefetch schedule should be considered in addition to the pixel data bw to avoid ms/mp mismatches. + * 1. oto bw should also be considered when calculating peak urgent bw to avoid situations oto/equ mismatches between ms and mp + * + * 2. equ bandwidth needs to be considered for calculating peak urgent bw when equ schedule is used in mode support. + * Some slight difference in variables may cause the pixel data bandwidth to be higher + * even though overall equ prefetch bandwidths can be lower going from ms to mp + */ + double RequiredPrefetchBWMax[DML2_MAX_PLANES]; double cursor_bw[DML2_MAX_PLANES]; double prefetch_cursor_bw[DML2_MAX_PLANES]; double prefetch_vmrow_bw[DML2_MAX_PLANES]; @@ -538,7 +600,44 @@ struct dml2_core_internal_mode_support { bool mall_comb_mcache_c[DML2_MAX_PLANES]; bool lc_comb_mcache[DML2_MAX_PLANES]; + unsigned int vmpg_width_y[DML2_MAX_PLANES]; + unsigned int vmpg_height_y[DML2_MAX_PLANES]; + unsigned int vmpg_width_c[DML2_MAX_PLANES]; + unsigned int vmpg_height_c[DML2_MAX_PLANES]; + + unsigned int meta_row_height_luma[DML2_MAX_PLANES]; + unsigned int meta_row_height_chroma[DML2_MAX_PLANES]; + unsigned int meta_row_bytes_per_row_ub_l[DML2_MAX_PLANES]; + unsigned int meta_row_bytes_per_row_ub_c[DML2_MAX_PLANES]; + unsigned int dpte_row_bytes_per_row_l[DML2_MAX_PLANES]; + unsigned int dpte_row_bytes_per_row_c[DML2_MAX_PLANES]; + + unsigned int pstate_bytes_required_l[DML2_MAX_PLANES]; + unsigned int pstate_bytes_required_c[DML2_MAX_PLANES]; + unsigned int cursor_bytes_per_chunk[DML2_MAX_PLANES]; + unsigned int cursor_bytes_per_line[DML2_MAX_PLANES]; + + unsigned int MaximumVStartup[DML2_MAX_PLANES]; + + double HostVMInefficiencyFactor; + double HostVMInefficiencyFactorPrefetch; + + unsigned int tdlut_pte_bytes_per_frame[DML2_MAX_PLANES]; + unsigned int tdlut_bytes_per_frame[DML2_MAX_PLANES]; + unsigned int tdlut_groups_per_2row_ub[DML2_MAX_PLANES]; + double tdlut_opt_time[DML2_MAX_PLANES]; + double tdlut_drain_time[DML2_MAX_PLANES]; + unsigned int tdlut_bytes_per_group[DML2_MAX_PLANES]; + + double Tvm_trips_flip[DML2_MAX_PLANES]; + double Tr0_trips_flip[DML2_MAX_PLANES]; + double Tvm_trips_flip_rounded[DML2_MAX_PLANES]; + double Tr0_trips_flip_rounded[DML2_MAX_PLANES]; + + unsigned int DSTYAfterScaler[DML2_MAX_PLANES]; + unsigned int DSTXAfterScaler[DML2_MAX_PLANES]; + enum dml2_pstate_method pstate_switch_modes[DML2_MAX_PLANES]; }; /// @brief A mega structure that houses various info for model programming step. @@ -548,6 +647,7 @@ struct dml2_core_internal_mode_program { double FabricClock; /// <brief Basically just the clock freq at the min (or given) state //double DCFCLK; /// <brief Basically just the clock freq at the min (or given) state and max combine setting double dram_bw_mbps; + double min_available_urgent_bandwidth_MBps; /// <brief Minimum guaranteed available urgent return bandwidth in MBps double uclk_freq_mhz; unsigned int NoOfDPP[DML2_MAX_PLANES]; enum dml2_odm_mode ODMMode[DML2_MAX_PLANES]; @@ -599,6 +699,8 @@ struct dml2_core_internal_mode_program { unsigned int MacroTileHeightC[DML2_MAX_PLANES]; unsigned int MacroTileWidthY[DML2_MAX_PLANES]; unsigned int MacroTileWidthC[DML2_MAX_PLANES]; + double MaximumSwathWidthLuma[DML2_MAX_PLANES]; + double MaximumSwathWidthChroma[DML2_MAX_PLANES]; bool surf_linear128_l[DML2_MAX_PLANES]; bool surf_linear128_c[DML2_MAX_PLANES]; @@ -631,6 +733,14 @@ struct dml2_core_internal_mode_program { double UrgentBurstFactorChroma[DML2_MAX_PLANES]; double UrgentBurstFactorChromaPre[DML2_MAX_PLANES]; + double MaximumSwathWidthInLineBufferLuma; + double MaximumSwathWidthInLineBufferChroma; + + unsigned int vmpg_width_y[DML2_MAX_PLANES]; + unsigned int vmpg_height_y[DML2_MAX_PLANES]; + unsigned int vmpg_width_c[DML2_MAX_PLANES]; + unsigned int vmpg_height_c[DML2_MAX_PLANES]; + double meta_row_bw[DML2_MAX_PLANES]; unsigned int meta_row_bytes[DML2_MAX_PLANES]; unsigned int meta_req_width[DML2_MAX_PLANES]; @@ -652,7 +762,9 @@ struct dml2_core_internal_mode_program { unsigned int PTERequestSizeC[DML2_MAX_PLANES]; double TWait[DML2_MAX_PLANES]; + double Tdmdl_vm_raw[DML2_MAX_PLANES]; double Tdmdl_vm[DML2_MAX_PLANES]; + double Tdmdl_raw[DML2_MAX_PLANES]; double Tdmdl[DML2_MAX_PLANES]; double TSetup[DML2_MAX_PLANES]; unsigned int dpde0_bytes_per_frame_ub_l[DML2_MAX_PLANES]; @@ -684,6 +796,38 @@ struct dml2_core_internal_mode_program { double TCalc; unsigned int TotImmediateFlipBytes; + unsigned int MaxTotalDETInKByte; + unsigned int NomDETInKByte; + unsigned int MinCompressedBufferSizeInKByte; + double PixelClockBackEnd[DML2_MAX_PLANES]; + double OutputBpp[DML2_MAX_PLANES]; + bool dsc_enable[DML2_MAX_PLANES]; + unsigned int num_dsc_slices[DML2_MAX_PLANES]; + unsigned int meta_row_bytes_per_row_ub_l[DML2_MAX_PLANES]; + unsigned int meta_row_bytes_per_row_ub_c[DML2_MAX_PLANES]; + unsigned int dpte_row_bytes_per_row_l[DML2_MAX_PLANES]; + unsigned int dpte_row_bytes_per_row_c[DML2_MAX_PLANES]; + unsigned int cursor_bytes_per_chunk[DML2_MAX_PLANES]; + unsigned int cursor_bytes_per_line[DML2_MAX_PLANES]; + unsigned int MaxVStartupLines[DML2_MAX_PLANES]; /// <brief more like vblank for the plane's OTG + double HostVMInefficiencyFactor; + double HostVMInefficiencyFactorPrefetch; + unsigned int tdlut_pte_bytes_per_frame[DML2_MAX_PLANES]; + unsigned int tdlut_bytes_per_frame[DML2_MAX_PLANES]; + unsigned int tdlut_groups_per_2row_ub[DML2_MAX_PLANES]; + double tdlut_opt_time[DML2_MAX_PLANES]; + double tdlut_drain_time[DML2_MAX_PLANES]; + unsigned int tdlut_bytes_per_group[DML2_MAX_PLANES]; + double Tvm_trips_flip[DML2_MAX_PLANES]; + double Tr0_trips_flip[DML2_MAX_PLANES]; + double Tvm_trips_flip_rounded[DML2_MAX_PLANES]; + double Tr0_trips_flip_rounded[DML2_MAX_PLANES]; + bool immediate_flip_required; // any pipes need immediate flip + double SOCCLK; /// <brief Basically just the clock freq at the min (or given) state + double TotalWRBandwidth; + double max_urgent_latency_us; + double df_response_time_us; + // ------------------- // Output // ------------------- @@ -694,9 +838,12 @@ struct dml2_core_internal_mode_program { // Support bool UrgVactiveBandwidthSupport; + bool PrefetchScheduleSupported; + bool UrgentBandwidthSupport; bool PrefetchModeSupported; // <brief Is the prefetch mode (bandwidth and latency) supported bool ImmediateFlipSupported; bool ImmediateFlipSupportedForPipe[DML2_MAX_PLANES]; + bool dcfclk_support; // Clock double Dcfclk; @@ -788,7 +935,7 @@ struct dml2_core_internal_mode_program { // RQ registers bool PTE_BUFFER_MODE[DML2_MAX_PLANES]; unsigned int BIGK_FRAGMENT_SIZE[DML2_MAX_PLANES]; - + double VActiveLatencyHidingUs[DML2_MAX_PLANES]; unsigned int SubViewportLinesNeededInMALL[DML2_MAX_PLANES]; bool is_using_mall_for_ss[DML2_MAX_PLANES]; @@ -1001,10 +1148,10 @@ struct dml2_core_calcs_mode_programming_locals { double dummy_bw[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max]; double surface_dummy_bw[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max][DML2_MAX_PLANES]; double surface_dummy_bw0[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max][DML2_MAX_PLANES]; - unsigned int dummy_integer_array[2][DML2_MAX_PLANES]; + unsigned int dummy_integer_array[4][DML2_MAX_PLANES]; enum dml2_output_encoder_class dummy_output_encoder_array[DML2_MAX_PLANES]; double dummy_single_array[2][DML2_MAX_PLANES]; - unsigned int dummy_long_array[4][DML2_MAX_PLANES]; + unsigned int dummy_long_array[8][DML2_MAX_PLANES]; bool dummy_boolean_array[2][DML2_MAX_PLANES]; bool dummy_boolean[2]; double dummy_single[2]; @@ -1028,7 +1175,6 @@ struct dml2_core_calcs_mode_programming_locals { double dlg_vblank_start; double LSetup; double blank_lines_remaining; - double TotalWRBandwidth; double WRBandwidth; struct dml2_core_internal_DmlPipe myPipe; double PixelClockBackEndFactor; @@ -1078,6 +1224,8 @@ struct dml2_core_calcs_mode_programming_locals { enum dml2_source_format_class pixel_format[DML2_MAX_PLANES]; unsigned int lb_source_lines_l[DML2_MAX_PLANES]; unsigned int lb_source_lines_c[DML2_MAX_PLANES]; + unsigned int num_dsc_slices[DML2_MAX_PLANES]; + bool dsc_enable[DML2_MAX_PLANES]; }; struct dml2_core_calcs_CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_locals { @@ -1151,6 +1299,7 @@ struct dml2_core_calcs_CalculateVMRowAndSwath_params { unsigned int HostVMMinPageSize; unsigned int DCCMetaBufferSizeBytes; bool mrq_present; + enum dml2_pstate_method pstate_switch_modes[DML2_MAX_PLANES]; // Output bool *PTEBufferSizeNotExceeded; @@ -1387,7 +1536,7 @@ struct dml2_core_shared_get_urgent_bandwidth_required_locals { double vm_row_bw; double flip_and_active_bw; double flip_and_prefetch_bw; - double flip_and_prefetch_bw_oto; + double flip_and_prefetch_bw_max; double active_and_excess_bw; }; @@ -1416,6 +1565,7 @@ struct dml2_core_shared_CalculateFlipSchedule_locals { struct dml2_core_shared_rq_dlg_get_dlg_reg_locals { unsigned int plane_idx; + unsigned int stream_idx; enum dml2_source_format_class source_format; const struct dml2_timing_cfg *timing; bool dual_plane; @@ -1623,6 +1773,9 @@ struct dml2_core_calcs_CalculateSwathAndDETConfiguration_params { double *BytePerPixDETC; unsigned int *DPPPerSurface; bool mrq_present; + unsigned int dummy[2][DML2_MAX_PLANES]; + unsigned int swath_width_luma_ub_single_dpp[DML2_MAX_PLANES]; + unsigned int swath_width_chroma_ub_single_dpp[DML2_MAX_PLANES]; // output unsigned int *req_per_swath_ub_l; @@ -1640,6 +1793,8 @@ struct dml2_core_calcs_CalculateSwathAndDETConfiguration_params { unsigned int *DETBufferSizeC; unsigned int *full_swath_bytes_l; unsigned int *full_swath_bytes_c; + unsigned int *full_swath_bytes_single_dpp_l; + unsigned int *full_swath_bytes_single_dpp_c; bool *UnboundedRequestEnabled; unsigned int *compbuf_reserved_space_64b; unsigned int *CompressedBufferSizeInkByte; @@ -1799,7 +1954,7 @@ struct dml2_core_calcs_CalculatePrefetchSchedule_params { double *VRatioPrefetchC; double *RequiredPrefetchPixelDataBWLuma; double *RequiredPrefetchPixelDataBWChroma; - double *RequiredPrefetchBWOTO; + double *RequiredPrefetchBWMax; bool *NotEnoughTimeForDynamicMetadata; double *Tno_bw; double *Tno_bw_flip; @@ -2036,7 +2191,7 @@ struct dml2_core_calcs_calculate_peak_bandwidth_required_params { double *surface_read_bandwidth_c; double *prefetch_bandwidth_l; double *prefetch_bandwidth_c; - double *prefetch_bandwidth_oto; + double *prefetch_bandwidth_max; double *excess_vactive_fill_bw_l; double *excess_vactive_fill_bw_c; double *cursor_bw; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.c index 2504d9c2ec349..5f301befed167 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.c @@ -82,7 +82,7 @@ bool dml2_core_utils_is_420(enum dml2_source_format_class source_format) val = 0; break; default: - DML2_ASSERT(0); + DML_ASSERT(0); break; } return val; @@ -145,7 +145,7 @@ bool dml2_core_utils_is_422_planar(enum dml2_source_format_class source_format) val = 0; break; default: - DML2_ASSERT(0); + DML_ASSERT(0); break; } return val; @@ -208,7 +208,7 @@ bool dml2_core_utils_is_422_packed(enum dml2_source_format_class source_format) val = 1; break; default: - DML2_ASSERT(0); + DML_ASSERT(0); break; } return val; @@ -216,104 +216,104 @@ bool dml2_core_utils_is_422_packed(enum dml2_source_format_class source_format) void dml2_core_utils_print_mode_support_info(const struct dml2_core_internal_mode_support_info *support, bool fail_only) { - dml2_printf("DML: ===================================== \n"); - dml2_printf("DML: DML_MODE_SUPPORT_INFO_ST\n"); + DML_LOG_VERBOSE("DML: ===================================== \n"); + DML_LOG_VERBOSE("DML: DML_MODE_SUPPORT_INFO_ST\n"); if (!fail_only || support->ScaleRatioAndTapsSupport == 0) - dml2_printf("DML: support: ScaleRatioAndTapsSupport = %d\n", support->ScaleRatioAndTapsSupport); + DML_LOG_VERBOSE("DML: support: ScaleRatioAndTapsSupport = %d\n", support->ScaleRatioAndTapsSupport); if (!fail_only || support->SourceFormatPixelAndScanSupport == 0) - dml2_printf("DML: support: SourceFormatPixelAndScanSupport = %d\n", support->SourceFormatPixelAndScanSupport); + DML_LOG_VERBOSE("DML: support: SourceFormatPixelAndScanSupport = %d\n", support->SourceFormatPixelAndScanSupport); if (!fail_only || support->ViewportSizeSupport == 0) - dml2_printf("DML: support: ViewportSizeSupport = %d\n", support->ViewportSizeSupport); + DML_LOG_VERBOSE("DML: support: ViewportSizeSupport = %d\n", support->ViewportSizeSupport); if (!fail_only || support->LinkRateDoesNotMatchDPVersion == 1) - dml2_printf("DML: support: LinkRateDoesNotMatchDPVersion = %d\n", support->LinkRateDoesNotMatchDPVersion); + DML_LOG_VERBOSE("DML: support: LinkRateDoesNotMatchDPVersion = %d\n", support->LinkRateDoesNotMatchDPVersion); if (!fail_only || support->LinkRateForMultistreamNotIndicated == 1) - dml2_printf("DML: support: LinkRateForMultistreamNotIndicated = %d\n", support->LinkRateForMultistreamNotIndicated); + DML_LOG_VERBOSE("DML: support: LinkRateForMultistreamNotIndicated = %d\n", support->LinkRateForMultistreamNotIndicated); if (!fail_only || support->BPPForMultistreamNotIndicated == 1) - dml2_printf("DML: support: BPPForMultistreamNotIndicated = %d\n", support->BPPForMultistreamNotIndicated); + DML_LOG_VERBOSE("DML: support: BPPForMultistreamNotIndicated = %d\n", support->BPPForMultistreamNotIndicated); if (!fail_only || support->MultistreamWithHDMIOreDP == 1) - dml2_printf("DML: support: MultistreamWithHDMIOreDP = %d\n", support->MultistreamWithHDMIOreDP); + DML_LOG_VERBOSE("DML: support: MultistreamWithHDMIOreDP = %d\n", support->MultistreamWithHDMIOreDP); if (!fail_only || support->ExceededMultistreamSlots == 1) - dml2_printf("DML: support: ExceededMultistreamSlots = %d\n", support->ExceededMultistreamSlots); + DML_LOG_VERBOSE("DML: support: ExceededMultistreamSlots = %d\n", support->ExceededMultistreamSlots); if (!fail_only || support->MSOOrODMSplitWithNonDPLink == 1) - dml2_printf("DML: support: MSOOrODMSplitWithNonDPLink = %d\n", support->MSOOrODMSplitWithNonDPLink); + DML_LOG_VERBOSE("DML: support: MSOOrODMSplitWithNonDPLink = %d\n", support->MSOOrODMSplitWithNonDPLink); if (!fail_only || support->NotEnoughLanesForMSO == 1) - dml2_printf("DML: support: NotEnoughLanesForMSO = %d\n", support->NotEnoughLanesForMSO); + DML_LOG_VERBOSE("DML: support: NotEnoughLanesForMSO = %d\n", support->NotEnoughLanesForMSO); if (!fail_only || support->P2IWith420 == 1) - dml2_printf("DML: support: P2IWith420 = %d\n", support->P2IWith420); + DML_LOG_VERBOSE("DML: support: P2IWith420 = %d\n", support->P2IWith420); if (!fail_only || support->DSC422NativeNotSupported == 1) - dml2_printf("DML: support: DSC422NativeNotSupported = %d\n", support->DSC422NativeNotSupported); + DML_LOG_VERBOSE("DML: support: DSC422NativeNotSupported = %d\n", support->DSC422NativeNotSupported); if (!fail_only || support->DSCSlicesODMModeSupported == 0) - dml2_printf("DML: support: DSCSlicesODMModeSupported = %d\n", support->DSCSlicesODMModeSupported); + DML_LOG_VERBOSE("DML: support: DSCSlicesODMModeSupported = %d\n", support->DSCSlicesODMModeSupported); if (!fail_only || support->NotEnoughDSCUnits == 1) - dml2_printf("DML: support: NotEnoughDSCUnits = %d\n", support->NotEnoughDSCUnits); + DML_LOG_VERBOSE("DML: support: NotEnoughDSCUnits = %d\n", support->NotEnoughDSCUnits); if (!fail_only || support->NotEnoughDSCSlices == 1) - dml2_printf("DML: support: NotEnoughDSCSlices = %d\n", support->NotEnoughDSCSlices); + DML_LOG_VERBOSE("DML: support: NotEnoughDSCSlices = %d\n", support->NotEnoughDSCSlices); if (!fail_only || support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe == 1) - dml2_printf("DML: support: ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = %d\n", support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe); + DML_LOG_VERBOSE("DML: support: ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = %d\n", support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe); if (!fail_only || support->InvalidCombinationOfMALLUseForPStateAndStaticScreen == 1) - dml2_printf("DML: support: InvalidCombinationOfMALLUseForPStateAndStaticScreen = %d\n", support->InvalidCombinationOfMALLUseForPStateAndStaticScreen); + DML_LOG_VERBOSE("DML: support: InvalidCombinationOfMALLUseForPStateAndStaticScreen = %d\n", support->InvalidCombinationOfMALLUseForPStateAndStaticScreen); if (!fail_only || support->DSCCLKRequiredMoreThanSupported == 1) - dml2_printf("DML: support: DSCCLKRequiredMoreThanSupported = %d\n", support->DSCCLKRequiredMoreThanSupported); + DML_LOG_VERBOSE("DML: support: DSCCLKRequiredMoreThanSupported = %d\n", support->DSCCLKRequiredMoreThanSupported); if (!fail_only || support->PixelsPerLinePerDSCUnitSupport == 0) - dml2_printf("DML: support: PixelsPerLinePerDSCUnitSupport = %d\n", support->PixelsPerLinePerDSCUnitSupport); + DML_LOG_VERBOSE("DML: support: PixelsPerLinePerDSCUnitSupport = %d\n", support->PixelsPerLinePerDSCUnitSupport); if (!fail_only || support->DTBCLKRequiredMoreThanSupported == 1) - dml2_printf("DML: support: DTBCLKRequiredMoreThanSupported = %d\n", support->DTBCLKRequiredMoreThanSupported); + DML_LOG_VERBOSE("DML: support: DTBCLKRequiredMoreThanSupported = %d\n", support->DTBCLKRequiredMoreThanSupported); if (!fail_only || support->InvalidCombinationOfMALLUseForPState == 1) - dml2_printf("DML: support: InvalidCombinationOfMALLUseForPState = %d\n", support->InvalidCombinationOfMALLUseForPState); + DML_LOG_VERBOSE("DML: support: InvalidCombinationOfMALLUseForPState = %d\n", support->InvalidCombinationOfMALLUseForPState); if (!fail_only || support->ROBSupport == 0) - dml2_printf("DML: support: ROBSupport = %d\n", support->ROBSupport); + DML_LOG_VERBOSE("DML: support: ROBSupport = %d\n", support->ROBSupport); if (!fail_only || support->OutstandingRequestsSupport == 0) - dml2_printf("DML: support: OutstandingRequestsSupport = %d\n", support->OutstandingRequestsSupport); + DML_LOG_VERBOSE("DML: support: OutstandingRequestsSupport = %d\n", support->OutstandingRequestsSupport); if (!fail_only || support->OutstandingRequestsUrgencyAvoidance == 0) - dml2_printf("DML: support: OutstandingRequestsUrgencyAvoidance = %d\n", support->OutstandingRequestsUrgencyAvoidance); + DML_LOG_VERBOSE("DML: support: OutstandingRequestsUrgencyAvoidance = %d\n", support->OutstandingRequestsUrgencyAvoidance); if (!fail_only || support->DISPCLK_DPPCLK_Support == 0) - dml2_printf("DML: support: DISPCLK_DPPCLK_Support = %d\n", support->DISPCLK_DPPCLK_Support); + DML_LOG_VERBOSE("DML: support: DISPCLK_DPPCLK_Support = %d\n", support->DISPCLK_DPPCLK_Support); if (!fail_only || support->TotalAvailablePipesSupport == 0) - dml2_printf("DML: support: TotalAvailablePipesSupport = %d\n", support->TotalAvailablePipesSupport); + DML_LOG_VERBOSE("DML: support: TotalAvailablePipesSupport = %d\n", support->TotalAvailablePipesSupport); if (!fail_only || support->NumberOfOTGSupport == 0) - dml2_printf("DML: support: NumberOfOTGSupport = %d\n", support->NumberOfOTGSupport); + DML_LOG_VERBOSE("DML: support: NumberOfOTGSupport = %d\n", support->NumberOfOTGSupport); if (!fail_only || support->NumberOfHDMIFRLSupport == 0) - dml2_printf("DML: support: NumberOfHDMIFRLSupport = %d\n", support->NumberOfHDMIFRLSupport); + DML_LOG_VERBOSE("DML: support: NumberOfHDMIFRLSupport = %d\n", support->NumberOfHDMIFRLSupport); if (!fail_only || support->NumberOfDP2p0Support == 0) - dml2_printf("DML: support: NumberOfDP2p0Support = %d\n", support->NumberOfDP2p0Support); + DML_LOG_VERBOSE("DML: support: NumberOfDP2p0Support = %d\n", support->NumberOfDP2p0Support); if (!fail_only || support->EnoughWritebackUnits == 0) - dml2_printf("DML: support: EnoughWritebackUnits = %d\n", support->EnoughWritebackUnits); + DML_LOG_VERBOSE("DML: support: EnoughWritebackUnits = %d\n", support->EnoughWritebackUnits); if (!fail_only || support->WritebackScaleRatioAndTapsSupport == 0) - dml2_printf("DML: support: WritebackScaleRatioAndTapsSupport = %d\n", support->WritebackScaleRatioAndTapsSupport); + DML_LOG_VERBOSE("DML: support: WritebackScaleRatioAndTapsSupport = %d\n", support->WritebackScaleRatioAndTapsSupport); if (!fail_only || support->WritebackLatencySupport == 0) - dml2_printf("DML: support: WritebackLatencySupport = %d\n", support->WritebackLatencySupport); + DML_LOG_VERBOSE("DML: support: WritebackLatencySupport = %d\n", support->WritebackLatencySupport); if (!fail_only || support->CursorSupport == 0) - dml2_printf("DML: support: CursorSupport = %d\n", support->CursorSupport); + DML_LOG_VERBOSE("DML: support: CursorSupport = %d\n", support->CursorSupport); if (!fail_only || support->PitchSupport == 0) - dml2_printf("DML: support: PitchSupport = %d\n", support->PitchSupport); + DML_LOG_VERBOSE("DML: support: PitchSupport = %d\n", support->PitchSupport); if (!fail_only || support->ViewportExceedsSurface == 1) - dml2_printf("DML: support: ViewportExceedsSurface = %d\n", support->ViewportExceedsSurface); + DML_LOG_VERBOSE("DML: support: ViewportExceedsSurface = %d\n", support->ViewportExceedsSurface); if (!fail_only || support->PrefetchSupported == 0) - dml2_printf("DML: support: PrefetchSupported = %d\n", support->PrefetchSupported); + DML_LOG_VERBOSE("DML: support: PrefetchSupported = %d\n", support->PrefetchSupported); if (!fail_only || support->EnoughUrgentLatencyHidingSupport == 0) - dml2_printf("DML: support: EnoughUrgentLatencyHidingSupport = %d\n", support->EnoughUrgentLatencyHidingSupport); + DML_LOG_VERBOSE("DML: support: EnoughUrgentLatencyHidingSupport = %d\n", support->EnoughUrgentLatencyHidingSupport); if (!fail_only || support->AvgBandwidthSupport == 0) - dml2_printf("DML: support: AvgBandwidthSupport = %d\n", support->AvgBandwidthSupport); + DML_LOG_VERBOSE("DML: support: AvgBandwidthSupport = %d\n", support->AvgBandwidthSupport); if (!fail_only || support->DynamicMetadataSupported == 0) - dml2_printf("DML: support: DynamicMetadataSupported = %d\n", support->DynamicMetadataSupported); + DML_LOG_VERBOSE("DML: support: DynamicMetadataSupported = %d\n", support->DynamicMetadataSupported); if (!fail_only || support->VRatioInPrefetchSupported == 0) - dml2_printf("DML: support: VRatioInPrefetchSupported = %d\n", support->VRatioInPrefetchSupported); + DML_LOG_VERBOSE("DML: support: VRatioInPrefetchSupported = %d\n", support->VRatioInPrefetchSupported); if (!fail_only || support->PTEBufferSizeNotExceeded == 0) - dml2_printf("DML: support: PTEBufferSizeNotExceeded = %d\n", support->PTEBufferSizeNotExceeded); + DML_LOG_VERBOSE("DML: support: PTEBufferSizeNotExceeded = %d\n", support->PTEBufferSizeNotExceeded); if (!fail_only || support->DCCMetaBufferSizeNotExceeded == 0) - dml2_printf("DML: support: DCCMetaBufferSizeNotExceeded = %d\n", support->DCCMetaBufferSizeNotExceeded); + DML_LOG_VERBOSE("DML: support: DCCMetaBufferSizeNotExceeded = %d\n", support->DCCMetaBufferSizeNotExceeded); if (!fail_only || support->ExceededMALLSize == 1) - dml2_printf("DML: support: ExceededMALLSize = %d\n", support->ExceededMALLSize); + DML_LOG_VERBOSE("DML: support: ExceededMALLSize = %d\n", support->ExceededMALLSize); if (!fail_only || support->g6_temp_read_support == 0) - dml2_printf("DML: support: g6_temp_read_support = %d\n", support->g6_temp_read_support); + DML_LOG_VERBOSE("DML: support: g6_temp_read_support = %d\n", support->g6_temp_read_support); if (!fail_only || support->ImmediateFlipSupport == 0) - dml2_printf("DML: support: ImmediateFlipSupport = %d\n", support->ImmediateFlipSupport); + DML_LOG_VERBOSE("DML: support: ImmediateFlipSupport = %d\n", support->ImmediateFlipSupport); if (!fail_only || support->LinkCapacitySupport == 0) - dml2_printf("DML: support: LinkCapacitySupport = %d\n", support->LinkCapacitySupport); + DML_LOG_VERBOSE("DML: support: LinkCapacitySupport = %d\n", support->LinkCapacitySupport); if (!fail_only || support->ModeSupport == 0) - dml2_printf("DML: support: ModeSupport = %d\n", support->ModeSupport); - dml2_printf("DML: ===================================== \n"); + DML_LOG_VERBOSE("DML: support: ModeSupport = %d\n", support->ModeSupport); + DML_LOG_VERBOSE("DML: ===================================== \n"); } const char *dml2_core_utils_internal_soc_state_type_str(enum dml2_core_internal_soc_state_type dml2_core_internal_soc_state_type) @@ -358,9 +358,9 @@ void dml2_core_utils_get_stream_output_bpp(double *out_bpp, const struct dml2_di out_bpp[k] = 0; } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: k=%d bpc=%f\n", __func__, k, bpc); - dml2_printf("DML::%s: k=%d dsc.enable=%d\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable); - dml2_printf("DML::%s: k=%d out_bpp=%f\n", __func__, k, out_bpp[k]); + DML_LOG_VERBOSE("DML::%s: k=%d bpc=%f\n", __func__, k, bpc); + DML_LOG_VERBOSE("DML::%s: k=%d dsc.enable=%d\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable); + DML_LOG_VERBOSE("DML::%s: k=%d out_bpp=%f\n", __func__, k, out_bpp[k]); #endif } } @@ -391,7 +391,7 @@ unsigned int dml2_core_util_get_num_active_pipes(int unsigned num_planes, const } #ifdef __DML_VBA_DEBUG__ - dml2_printf("DML::%s: num_active_pipes = %d\n", __func__, num_active_pipes); + DML_LOG_VERBOSE("DML::%s: num_active_pipes = %d\n", __func__, num_active_pipes); #endif return num_active_pipes; } @@ -452,7 +452,7 @@ unsigned int dml2_core_utils_get_tile_block_size_bytes(enum dml2_swizzle_mode sw else if (sw_mode == dml2_gfx11_sw_256kb_r_x) return 262144; else { - DML2_ASSERT(0); + DML_ASSERT(0); return 256; }; } @@ -464,7 +464,7 @@ bool dml2_core_utils_get_segment_horizontal_contiguous(enum dml2_swizzle_mode sw bool dml2_core_utils_is_linear(enum dml2_swizzle_mode sw_mode) { - return (sw_mode == dml2_sw_linear || sw_mode == dml2_sw_linear_256b || sw_mode == dml2_linear_64elements); + return sw_mode == dml2_sw_linear; }; @@ -498,8 +498,8 @@ int unsigned dml2_core_utils_get_gfx_version(enum dml2_swizzle_mode sw_mode) sw_mode == dml2_gfx11_sw_256kb_r_x) version = 11; else { - dml2_printf("ERROR: Invalid sw_mode setting! val=%u\n", sw_mode); - DML2_ASSERT(0); + DML_LOG_VERBOSE("ERROR: Invalid sw_mode setting! val=%u\n", sw_mode); + DML_ASSERT(0); } return version; @@ -511,7 +511,7 @@ unsigned int dml2_core_utils_get_qos_param_index(unsigned long uclk_freq_khz, co unsigned int index = 0; for (i = 0; i < DML_MAX_CLK_TABLE_SIZE; i++) { - dml2_printf("DML::%s: per_uclk_dpm_params[%d].minimum_uclk_khz = %d\n", __func__, i, per_uclk_dpm_params[i].minimum_uclk_khz); + DML_LOG_VERBOSE("DML::%s: per_uclk_dpm_params[%d].minimum_uclk_khz = %ld\n", __func__, i, per_uclk_dpm_params[i].minimum_uclk_khz); if (i == 0) index = 0; @@ -524,8 +524,8 @@ unsigned int dml2_core_utils_get_qos_param_index(unsigned long uclk_freq_khz, co } } #if defined(__DML_VBA_DEBUG__) - dml2_printf("DML::%s: uclk_freq_khz = %d\n", __func__, uclk_freq_khz); - dml2_printf("DML::%s: index = %d\n", __func__, index); + DML_LOG_VERBOSE("DML::%s: uclk_freq_khz = %ld\n", __func__, uclk_freq_khz); + DML_LOG_VERBOSE("DML::%s: index = %d\n", __func__, index); #endif return index; } @@ -533,32 +533,32 @@ unsigned int dml2_core_utils_get_qos_param_index(unsigned long uclk_freq_khz, co unsigned int dml2_core_utils_get_active_min_uclk_dpm_index(unsigned long uclk_freq_khz, const struct dml2_soc_state_table *clk_table) { unsigned int i; - bool clk_entry_found = 0; + bool clk_entry_found = false; for (i = 0; i < clk_table->uclk.num_clk_values; i++) { - dml2_printf("DML::%s: clk_table.uclk.clk_values_khz[%d] = %d\n", __func__, i, clk_table->uclk.clk_values_khz[i]); + DML_LOG_VERBOSE("DML::%s: clk_table.uclk.clk_values_khz[%d] = %ld\n", __func__, i, clk_table->uclk.clk_values_khz[i]); if (uclk_freq_khz == clk_table->uclk.clk_values_khz[i]) { - clk_entry_found = 1; + clk_entry_found = true; break; } } if (!clk_entry_found) - DML2_ASSERT(clk_entry_found); + DML_ASSERT(clk_entry_found); #if defined(__DML_VBA_DEBUG__) - dml2_printf("DML::%s: uclk_freq_khz = %ld\n", __func__, uclk_freq_khz); - dml2_printf("DML::%s: index = %d\n", __func__, i); + DML_LOG_VERBOSE("DML::%s: uclk_freq_khz = %ld\n", __func__, uclk_freq_khz); + DML_LOG_VERBOSE("DML::%s: index = %d\n", __func__, i); #endif return i; } bool dml2_core_utils_is_dual_plane(enum dml2_source_format_class source_format) { - bool ret_val = 0; + bool ret_val = false; if (dml2_core_utils_is_420(source_format) || dml2_core_utils_is_422_planar(source_format) || (source_format == dml2_rgbe_alpha)) - ret_val = 1; + ret_val = true; return ret_val; } diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c index 15507926f3a41..22969a533a7b5 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c @@ -389,9 +389,6 @@ static bool map_min_clocks_to_dpm(const struct dml2_core_mode_support_result *mo if (result) result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.dispclk_khz, &state_table->dispclk); - if (result) - result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.deepsleep_dcfclk_khz, &state_table->dcfclk); - for (i = 0; i < DML2_MAX_DCN_PIPES; i++) { if (result) result = round_up_to_next_dpm(&display_cfg->plane_programming[i].min_clocks.dcn4x.dppclk_khz, &state_table->dppclk); @@ -754,6 +751,8 @@ bool dpmm_dcn4_map_watermarks(struct dml2_dpmm_map_watermarks_params_in_out *in_ dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].fclk_pstate = (int unsigned)(mode_lib->mp.Watermark.FCLKChangeWatermark * refclk_freq_in_mhz); dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].sr_enter = (int unsigned)(mode_lib->mp.Watermark.StutterEnterPlusExitWatermark * refclk_freq_in_mhz); dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].sr_exit = (int unsigned)(mode_lib->mp.Watermark.StutterExitWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].sr_enter_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterEnterPlusExitWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].sr_exit_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterExitWatermark * refclk_freq_in_mhz); dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].temp_read_or_ppt = (int unsigned)(mode_lib->mp.Watermark.temp_read_or_ppt_watermark_us * refclk_freq_in_mhz); dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].uclk_pstate = (int unsigned)(mode_lib->mp.Watermark.DRAMClockChangeWatermark * refclk_freq_in_mhz); dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].urgent = (int unsigned)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); @@ -768,6 +767,8 @@ bool dpmm_dcn4_map_watermarks(struct dml2_dpmm_map_watermarks_params_in_out *in_ dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].fclk_pstate = (int unsigned)(mode_lib->mp.Watermark.FCLKChangeWatermark * refclk_freq_in_mhz); dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].sr_enter = (int unsigned)(mode_lib->mp.Watermark.StutterEnterPlusExitWatermark * refclk_freq_in_mhz); dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].sr_exit = (int unsigned)(mode_lib->mp.Watermark.StutterExitWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].sr_enter_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterEnterPlusExitWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].sr_exit_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterExitWatermark * refclk_freq_in_mhz); dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].temp_read_or_ppt = (int unsigned)(mode_lib->mp.Watermark.temp_read_or_ppt_watermark_us * refclk_freq_in_mhz); dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].uclk_pstate = (int unsigned)(mode_lib->mp.Watermark.DRAMClockChangeWatermark * refclk_freq_in_mhz); dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].urgent = (int unsigned)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.c index f4b1a7d02d426..a265f254152c7 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.c @@ -182,6 +182,10 @@ static bool build_min_clock_table(const struct dml2_soc_bb *soc_bb, struct dml2_ min_table->max_clocks_khz.dtbclk = soc_bb->clk_table.dtbclk.clk_values_khz[soc_bb->clk_table.dtbclk.num_clk_values - 1]; min_table->max_clocks_khz.phyclk = soc_bb->clk_table.phyclk.clk_values_khz[soc_bb->clk_table.phyclk.num_clk_values - 1]; + min_table->max_ss_clocks_khz.dispclk = (unsigned int)((double)min_table->max_clocks_khz.dispclk / (1.0 + soc_bb->dcn_downspread_percent / 100.0)); + min_table->max_ss_clocks_khz.dppclk = (unsigned int)((double)min_table->max_clocks_khz.dppclk / (1.0 + soc_bb->dcn_downspread_percent / 100.0)); + min_table->max_ss_clocks_khz.dtbclk = (unsigned int)((double)min_table->max_clocks_khz.dtbclk / (1.0 + soc_bb->dcn_downspread_percent / 100.0)); + min_table->max_clocks_khz.dcfclk = soc_bb->clk_table.dcfclk.clk_values_khz[soc_bb->clk_table.dcfclk.num_clk_values - 1]; min_table->max_clocks_khz.fclk = soc_bb->clk_table.fclk.clk_values_khz[soc_bb->clk_table.fclk.num_clk_values - 1]; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c index f50662b832966..d88b3e0082dd4 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c @@ -659,7 +659,7 @@ bool pmo_dcn4_fams2_initialize(struct dml2_pmo_initialize_in_out *in_out) for (i = 1; i <= PMO_DCN4_MAX_DISPLAYS; i++) { switch (i) { case 1: - DML2_ASSERT(base_strategy_list_1_display_size <= PMO_DCN4_MAX_BASE_STRATEGIES); + DML_ASSERT(base_strategy_list_1_display_size <= PMO_DCN4_MAX_BASE_STRATEGIES); /* populate list */ pmo_dcn4_fams2_expand_base_pstate_strategies( @@ -670,7 +670,7 @@ bool pmo_dcn4_fams2_initialize(struct dml2_pmo_initialize_in_out *in_out) &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i - 1]); break; case 2: - DML2_ASSERT(base_strategy_list_2_display_size <= PMO_DCN4_MAX_BASE_STRATEGIES); + DML_ASSERT(base_strategy_list_2_display_size <= PMO_DCN4_MAX_BASE_STRATEGIES); /* populate list */ pmo_dcn4_fams2_expand_base_pstate_strategies( @@ -681,7 +681,7 @@ bool pmo_dcn4_fams2_initialize(struct dml2_pmo_initialize_in_out *in_out) &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i - 1]); break; case 3: - DML2_ASSERT(base_strategy_list_3_display_size <= PMO_DCN4_MAX_BASE_STRATEGIES); + DML_ASSERT(base_strategy_list_3_display_size <= PMO_DCN4_MAX_BASE_STRATEGIES); /* populate list */ pmo_dcn4_fams2_expand_base_pstate_strategies( @@ -692,7 +692,7 @@ bool pmo_dcn4_fams2_initialize(struct dml2_pmo_initialize_in_out *in_out) &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i - 1]); break; case 4: - DML2_ASSERT(base_strategy_list_4_display_size <= PMO_DCN4_MAX_BASE_STRATEGIES); + DML_ASSERT(base_strategy_list_4_display_size <= PMO_DCN4_MAX_BASE_STRATEGIES); /* populate list */ pmo_dcn4_fams2_expand_base_pstate_strategies( diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c index dc2ce5e77f579..4a7c4c62111e7 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c @@ -761,7 +761,7 @@ bool dml2_top_mcache_calc_mcache_count_and_offsets(struct top_mcache_calc_mcache total_mcaches_required--; } } - dml2_printf("DML_CORE_DCN3::%s: plane_%d, total_mcaches_required=%d\n", __func__, i, total_mcaches_required); + DML_LOG_VERBOSE("DML_CORE_DCN3::%s: plane_%d, total_mcaches_required=%d\n", __func__, i, total_mcaches_required); if (total_mcaches_required > dml->soc_bbox.num_dcc_mcaches) { result = false; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.c deleted file mode 100644 index c506667897c4c..0000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.c +++ /dev/null @@ -1,31 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "dml2_debug.h" - -int dml2_log_internal(const char *format, ...) -{ - return 0; -} - -int dml2_printf(const char *format, ...) -{ -#ifdef _DEBUG -#ifdef _DEBUG_PRINTS - int result; - va_list args; - va_start(args, format); - - result = vprintf(format, args); - - va_end(args); - - return result; -#else - return 0; -#endif -#else - return 0; -#endif -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.h index bfe6f236d2e44..611c80f4f1bf1 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.h @@ -5,55 +5,185 @@ #ifndef __DML2_DEBUG_H__ #define __DML2_DEBUG_H__ -#ifndef DML2_ASSERT -#define DML2_ASSERT(condition) ((void)0) -#endif +#include "os_types.h" +#define DML_ASSERT(condition) ASSERT(condition) +#define DML_LOG_LEVEL_DEFAULT DML_LOG_LEVEL_WARN +#define DML_LOG_INTERNAL(fmt, ...) dm_output_to_console(fmt, ## __VA_ARGS__) -/* - * DML_LOG_FATAL - fatal errors for unrecoverable DML states until a restart. - * DML_LOG_ERROR - unexpected but recoverable failures inside DML - * DML_LOG_WARN - unexpected inputs or events to DML - * DML_LOG_INFO - high level tracing of DML interfaces - * DML_LOG_DEBUG - detailed tracing of DML internal components - * DML_LOG_VERBOSE - detailed tracing of DML calculation procedure - */ -#if !defined(DML_LOG_LEVEL) -#if defined(_DEBUG) && defined(_DEBUG_PRINTS) -/* for backward compatibility with old macros */ -#define DML_LOG_LEVEL 5 -#else -#define DML_LOG_LEVEL 0 -#endif -#endif +/* private helper macros */ +#define _BOOL_FORMAT(field) "%s", field ? "true" : "false" +#define _UINT_FORMAT(field) "%u", field +#define _INT_FORMAT(field) "%d", field +#define _DOUBLE_FORMAT(field) "%lf", field +#define _ELEMENT_FUNC "function" +#define _ELEMENT_COMP_IF "component_interface" +#define _ELEMENT_TOP_IF "top_interface" +#define _LOG_ENTRY(element) do { \ + DML_LOG_INTERNAL("<"element" name=\""); \ + DML_LOG_INTERNAL(__func__); \ + DML_LOG_INTERNAL("\">\n"); \ +} while (0) +#define _LOG_EXIT(element) DML_LOG_INTERNAL("</"element">\n") +#define _LOG_SCALAR(field, format) do { \ + DML_LOG_INTERNAL(#field" = "format(field)); \ + DML_LOG_INTERNAL("\n"); \ +} while (0) +#define _LOG_ARRAY(field, size, format) do { \ + DML_LOG_INTERNAL(#field " = ["); \ + for (int _i = 0; _i < (int) size; _i++) { \ + DML_LOG_INTERNAL(format(field[_i])); \ + if (_i + 1 == (int) size) \ + DML_LOG_INTERNAL("]\n"); \ + else \ + DML_LOG_INTERNAL(", "); \ +}} while (0) +#define _LOG_2D_ARRAY(field, size0, size1, format) do { \ + DML_LOG_INTERNAL(#field" = ["); \ + for (int _i = 0; _i < (int) size0; _i++) { \ + DML_LOG_INTERNAL("\n\t["); \ + for (int _j = 0; _j < (int) size1; _j++) { \ + DML_LOG_INTERNAL(format(field[_i][_j])); \ + if (_j + 1 == (int) size1) \ + DML_LOG_INTERNAL("]"); \ + else \ + DML_LOG_INTERNAL(", "); \ + } \ + if (_i + 1 == (int) size0) \ + DML_LOG_INTERNAL("]\n"); \ + else \ + DML_LOG_INTERNAL(", "); \ + } \ +} while (0) +#define _LOG_3D_ARRAY(field, size0, size1, size2, format) do { \ + DML_LOG_INTERNAL(#field" = ["); \ + for (int _i = 0; _i < (int) size0; _i++) { \ + DML_LOG_INTERNAL("\n\t["); \ + for (int _j = 0; _j < (int) size1; _j++) { \ + DML_LOG_INTERNAL("["); \ + for (int _k = 0; _k < (int) size2; _k++) { \ + DML_LOG_INTERNAL(format(field[_i][_j][_k])); \ + if (_k + 1 == (int) size2) \ + DML_LOG_INTERNAL("]"); \ + else \ + DML_LOG_INTERNAL(", "); \ + } \ + if (_j + 1 == (int) size1) \ + DML_LOG_INTERNAL("]"); \ + else \ + DML_LOG_INTERNAL(", "); \ + } \ + if (_i + 1 == (int) size0) \ + DML_LOG_INTERNAL("]\n"); \ + else \ + DML_LOG_INTERNAL(", "); \ + } \ +} while (0) -#define DML_LOG_FATAL(fmt, ...) dml2_log_internal(fmt, ## __VA_ARGS__) -#if DML_LOG_LEVEL >= 1 -#define DML_LOG_ERROR(fmt, ...) dml2_log_internal(fmt, ## __VA_ARGS__) +/* fatal errors for unrecoverable DML states until a full reset */ +#define DML_LOG_LEVEL_FATAL 0 +/* unexpected but recoverable failures inside DML */ +#define DML_LOG_LEVEL_ERROR 1 +/* unexpected inputs or events to DML */ +#define DML_LOG_LEVEL_WARN 2 +/* high level tracing of DML interfaces */ +#define DML_LOG_LEVEL_INFO 3 +/* tracing of DML internal executions */ +#define DML_LOG_LEVEL_DEBUG 4 +/* detailed tracing of DML calculation procedure */ +#define DML_LOG_LEVEL_VERBOSE 5 + +#ifndef DML_LOG_LEVEL +#define DML_LOG_LEVEL DML_LOG_LEVEL_DEFAULT +#endif /* #ifndef DML_LOG_LEVEL */ + +/* public macros for DML_LOG_LEVEL_FATAL and up */ +#define DML_LOG_FATAL(fmt, ...) DML_LOG_INTERNAL("[DML FATAL] " fmt, ## __VA_ARGS__) + +/* public macros for DML_LOG_LEVEL_ERROR and up */ +#if DML_LOG_LEVEL >= DML_LOG_LEVEL_ERROR +#define DML_LOG_ERROR(fmt, ...) DML_LOG_INTERNAL("[DML ERROR] "fmt, ## __VA_ARGS__) +#define DML_ASSERT_MSG(condition, fmt, ...) \ + do { \ + if (!(condition)) { \ + DML_LOG_ERROR("ASSERT hit in %s line %d\n", __func__, __LINE__); \ + DML_LOG_ERROR(fmt, ## __VA_ARGS__); \ + DML_ASSERT(condition); \ + } \ + } while (0) #else #define DML_LOG_ERROR(fmt, ...) ((void)0) +#define DML_ASSERT_MSG(condition, fmt, ...) ((void)0) #endif -#if DML_LOG_LEVEL >= 2 -#define DML_LOG_WARN(fmt, ...) dml2_log_internal(fmt, ## __VA_ARGS__) + +/* public macros for DML_LOG_LEVEL_WARN and up */ +#if DML_LOG_LEVEL >= DML_LOG_LEVEL_WARN +#define DML_LOG_WARN(fmt, ...) DML_LOG_INTERNAL("[DML WARN] "fmt, ## __VA_ARGS__) #else #define DML_LOG_WARN(fmt, ...) ((void)0) #endif -#if DML_LOG_LEVEL >= 3 -#define DML_LOG_INFO(fmt, ...) dml2_log_internal(fmt, ## __VA_ARGS__) + +/* public macros for DML_LOG_LEVEL_INFO and up */ +#if DML_LOG_LEVEL >= DML_LOG_LEVEL_INFO +#define DML_LOG_INFO(fmt, ...) DML_LOG_INTERNAL("[DML INFO] "fmt, ## __VA_ARGS__) +#define DML_LOG_TOP_IF_ENTER() _LOG_ENTRY(_ELEMENT_TOP_IF) +#define DML_LOG_TOP_IF_EXIT() _LOG_EXIT(_ELEMENT_TOP_IF) #else #define DML_LOG_INFO(fmt, ...) ((void)0) +#define DML_LOG_TOP_IF_ENTER() ((void)0) +#define DML_LOG_TOP_IF_EXIT() ((void)0) #endif -#if DML_LOG_LEVEL >= 4 -#define DML_LOG_DEBUG(fmt, ...) dml2_log_internal(fmt, ## __VA_ARGS__) + +/* public macros for DML_LOG_LEVEL_DEBUG and up */ +#if DML_LOG_LEVEL >= DML_LOG_LEVEL_DEBUG +#define DML_LOG_DEBUG(fmt, ...) DML_LOG_INTERNAL(fmt, ## __VA_ARGS__) +#define DML_LOG_COMP_IF_ENTER() _LOG_ENTRY(_ELEMENT_COMP_IF) +#define DML_LOG_COMP_IF_EXIT() _LOG_EXIT(_ELEMENT_COMP_IF) +#define DML_LOG_FUNC_ENTER() _LOG_ENTRY(_ELEMENT_FUNC) +#define DML_LOG_FUNC_EXIT() _LOG_EXIT(_ELEMENT_FUNC) +#define DML_LOG_DEBUG_BOOL(field) _LOG_SCALAR(field, _BOOL_FORMAT) +#define DML_LOG_DEBUG_UINT(field) _LOG_SCALAR(field, _UINT_FORMAT) +#define DML_LOG_DEBUG_INT(field) _LOG_SCALAR(field, _INT_FORMAT) +#define DML_LOG_DEBUG_DOUBLE(field) _LOG_SCALAR(field, _DOUBLE_FORMAT) +#define DML_LOG_DEBUG_ARRAY_BOOL(field, size) _LOG_ARRAY(field, size, _BOOL_FORMAT) +#define DML_LOG_DEBUG_ARRAY_UINT(field, size) _LOG_ARRAY(field, size, _UINT_FORMAT) +#define DML_LOG_DEBUG_ARRAY_INT(field, size) _LOG_ARRAY(field, size, _INT_FORMAT) +#define DML_LOG_DEBUG_ARRAY_DOUBLE(field, size) _LOG_ARRAY(field, size, _DOUBLE_FORMAT) +#define DML_LOG_DEBUG_2D_ARRAY_BOOL(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _BOOL_FORMAT) +#define DML_LOG_DEBUG_2D_ARRAY_UINT(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _UINT_FORMAT) +#define DML_LOG_DEBUG_2D_ARRAY_INT(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _INT_FORMAT) +#define DML_LOG_DEBUG_2D_ARRAY_DOUBLE(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _DOUBLE_FORMAT) +#define DML_LOG_DEBUG_3D_ARRAY_BOOL(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _BOOL_FORMAT) +#define DML_LOG_DEBUG_3D_ARRAY_UINT(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _UINT_FORMAT) +#define DML_LOG_DEBUG_3D_ARRAY_INT(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _INT_FORMAT) +#define DML_LOG_DEBUG_3D_ARRAY_DOUBLE(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _DOUBLE_FORMAT) #else #define DML_LOG_DEBUG(fmt, ...) ((void)0) +#define DML_LOG_COMP_IF_ENTER() ((void)0) +#define DML_LOG_COMP_IF_EXIT() ((void)0) +#define DML_LOG_FUNC_ENTER() ((void)0) +#define DML_LOG_FUNC_EXIT() ((void)0) +#define DML_LOG_DEBUG_BOOL(field) ((void)0) +#define DML_LOG_DEBUG_UINT(field) ((void)0) +#define DML_LOG_DEBUG_INT(field) ((void)0) +#define DML_LOG_DEBUG_DOUBLE(field) ((void)0) +#define DML_LOG_DEBUG_ARRAY_BOOL(field, size) ((void)0) +#define DML_LOG_DEBUG_ARRAY_UINT(field, size) ((void)0) +#define DML_LOG_DEBUG_ARRAY_INT(field, size) ((void)0) +#define DML_LOG_DEBUG_ARRAY_DOUBLE(field, size) ((void)0) +#define DML_LOG_DEBUG_2D_ARRAY_BOOL(field, size0, size1) ((void)0) +#define DML_LOG_DEBUG_2D_ARRAY_UINT(field, size0, size1) ((void)0) +#define DML_LOG_DEBUG_2D_ARRAY_INT(field, size0, size1) ((void)0) +#define DML_LOG_DEBUG_2D_ARRAY_DOUBLE(field, size0, size1) ((void)0) +#define DML_LOG_DEBUG_3D_ARRAY_BOOL(field, size0, size1, size2) ((void)0) +#define DML_LOG_DEBUG_3D_ARRAY_UINT(field, size0, size1, size2) ((void)0) +#define DML_LOG_DEBUG_3D_ARRAY_INT(field, size0, size1, size2) ((void)0) +#define DML_LOG_DEBUG_3D_ARRAY_DOUBLE(field, size0, size1, size2) ((void)0) #endif -#if DML_LOG_LEVEL >= 5 -#define DML_LOG_VERBOSE(fmt, ...) dml2_log_internal(fmt, ## __VA_ARGS__) + +/* public macros for DML_LOG_LEVEL_VERBOSE */ +#if DML_LOG_LEVEL >= DML_LOG_LEVEL_VERBOSE +#define DML_LOG_VERBOSE(fmt, ...) DML_LOG_INTERNAL(fmt, ## __VA_ARGS__) #else #define DML_LOG_VERBOSE(fmt, ...) ((void)0) -#endif - -int dml2_log_internal(const char *format, ...); -int dml2_printf(const char *format, ...); - -#endif +#endif /* #if DML_LOG_LEVEL >= DML_LOG_LEVEL_VERBOSE */ +#endif /* __DML2_DEBUG_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_internal_shared_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_internal_shared_types.h index d8d01dceacdd4..d52aa82283b36 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_internal_shared_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_internal_shared_types.h @@ -38,6 +38,12 @@ struct dml2_mcg_min_clock_table { } max_clocks_khz; struct { + unsigned int dispclk; + unsigned int dppclk; + unsigned int dtbclk; + } max_ss_clocks_khz; + + struct { unsigned int dprefclk; unsigned int xtalclk; unsigned int pcierefclk; @@ -196,6 +202,8 @@ struct dml2_core_mode_support_result { } active; unsigned int dispclk_khz; + unsigned int dpprefclk_khz; + unsigned int dtbrefclk_khz; unsigned int dcfclk_deepsleep_khz; unsigned int socclk_khz; @@ -440,13 +448,17 @@ struct dml2_core_internal_state_intermediates { }; struct dml2_core_mode_support_locals { - struct dml2_core_calcs_mode_support_ex mode_support_ex_params; + union { + struct dml2_core_calcs_mode_support_ex mode_support_ex_params; + }; struct dml2_display_cfg svp_expanded_display_cfg; struct dml2_calculate_mcache_allocation_in_out calc_mcache_allocation_params; }; struct dml2_core_mode_programming_locals { - struct dml2_core_calcs_mode_programming_ex mode_programming_ex_params; + union { + struct dml2_core_calcs_mode_programming_ex mode_programming_ex_params; + }; struct dml2_display_cfg svp_expanded_display_cfg; }; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.c index a966abd407881..5f1b49a500495 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.c @@ -1082,22 +1082,22 @@ bool dml2_map_dc_pipes(struct dml2_context *ctx, struct dc_state *state, const s if (stream_disp_cfg_index >= disp_cfg_index_max) continue; - if (ODMMode[stream_disp_cfg_index] == dml_odm_mode_bypass) { - scratch.odm_info.odm_factor = 1; - } else if (ODMMode[stream_disp_cfg_index] == dml_odm_mode_combine_2to1) { - scratch.odm_info.odm_factor = 2; - } else if (ODMMode[stream_disp_cfg_index] == dml_odm_mode_combine_4to1) { - scratch.odm_info.odm_factor = 4; - } else { - ASSERT(false); - scratch.odm_info.odm_factor = 1; - } - + if (ctx->architecture == dml2_architecture_20) { + if (ODMMode[stream_disp_cfg_index] == dml_odm_mode_bypass) { + scratch.odm_info.odm_factor = 1; + } else if (ODMMode[stream_disp_cfg_index] == dml_odm_mode_combine_2to1) { + scratch.odm_info.odm_factor = 2; + } else if (ODMMode[stream_disp_cfg_index] == dml_odm_mode_combine_4to1) { + scratch.odm_info.odm_factor = 4; + } else { + ASSERT(false); + scratch.odm_info.odm_factor = 1; + } + } else if (ctx->architecture == dml2_architecture_21) { /* After DML2.1 update, ODM interpretation needs to change and is no longer same as for DML2.0. * This is not an issue with new resource management logic. This block ensure backcompat * with legacy pipe management with updated DML. * */ - if (ctx->architecture == dml2_architecture_21) { if (ODMMode[stream_disp_cfg_index] == 1) { scratch.odm_info.odm_factor = 1; } else if (ODMMode[stream_disp_cfg_index] == 2) { diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.c index 6b3b8803e0aee..a56e75cdf7123 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.c @@ -868,7 +868,7 @@ bool dml2_svp_remove_all_phantom_pipes(struct dml2_context *ctx, struct dc_state /* Conditions for setting up phantom pipes for SubVP: * 1. Not force disable SubVP - * 2. Full update (i.e. !fast_validate) + * 2. Full update (i.e. DC_VALIDATE_MODE_AND_PROGRAMMING) * 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) diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c index ab6baf2698012..3b866e876bf4d 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c @@ -896,7 +896,7 @@ static void populate_dummy_dml_surface_cfg(struct dml_surface_cfg_st *out, unsig out->SurfaceWidthC[location] = in->timing.h_addressable; out->SurfaceHeightC[location] = in->timing.v_addressable; out->PitchY[location] = ((out->SurfaceWidthY[location] + 127) / 128) * 128; - out->PitchC[location] = 0; + out->PitchC[location] = 1; out->DCCEnable[location] = false; out->DCCMetaPitchY[location] = 0; out->DCCMetaPitchC[location] = 0; @@ -953,6 +953,7 @@ static void populate_dml_surface_cfg_from_plane_state(enum dml_project_id dml2_p out->SourcePixelFormat[location] = dml_420_10; break; case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616: + case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616: case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F: case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F: out->SourcePixelFormat[location] = dml_444_64; @@ -1188,22 +1189,6 @@ static unsigned int map_plane_to_dml_display_cfg(const struct dml2_context *dml2 return location; } -static void apply_legacy_svp_drr_settings(struct dml2_context *dml2, const struct dc_state *state, struct dml_display_cfg_st *dml_dispcfg) -{ - int i; - - if (state->bw_ctx.bw.dcn.clk.fw_based_mclk_switching) { - ASSERT(state->stream_count == 1); - dml_dispcfg->timing.DRRDisplay[0] = true; - } else if (state->bw_ctx.bw.dcn.legacy_svp_drr_stream_index_valid) { - - for (i = 0; i < dml_dispcfg->num_timings; i++) { - if (dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[i] == state->streams[state->bw_ctx.bw.dcn.legacy_svp_drr_stream_index]->stream_id) - dml_dispcfg->timing.DRRDisplay[i] = true; - } - } -} - static void dml2_populate_pipe_to_plane_index_mapping(struct dml2_context *dml2, struct dc_state *state) { unsigned int i; @@ -1436,9 +1421,6 @@ void map_dc_state_into_dml_display_cfg(struct dml2_context *dml2, struct dc_stat } } } - - if (!dml2->config.use_native_pstate_optimization) - apply_legacy_svp_drr_settings(dml2, context, dml_dispcfg); } void dml2_update_pipe_ctx_dchub_regs(struct _vcs_dpi_dml_display_rq_regs_st *rq_regs, diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c index e89571874185e..0318260370edf 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c @@ -24,8 +24,6 @@ * */ -#include <linux/vmalloc.h> - #include "display_mode_core.h" #include "dml2_internal_types.h" #include "dml2_utils.h" @@ -95,12 +93,17 @@ static void map_hw_resources(struct dml2_context *dml2, static unsigned int pack_and_call_dml_mode_support_ex(struct dml2_context *dml2, const struct dml_display_cfg_st *display_cfg, - struct dml_mode_support_info_st *evaluation_info) + struct dml_mode_support_info_st *evaluation_info, + enum dc_validate_mode validate_mode) { struct dml2_wrapper_scratch *s = &dml2->v20.scratch; s->mode_support_params.mode_lib = &dml2->v20.dml_core_ctx; s->mode_support_params.in_display_cfg = display_cfg; + if (validate_mode == DC_VALIDATE_MODE_ONLY) + s->mode_support_params.in_start_state_idx = dml2->v20.dml_core_ctx.states.num_states - 1; + else + s->mode_support_params.in_start_state_idx = 0; s->mode_support_params.out_evaluation_info = evaluation_info; memset(evaluation_info, 0, sizeof(struct dml_mode_support_info_st)); @@ -112,10 +115,8 @@ static unsigned int pack_and_call_dml_mode_support_ex(struct dml2_context *dml2, static bool optimize_configuration(struct dml2_context *dml2, struct dml2_wrapper_optimize_configuration_params *p) { int unused_dpps = p->ip_params->max_num_dpp; - int i, j; - int odms_needed, refresh_rate_hz, dpps_needed, subvp_height, pstate_width_fw_delay_lines, surface_count; - int subvp_timing_to_add, new_timing_index, subvp_surface_to_add, new_surface_index; - float frame_time_sec, max_frame_time_sec; + int i; + int odms_needed; int largest_blend_and_timing = 0; bool optimization_done = false; @@ -130,79 +131,6 @@ static bool optimize_configuration(struct dml2_context *dml2, struct dml2_wrappe if (p->new_display_config != p->cur_display_config) *p->new_display_config = *p->cur_display_config; - // Optimize P-State Support - if (dml2->config.use_native_pstate_optimization) { - if (p->cur_mode_support_info->DRAMClockChangeSupport[0] == dml_dram_clock_change_unsupported) { - // Find a display with < 120Hz refresh rate with maximal refresh rate that's not already subvp - subvp_timing_to_add = -1; - subvp_surface_to_add = -1; - max_frame_time_sec = 0; - surface_count = 0; - for (i = 0; i < (int) p->cur_display_config->num_timings; i++) { - refresh_rate_hz = (int)div_u64((unsigned long long) p->cur_display_config->timing.PixelClock[i] * 1000 * 1000, - (p->cur_display_config->timing.HTotal[i] * p->cur_display_config->timing.VTotal[i])); - if (refresh_rate_hz < 120) { - // Check its upstream surfaces to see if this one could be converted to subvp. - dpps_needed = 0; - for (j = 0; j < (int) p->cur_display_config->num_surfaces; j++) { - if (p->cur_display_config->plane.BlendingAndTiming[j] == i && - p->cur_display_config->plane.UseMALLForPStateChange[j] == dml_use_mall_pstate_change_disable) { - dpps_needed += p->cur_mode_support_info->DPPPerSurface[j]; - subvp_surface_to_add = j; - surface_count++; - } - } - - if (surface_count == 1 && dpps_needed > 0 && dpps_needed <= unused_dpps) { - frame_time_sec = (float)1 / refresh_rate_hz; - if (frame_time_sec > max_frame_time_sec) { - max_frame_time_sec = frame_time_sec; - subvp_timing_to_add = i; - } - } - } - } - if (subvp_timing_to_add >= 0) { - new_timing_index = p->new_display_config->num_timings++; - new_surface_index = p->new_display_config->num_surfaces++; - // Add a phantom pipe reflecting the main pipe's timing - dml2_util_copy_dml_timing(&p->new_display_config->timing, new_timing_index, subvp_timing_to_add); - - pstate_width_fw_delay_lines = (int)(((double)(p->config->svp_pstate.subvp_fw_processing_delay_us + - p->config->svp_pstate.subvp_pstate_allow_width_us) / 1000000) * - (p->new_display_config->timing.PixelClock[subvp_timing_to_add] * 1000 * 1000) / - (double)p->new_display_config->timing.HTotal[subvp_timing_to_add]); - - subvp_height = p->cur_mode_support_info->SubViewportLinesNeededInMALL[subvp_timing_to_add] + pstate_width_fw_delay_lines; - - p->new_display_config->timing.VActive[new_timing_index] = subvp_height; - p->new_display_config->timing.VTotal[new_timing_index] = subvp_height + - p->new_display_config->timing.VTotal[subvp_timing_to_add] - p->new_display_config->timing.VActive[subvp_timing_to_add]; - - p->new_display_config->output.OutputDisabled[new_timing_index] = true; - - p->new_display_config->plane.UseMALLForPStateChange[subvp_surface_to_add] = dml_use_mall_pstate_change_sub_viewport; - - dml2_util_copy_dml_plane(&p->new_display_config->plane, new_surface_index, subvp_surface_to_add); - dml2_util_copy_dml_surface(&p->new_display_config->surface, new_surface_index, subvp_surface_to_add); - - p->new_display_config->plane.ViewportHeight[new_surface_index] = subvp_height; - p->new_display_config->plane.ViewportHeightChroma[new_surface_index] = subvp_height; - p->new_display_config->plane.ViewportStationary[new_surface_index] = false; - - p->new_display_config->plane.UseMALLForStaticScreen[new_surface_index] = dml_use_mall_static_screen_disable; - p->new_display_config->plane.UseMALLForPStateChange[new_surface_index] = dml_use_mall_pstate_change_phantom_pipe; - - p->new_display_config->plane.NumberOfCursors[new_surface_index] = 0; - - p->new_policy->ImmediateFlipRequirement[new_surface_index] = dml_immediate_flip_not_required; - - p->new_display_config->plane.BlendingAndTiming[new_surface_index] = new_timing_index; - - optimization_done = true; - } - } - } // Optimize Clocks if (!optimization_done) { @@ -226,7 +154,8 @@ static bool optimize_configuration(struct dml2_context *dml2, struct dml2_wrappe return optimization_done; } -static int calculate_lowest_supported_state_for_temp_read(struct dml2_context *dml2, struct dc_state *display_state) +static int calculate_lowest_supported_state_for_temp_read(struct dml2_context *dml2, struct dc_state *display_state, + enum dc_validate_mode validate_mode) { struct dml2_calculate_lowest_supported_state_for_temp_read_scratch *s = &dml2->v20.scratch.dml2_calculate_lowest_supported_state_for_temp_read_scratch; struct dml2_wrapper_scratch *s_global = &dml2->v20.scratch; @@ -268,7 +197,8 @@ static int calculate_lowest_supported_state_for_temp_read(struct dml2_context *d dml2->v20.dml_core_ctx.states.state_array[j].dram_clock_change_latency_us = s_global->dummy_pstate_table[i].dummy_pstate_latency_us; } - dml_result = pack_and_call_dml_mode_support_ex(dml2, &s->cur_display_config, &s->evaluation_info); + dml_result = pack_and_call_dml_mode_support_ex(dml2, &s->cur_display_config, &s->evaluation_info, + validate_mode); if (dml_result && s->evaluation_info.DRAMClockChangeSupport[0] == dml_dram_clock_change_vactive) { map_hw_resources(dml2, &s->cur_display_config, &s->evaluation_info); @@ -333,7 +263,8 @@ static bool does_configuration_meet_sw_policies(struct dml2_context *ctx, const } static bool dml_mode_support_wrapper(struct dml2_context *dml2, - struct dc_state *display_state) + struct dc_state *display_state, + enum dc_validate_mode validate_mode) { struct dml2_wrapper_scratch *s = &dml2->v20.scratch; unsigned int result = 0, i; @@ -369,7 +300,8 @@ static bool dml_mode_support_wrapper(struct dml2_context *dml2, result = pack_and_call_dml_mode_support_ex(dml2, &s->cur_display_config, - &s->mode_support_info); + &s->mode_support_info, + validate_mode); if (result) result = does_configuration_meet_sw_policies(dml2, &s->cur_display_config, &s->mode_support_info); @@ -390,7 +322,8 @@ static bool dml_mode_support_wrapper(struct dml2_context *dml2, dml2->v20.dml_core_ctx.policy = s->new_policy; optimized_result = pack_and_call_dml_mode_support_ex(dml2, &s->new_display_config, - &s->mode_support_info); + &s->mode_support_info, + validate_mode); if (optimized_result) optimized_result = does_configuration_meet_sw_policies(dml2, &s->new_display_config, &s->mode_support_info); @@ -409,7 +342,8 @@ static bool dml_mode_support_wrapper(struct dml2_context *dml2, if (!optimized_result) { result = pack_and_call_dml_mode_support_ex(dml2, &s->cur_display_config, - &s->mode_support_info); + &s->mode_support_info, + validate_mode); } } @@ -419,118 +353,7 @@ static bool dml_mode_support_wrapper(struct dml2_context *dml2, return result; } -static int find_drr_eligible_stream(struct dc_state *display_state) -{ - int i; - - for (i = 0; i < display_state->stream_count; i++) { - if (dc_state_get_stream_subvp_type(display_state, display_state->streams[i]) == SUBVP_NONE - && display_state->streams[i]->ignore_msa_timing_param) { - // Use ignore_msa_timing_param flag to identify as DRR - return i; - } - } - - return -1; -} - -static bool optimize_pstate_with_svp_and_drr(struct dml2_context *dml2, struct dc_state *display_state) -{ - struct dml2_wrapper_scratch *s = &dml2->v20.scratch; - bool pstate_optimization_done = false; - bool pstate_optimization_success = false; - bool result = false; - int drr_display_index = 0, non_svp_streams = 0; - bool force_svp = dml2->config.svp_pstate.force_enable_subvp; - - display_state->bw_ctx.bw.dcn.clk.fw_based_mclk_switching = false; - display_state->bw_ctx.bw.dcn.legacy_svp_drr_stream_index_valid = false; - - result = dml_mode_support_wrapper(dml2, display_state); - - if (!result) { - pstate_optimization_done = true; - } else if (s->mode_support_info.DRAMClockChangeSupport[0] != dml_dram_clock_change_unsupported && !force_svp) { - pstate_optimization_success = true; - pstate_optimization_done = true; - } - - if (display_state->stream_count == 1 && dml2->config.callbacks.can_support_mclk_switch_using_fw_based_vblank_stretch(dml2->config.callbacks.dc, display_state)) { - display_state->bw_ctx.bw.dcn.clk.fw_based_mclk_switching = true; - - result = dml_mode_support_wrapper(dml2, display_state); - } else { - non_svp_streams = display_state->stream_count; - - while (!pstate_optimization_done) { - result = dml_mode_programming(&dml2->v20.dml_core_ctx, s->mode_support_params.out_lowest_state_idx, &s->cur_display_config, true); - - // Always try adding SVP first - if (result) - result = dml2_svp_add_phantom_pipe_to_dc_state(dml2, display_state, &s->mode_support_info); - else - pstate_optimization_done = true; - - - if (result) { - result = dml_mode_support_wrapper(dml2, display_state); - } else { - pstate_optimization_done = true; - } - - if (result) { - non_svp_streams--; - - if (s->mode_support_info.DRAMClockChangeSupport[0] != dml_dram_clock_change_unsupported) { - if (dml2_svp_validate_static_schedulability(dml2, display_state, s->mode_support_info.DRAMClockChangeSupport[0])) { - pstate_optimization_success = true; - pstate_optimization_done = true; - } else { - pstate_optimization_success = false; - pstate_optimization_done = false; - } - } else { - drr_display_index = find_drr_eligible_stream(display_state); - - // If there is only 1 remaining non SubVP pipe that is DRR, check static - // schedulability for SubVP + DRR. - if (non_svp_streams == 1 && drr_display_index >= 0) { - if (dml2_svp_drr_schedulable(dml2, display_state, &display_state->streams[drr_display_index]->timing)) { - display_state->bw_ctx.bw.dcn.legacy_svp_drr_stream_index_valid = true; - display_state->bw_ctx.bw.dcn.legacy_svp_drr_stream_index = drr_display_index; - result = dml_mode_support_wrapper(dml2, display_state); - } - - if (result && s->mode_support_info.DRAMClockChangeSupport[0] != dml_dram_clock_change_unsupported) { - pstate_optimization_success = true; - pstate_optimization_done = true; - } else { - pstate_optimization_success = false; - pstate_optimization_done = false; - } - } - - if (pstate_optimization_success) { - pstate_optimization_done = true; - } else { - pstate_optimization_done = false; - } - } - } - } - } - - if (!pstate_optimization_success) { - dml2_svp_remove_all_phantom_pipes(dml2, display_state); - display_state->bw_ctx.bw.dcn.clk.fw_based_mclk_switching = false; - display_state->bw_ctx.bw.dcn.legacy_svp_drr_stream_index_valid = false; - result = dml_mode_support_wrapper(dml2, display_state); - } - - return result; -} - -static bool call_dml_mode_support_and_programming(struct dc_state *context) +static bool call_dml_mode_support_and_programming(struct dc_state *context, enum dc_validate_mode validate_mode) { unsigned int result = 0; unsigned int min_state = 0; @@ -544,16 +367,13 @@ static bool call_dml_mode_support_and_programming(struct dc_state *context) struct dml2_wrapper_scratch *s = &dml2->v20.scratch; if (!context->streams[0]->sink->link->dc->caps.is_apu) { - min_state_for_g6_temp_read = calculate_lowest_supported_state_for_temp_read(dml2, context); + min_state_for_g6_temp_read = calculate_lowest_supported_state_for_temp_read(dml2, context, + validate_mode); ASSERT(min_state_for_g6_temp_read >= 0); } - if (!dml2->config.use_native_pstate_optimization) { - result = optimize_pstate_with_svp_and_drr(dml2, context); - } else { - result = dml_mode_support_wrapper(dml2, context); - } + result = dml_mode_support_wrapper(dml2, context, validate_mode); /* Upon trying to sett certain frequencies in FRL, min_state_for_g6_temp_read is reported as -1. This leads to an invalid value of min_state causing crashes later on. * Use the default logic for min_state only when min_state_for_g6_temp_read is a valid value. In other cases, use the value calculated by the DML directly. @@ -575,7 +395,8 @@ static bool call_dml_mode_support_and_programming(struct dc_state *context) return result; } -static bool dml2_validate_and_build_resource(const struct dc *in_dc, struct dc_state *context) +static bool dml2_validate_and_build_resource(const struct dc *in_dc, struct dc_state *context, + enum dc_validate_mode validate_mode) { struct dml2_context *dml2 = context->bw_ctx.dml2; struct dml2_wrapper_scratch *s = &dml2->v20.scratch; @@ -611,7 +432,7 @@ static bool dml2_validate_and_build_resource(const struct dc *in_dc, struct dc_s copy_dummy_pstate_table(s->dummy_pstate_table, in_dc->clk_mgr->bw_params->dummy_pstate_table, 4); - result = call_dml_mode_support_and_programming(context); + result = call_dml_mode_support_and_programming(context, validate_mode); /* Call map dc pipes to map the pipes based on the DML output. For correctly determining if recalculation * is required or not, the resource context needs to correctly reflect the number of active pipes. We would * only know the correct number if active pipes after dml2_map_dc_pipes is called. @@ -628,7 +449,7 @@ static bool dml2_validate_and_build_resource(const struct dc *in_dc, struct dc_s need_recalculation = dml2_verify_det_buffer_configuration(dml2, context, &dml2->det_helper_scratch); if (need_recalculation) { /* Engage the DML again if recalculation is required. */ - call_dml_mode_support_and_programming(context); + call_dml_mode_support_and_programming(context, validate_mode); if (!dml2->config.skip_hw_state_mapping) { dml2_map_dc_pipes(dml2, context, &s->cur_display_config, &s->dml_to_dc_pipe_mapping, in_dc->current_state); } @@ -663,7 +484,10 @@ static bool dml2_validate_and_build_resource(const struct dc *in_dc, struct dc_s dml2_copy_clocks_to_dc_state(&out_clks, context); dml2_extract_watermark_set(&context->bw_ctx.bw.dcn.watermarks.a, &dml2->v20.dml_core_ctx); dml2_extract_watermark_set(&context->bw_ctx.bw.dcn.watermarks.b, &dml2->v20.dml_core_ctx); - memcpy(&context->bw_ctx.bw.dcn.watermarks.c, &dml2->v20.g6_temp_read_watermark_set, sizeof(context->bw_ctx.bw.dcn.watermarks.c)); + if (context->streams[0]->sink->link->dc->caps.is_apu) + dml2_extract_watermark_set(&context->bw_ctx.bw.dcn.watermarks.c, &dml2->v20.dml_core_ctx); + else + memcpy(&context->bw_ctx.bw.dcn.watermarks.c, &dml2->v20.g6_temp_read_watermark_set, sizeof(context->bw_ctx.bw.dcn.watermarks.c)); dml2_extract_watermark_set(&context->bw_ctx.bw.dcn.watermarks.d, &dml2->v20.dml_core_ctx); dml2_extract_writeback_wm(context, &dml2->v20.dml_core_ctx); //copy for deciding zstate use @@ -681,7 +505,7 @@ static bool dml2_validate_and_build_resource(const struct dc *in_dc, struct dc_s return result; } -static bool dml2_validate_only(struct dc_state *context) +static bool dml2_validate_only(struct dc_state *context, enum dc_validate_mode validate_mode) { struct dml2_context *dml2; unsigned int result = 0; @@ -705,7 +529,8 @@ static bool dml2_validate_only(struct dc_state *context) result = pack_and_call_dml_mode_support_ex(dml2, &dml2->v20.scratch.cur_display_config, - &dml2->v20.scratch.mode_support_info); + &dml2->v20.scratch.mode_support_info, + validate_mode); if (result) result = does_configuration_meet_sw_policies(dml2, &dml2->v20.scratch.cur_display_config, &dml2->v20.scratch.mode_support_info); @@ -720,7 +545,8 @@ static void dml2_apply_debug_options(const struct dc *dc, struct dml2_context *d } } -bool dml2_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml2, bool fast_validate) +bool dml2_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml2, + enum dc_validate_mode validate_mode) { bool out = false; @@ -730,17 +556,17 @@ bool dml2_validate(const struct dc *in_dc, struct dc_state *context, struct dml2 /* DML2.1 validation path */ if (dml2->architecture == dml2_architecture_21) { - out = dml21_validate(in_dc, context, dml2, fast_validate); + out = dml21_validate(in_dc, context, dml2, validate_mode); return out; } DC_FP_START(); - /* Use dml_validate_only for fast_validate path */ - if (fast_validate) - out = dml2_validate_only(context); + /* Use dml_validate_only for DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX path */ + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) + out = dml2_validate_only(context, validate_mode); else - out = dml2_validate_and_build_resource(in_dc, context); + out = dml2_validate_and_build_resource(in_dc, context, validate_mode); DC_FP_END(); @@ -754,8 +580,8 @@ static inline struct dml2_context *dml2_allocate_memory(void) static void dml2_init(const struct dc *in_dc, const struct dml2_configuration_options *config, struct dml2_context **dml2) { - if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version == DCN_VERSION_4_01)) { - dml21_reinit(in_dc, dml2, config); + if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version >= DCN_VERSION_4_01)) { + dml21_reinit(in_dc, *dml2, config); return; } @@ -800,9 +626,7 @@ static void dml2_init(const struct dc *in_dc, const struct dml2_configuration_op bool dml2_create(const struct dc *in_dc, const struct dml2_configuration_options *config, struct dml2_context **dml2) { // TODO : Temporarily add DCN_VERSION_3_2 for N-1 validation. Remove DCN_VERSION_3_2 after N-1 validation phase is complete. - if ((in_dc->debug.using_dml21) - && (in_dc->ctx->dce_version == DCN_VERSION_4_01 - )) + if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version >= DCN_VERSION_4_01)) return dml21_create(in_dc, dml2, config); // Allocate Mode Lib Ctx @@ -871,8 +695,8 @@ void dml2_reinit(const struct dc *in_dc, const struct dml2_configuration_options *config, struct dml2_context **dml2) { - if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version == DCN_VERSION_4_01)) { - dml21_reinit(in_dc, dml2, config); + if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version >= DCN_VERSION_4_01)) { + dml21_reinit(in_dc, *dml2, config); return; } diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h index 785226945699d..c384e141cebc8 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h @@ -40,6 +40,7 @@ struct dc_sink; struct dc_stream_state; struct resource_context; struct display_stream_compressor; +struct dc_mcache_params; // Configuration of the MALL on the SoC struct dml2_soc_mall_info { @@ -107,6 +108,7 @@ struct dml2_dc_callbacks { unsigned int (*get_max_flickerless_instant_vtotal_increase)( struct dc_stream_state *stream, bool is_gaming); + bool (*allocate_mcache)(struct dc_state *context, const struct dc_mcache_params *mcache_params); }; struct dml2_dc_svp_callbacks { @@ -238,7 +240,7 @@ struct dml2_configuration_options { bool use_clock_dc_limits; bool gpuvm_enable; bool force_tdlut_enable; - struct dml2_soc_bb *bb_from_dmub; + void *bb_from_dmub; }; /* @@ -270,7 +272,7 @@ void dml2_reinit(const struct dc *in_dc, * dml2_validate - Determines if a display configuration is supported or not. * @in_dc: dc. * @context: dc_state to be validated. - * @fast_validate: Fast validate will not populate context.res_ctx. + * @validate_mode: DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX will not populate context.res_ctx. * * DML1.0 compatible interface for validation. * @@ -293,7 +295,7 @@ void dml2_reinit(const struct dc *in_dc, bool dml2_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml2, - bool fast_validate); + enum dc_validate_mode validate_mode); /* * dml2_extract_dram_and_fclk_change_support - Extracts the FCLK and UCLK change support info. |