diff options
Diffstat (limited to 'drivers/gpu/drm/amd/powerplay')
33 files changed, 1320 insertions, 402 deletions
diff --git a/drivers/gpu/drm/amd/powerplay/amd_powerplay.c b/drivers/gpu/drm/amd/powerplay/amd_powerplay.c index 7932eb163a00..c195575366a3 100644 --- a/drivers/gpu/drm/amd/powerplay/amd_powerplay.c +++ b/drivers/gpu/drm/amd/powerplay/amd_powerplay.c @@ -48,7 +48,6 @@ static int amd_powerplay_create(struct amdgpu_device *adev) hwmgr->adev = adev; hwmgr->not_vf = !amdgpu_sriov_vf(adev); - hwmgr->pm_en = (amdgpu_dpm && hwmgr->not_vf) ? true : false; hwmgr->device = amdgpu_cgs_create_device(adev); mutex_init(&hwmgr->smu_lock); hwmgr->chip_family = adev->family; @@ -928,9 +927,12 @@ static int pp_dpm_set_mp1_state(void *handle, enum pp_mp1_state mp1_state) { struct pp_hwmgr *hwmgr = handle; - if (!hwmgr || !hwmgr->pm_en) + if (!hwmgr) return -EINVAL; + if (!hwmgr->pm_en) + return 0; + if (hwmgr->hwmgr_func->set_mp1_state) return hwmgr->hwmgr_func->set_mp1_state(hwmgr, mp1_state); diff --git a/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c b/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c index a23729d3174b..99469479e277 100644 --- a/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c +++ b/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c @@ -356,6 +356,35 @@ int smu_get_dpm_level_count(struct smu_context *smu, enum smu_clk_type clk_type, return smu_get_dpm_freq_by_index(smu, clk_type, 0xff, value); } +int smu_get_dpm_level_range(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t *min_value, uint32_t *max_value) +{ + int ret = 0; + uint32_t level_count = 0; + + if (!min_value && !max_value) + return -EINVAL; + + if (min_value) { + /* by default, level 0 clock value as min value */ + ret = smu_get_dpm_freq_by_index(smu, clk_type, 0, min_value); + if (ret) + return ret; + } + + if (max_value) { + ret = smu_get_dpm_level_count(smu, clk_type, &level_count); + if (ret) + return ret; + + ret = smu_get_dpm_freq_by_index(smu, clk_type, level_count - 1, max_value); + if (ret) + return ret; + } + + return ret; +} + bool smu_clk_dpm_is_enabled(struct smu_context *smu, enum smu_clk_type clk_type) { enum smu_feature_mask feature_id = 0; @@ -404,10 +433,10 @@ int smu_dpm_set_power_gate(struct smu_context *smu, uint32_t block_type, switch (block_type) { case AMD_IP_BLOCK_TYPE_UVD: - ret = smu_dpm_set_uvd_enable(smu, gate); + ret = smu_dpm_set_uvd_enable(smu, !gate); break; case AMD_IP_BLOCK_TYPE_VCE: - ret = smu_dpm_set_vce_enable(smu, gate); + ret = smu_dpm_set_vce_enable(smu, !gate); break; case AMD_IP_BLOCK_TYPE_GFX: ret = smu_gfx_off_control(smu, gate); @@ -415,6 +444,9 @@ int smu_dpm_set_power_gate(struct smu_context *smu, uint32_t block_type, case AMD_IP_BLOCK_TYPE_SDMA: ret = smu_powergate_sdma(smu, gate); break; + case AMD_IP_BLOCK_TYPE_JPEG: + ret = smu_dpm_set_jpeg_enable(smu, !gate); + break; default: break; } @@ -487,26 +519,25 @@ int smu_update_table(struct smu_context *smu, enum smu_table_id table_index, int { struct smu_table_context *smu_table = &smu->smu_table; struct amdgpu_device *adev = smu->adev; - struct smu_table *table = NULL; - int ret = 0; + struct smu_table *table = &smu_table->driver_table; int table_id = smu_table_get_index(smu, table_index); + uint32_t table_size; + int ret = 0; if (!table_data || table_id >= SMU_TABLE_COUNT || table_id < 0) return -EINVAL; - table = &smu_table->tables[table_index]; + table_size = smu_table->tables[table_index].size; - if (drv2smu) - memcpy(table->cpu_addr, table_data, table->size); + if (drv2smu) { + memcpy(table->cpu_addr, table_data, table_size); + /* + * Flush hdp cache: to guard the content seen by + * GPU is consitent with CPU. + */ + amdgpu_asic_flush_hdp(adev, NULL); + } - ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetDriverDramAddrHigh, - upper_32_bits(table->mc_address)); - if (ret) - return ret; - ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetDriverDramAddrLow, - lower_32_bits(table->mc_address)); - if (ret) - return ret; ret = smu_send_smc_msg_with_param(smu, drv2smu ? SMU_MSG_TransferTableDram2Smu : SMU_MSG_TransferTableSmu2Dram, @@ -514,11 +545,10 @@ int smu_update_table(struct smu_context *smu, enum smu_table_id table_index, int if (ret) return ret; - /* flush hdp cache */ - adev->nbio.funcs->hdp_flush(adev, NULL); - - if (!drv2smu) - memcpy(table_data, table->cpu_addr, table->size); + if (!drv2smu) { + amdgpu_asic_flush_hdp(adev, NULL); + memcpy(table_data, table->cpu_addr, table_size); + } return ret; } @@ -527,9 +557,12 @@ bool is_support_sw_smu(struct amdgpu_device *adev) { if (adev->asic_type == CHIP_VEGA20) return (amdgpu_dpm == 2) ? true : false; - else if (adev->asic_type >= CHIP_ARCTURUS) - return true; - else + else if (adev->asic_type >= CHIP_ARCTURUS) { + if (amdgpu_sriov_vf(adev)&& !amdgpu_sriov_is_pp_one_vf(adev)) + return false; + else + return true; + } else return false; } @@ -637,12 +670,11 @@ int smu_feature_init_dpm(struct smu_context *smu) int smu_feature_is_enabled(struct smu_context *smu, enum smu_feature_mask mask) { - struct amdgpu_device *adev = smu->adev; struct smu_feature *feature = &smu->smu_feature; int feature_id; int ret = 0; - if (adev->flags & AMD_IS_APU) + if (smu->is_apu) return 1; feature_id = smu_feature_get_index(smu, mask); @@ -942,32 +974,56 @@ static int smu_init_fb_allocations(struct smu_context *smu) struct amdgpu_device *adev = smu->adev; struct smu_table_context *smu_table = &smu->smu_table; struct smu_table *tables = smu_table->tables; + struct smu_table *driver_table = &(smu_table->driver_table); + uint32_t max_table_size = 0; int ret, i; - for (i = 0; i < SMU_TABLE_COUNT; i++) { - if (tables[i].size == 0) - continue; + /* VRAM allocation for tool table */ + if (tables[SMU_TABLE_PMSTATUSLOG].size) { ret = amdgpu_bo_create_kernel(adev, - tables[i].size, - tables[i].align, - tables[i].domain, - &tables[i].bo, - &tables[i].mc_address, - &tables[i].cpu_addr); - if (ret) - goto failed; + tables[SMU_TABLE_PMSTATUSLOG].size, + tables[SMU_TABLE_PMSTATUSLOG].align, + tables[SMU_TABLE_PMSTATUSLOG].domain, + &tables[SMU_TABLE_PMSTATUSLOG].bo, + &tables[SMU_TABLE_PMSTATUSLOG].mc_address, + &tables[SMU_TABLE_PMSTATUSLOG].cpu_addr); + if (ret) { + pr_err("VRAM allocation for tool table failed!\n"); + return ret; + } } - return 0; -failed: - while (--i >= 0) { + /* VRAM allocation for driver table */ + for (i = 0; i < SMU_TABLE_COUNT; i++) { if (tables[i].size == 0) continue; - amdgpu_bo_free_kernel(&tables[i].bo, - &tables[i].mc_address, - &tables[i].cpu_addr); + if (i == SMU_TABLE_PMSTATUSLOG) + continue; + + if (max_table_size < tables[i].size) + max_table_size = tables[i].size; + } + + driver_table->size = max_table_size; + driver_table->align = PAGE_SIZE; + driver_table->domain = AMDGPU_GEM_DOMAIN_VRAM; + + ret = amdgpu_bo_create_kernel(adev, + driver_table->size, + driver_table->align, + driver_table->domain, + &driver_table->bo, + &driver_table->mc_address, + &driver_table->cpu_addr); + if (ret) { + pr_err("VRAM allocation for driver table failed!\n"); + if (tables[SMU_TABLE_PMSTATUSLOG].mc_address) + amdgpu_bo_free_kernel(&tables[SMU_TABLE_PMSTATUSLOG].bo, + &tables[SMU_TABLE_PMSTATUSLOG].mc_address, + &tables[SMU_TABLE_PMSTATUSLOG].cpu_addr); } + return ret; } @@ -975,18 +1031,19 @@ static int smu_fini_fb_allocations(struct smu_context *smu) { struct smu_table_context *smu_table = &smu->smu_table; struct smu_table *tables = smu_table->tables; - uint32_t i = 0; + struct smu_table *driver_table = &(smu_table->driver_table); if (!tables) return 0; - for (i = 0; i < SMU_TABLE_COUNT; i++) { - if (tables[i].size == 0) - continue; - amdgpu_bo_free_kernel(&tables[i].bo, - &tables[i].mc_address, - &tables[i].cpu_addr); - } + if (tables[SMU_TABLE_PMSTATUSLOG].mc_address) + amdgpu_bo_free_kernel(&tables[SMU_TABLE_PMSTATUSLOG].bo, + &tables[SMU_TABLE_PMSTATUSLOG].mc_address, + &tables[SMU_TABLE_PMSTATUSLOG].cpu_addr); + + amdgpu_bo_free_kernel(&driver_table->bo, + &driver_table->mc_address, + &driver_table->cpu_addr); return 0; } @@ -1056,28 +1113,31 @@ static int smu_smc_table_hw_init(struct smu_context *smu, } /* smu_dump_pptable(smu); */ + if (!amdgpu_sriov_vf(adev)) { + ret = smu_set_driver_table_location(smu); + if (ret) + return ret; - /* - * Copy pptable bo in the vram to smc with SMU MSGs such as - * SetDriverDramAddr and TransferTableDram2Smu. - */ - ret = smu_write_pptable(smu); - if (ret) - return ret; - - /* issue Run*Btc msg */ - ret = smu_run_btc(smu); - if (ret) - return ret; - - ret = smu_feature_set_allowed_mask(smu); - if (ret) - return ret; + /* + * Copy pptable bo in the vram to smc with SMU MSGs such as + * SetDriverDramAddr and TransferTableDram2Smu. + */ + ret = smu_write_pptable(smu); + if (ret) + return ret; - ret = smu_system_features_control(smu, true); - if (ret) - return ret; + /* issue Run*Btc msg */ + ret = smu_run_btc(smu); + if (ret) + return ret; + ret = smu_feature_set_allowed_mask(smu); + if (ret) + return ret; + ret = smu_system_features_control(smu, true); + if (ret) + return ret; + } if (adev->asic_type != CHIP_ARCTURUS) { ret = smu_notify_display_change(smu); if (ret) @@ -1130,8 +1190,9 @@ static int smu_smc_table_hw_init(struct smu_context *smu, /* * Set PMSTATUSLOG table bo address with SetToolsDramAddr MSG for tools. */ - ret = smu_set_tool_table_location(smu); - + if (!amdgpu_sriov_vf(adev)) { + ret = smu_set_tool_table_location(smu); + } if (!smu_is_dpm_running(smu)) pr_info("dpm has been disabled\n"); @@ -1187,10 +1248,9 @@ static int smu_free_memory_pool(struct smu_context *smu) { struct smu_table_context *smu_table = &smu->smu_table; struct smu_table *memory_pool = &smu_table->memory_pool; - int ret = 0; if (memory_pool->size == SMU_MEMORY_POOL_SIZE_ZERO) - return ret; + return 0; amdgpu_bo_free_kernel(&memory_pool->bo, &memory_pool->mc_address, @@ -1198,7 +1258,7 @@ static int smu_free_memory_pool(struct smu_context *smu) memset(memory_pool, 0, sizeof(struct smu_table)); - return ret; + return 0; } static int smu_start_smc_engine(struct smu_context *smu) @@ -1237,12 +1297,16 @@ static int smu_hw_init(void *handle) return ret; } - if (adev->flags & AMD_IS_APU) { + if (smu->is_apu) { smu_powergate_sdma(&adev->smu, false); smu_powergate_vcn(&adev->smu, false); + smu_powergate_jpeg(&adev->smu, false); smu_set_gfx_cgpg(&adev->smu, true); } + if (amdgpu_sriov_vf(adev) && !amdgpu_sriov_is_pp_one_vf(adev)) + return 0; + if (!smu->pm_enabled) return 0; @@ -1285,7 +1349,7 @@ failed: static int smu_stop_dpms(struct smu_context *smu) { - return smu_send_smc_msg(smu, SMU_MSG_DisableAllSmuFeatures); + return smu_system_features_control(smu, false); } static int smu_hw_fini(void *handle) @@ -1295,36 +1359,45 @@ static int smu_hw_fini(void *handle) struct smu_table_context *table_context = &smu->smu_table; int ret = 0; - if (adev->flags & AMD_IS_APU) { + if (amdgpu_sriov_vf(adev)&& !amdgpu_sriov_is_pp_one_vf(adev)) + return 0; + + if (smu->is_apu) { smu_powergate_sdma(&adev->smu, true); smu_powergate_vcn(&adev->smu, true); + smu_powergate_jpeg(&adev->smu, true); } - ret = smu_stop_thermal_control(smu); - if (ret) { - pr_warn("Fail to stop thermal control!\n"); - return ret; - } + if (!smu->pm_enabled) + return 0; - /* - * For custom pptable uploading, skip the DPM features - * disable process on Navi1x ASICs. - * - As the gfx related features are under control of - * RLC on those ASICs. RLC reinitialization will be - * needed to reenable them. That will cost much more - * efforts. - * - * - SMU firmware can handle the DPM reenablement - * properly. - */ - if (!smu->uploading_custom_pp_table || - !((adev->asic_type >= CHIP_NAVI10) && - (adev->asic_type <= CHIP_NAVI12))) { - ret = smu_stop_dpms(smu); + if (!amdgpu_sriov_vf(adev)){ + ret = smu_stop_thermal_control(smu); if (ret) { - pr_warn("Fail to stop Dpms!\n"); + pr_warn("Fail to stop thermal control!\n"); return ret; } + + /* + * For custom pptable uploading, skip the DPM features + * disable process on Navi1x ASICs. + * - As the gfx related features are under control of + * RLC on those ASICs. RLC reinitialization will be + * needed to reenable them. That will cost much more + * efforts. + * + * - SMU firmware can handle the DPM reenablement + * properly. + */ + if (!smu->uploading_custom_pp_table || + !((adev->asic_type >= CHIP_NAVI10) && + (adev->asic_type <= CHIP_NAVI12))) { + ret = smu_stop_dpms(smu); + if (ret) { + pr_warn("Fail to stop Dpms!\n"); + return ret; + } + } } kfree(table_context->driver_pptable); @@ -1370,14 +1443,17 @@ static int smu_suspend(void *handle) struct smu_context *smu = &adev->smu; bool baco_feature_is_enabled = false; - if(!(adev->flags & AMD_IS_APU)) + if (!smu->pm_enabled) + return 0; + + if(!smu->is_apu) baco_feature_is_enabled = smu_feature_is_enabled(smu, SMU_FEATURE_BACO_BIT); ret = smu_system_features_control(smu, false); if (ret) return ret; - if (adev->in_gpu_reset && baco_feature_is_enabled) { + if (baco_feature_is_enabled) { ret = smu_feature_set_enabled(smu, SMU_FEATURE_BACO_BIT, true); if (ret) { pr_warn("set BACO feature enabled failed, return %d\n", ret); @@ -1402,6 +1478,12 @@ static int smu_resume(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; struct smu_context *smu = &adev->smu; + if (amdgpu_sriov_vf(adev)&& !amdgpu_sriov_is_pp_one_vf(adev)) + return 0; + + if (!smu->pm_enabled) + return 0; + pr_info("SMU is resuming...\n"); ret = smu_start_smc_engine(smu); @@ -1600,43 +1682,6 @@ static int smu_enable_umd_pstate(void *handle, return 0; } -static int smu_default_set_performance_level(struct smu_context *smu, enum amd_dpm_forced_level level) -{ - int ret = 0; - uint32_t sclk_mask, mclk_mask, soc_mask; - - switch (level) { - case AMD_DPM_FORCED_LEVEL_HIGH: - ret = smu_force_dpm_limit_value(smu, true); - break; - case AMD_DPM_FORCED_LEVEL_LOW: - ret = smu_force_dpm_limit_value(smu, false); - break; - case AMD_DPM_FORCED_LEVEL_AUTO: - case AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD: - ret = smu_unforce_dpm_levels(smu); - break; - case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK: - case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK: - case AMD_DPM_FORCED_LEVEL_PROFILE_PEAK: - ret = smu_get_profiling_clk_mask(smu, level, - &sclk_mask, - &mclk_mask, - &soc_mask); - if (ret) - return ret; - smu_force_clk_levels(smu, SMU_SCLK, 1 << sclk_mask, false); - smu_force_clk_levels(smu, SMU_MCLK, 1 << mclk_mask, false); - smu_force_clk_levels(smu, SMU_SOCCLK, 1 << soc_mask, false); - break; - case AMD_DPM_FORCED_LEVEL_MANUAL: - case AMD_DPM_FORCED_LEVEL_PROFILE_EXIT: - default: - break; - } - return ret; -} - int smu_adjust_power_state_dynamic(struct smu_context *smu, enum amd_dpm_forced_level level, bool skip_display_settings) @@ -1664,7 +1709,7 @@ int smu_adjust_power_state_dynamic(struct smu_context *smu, } if (!skip_display_settings) { - ret = smu_notify_smc_dispaly_config(smu); + ret = smu_notify_smc_display_config(smu); if (ret) { pr_err("Failed to notify smc display config!"); return ret; @@ -1674,11 +1719,8 @@ int smu_adjust_power_state_dynamic(struct smu_context *smu, if (smu_dpm_ctx->dpm_level != level) { ret = smu_asic_set_performance_level(smu, level); if (ret) { - ret = smu_default_set_performance_level(smu, level); - if (ret) { - pr_err("Failed to set performance level!"); - return ret; - } + pr_err("Failed to set performance level!"); + return ret; } /* update the saved copy */ @@ -1920,27 +1962,25 @@ int smu_set_df_cstate(struct smu_context *smu, int smu_write_watermarks_table(struct smu_context *smu) { - int ret = 0; - struct smu_table_context *smu_table = &smu->smu_table; - struct smu_table *table = NULL; - - table = &smu_table->tables[SMU_TABLE_WATERMARKS]; + void *watermarks_table = smu->smu_table.watermarks_table; - if (!table->cpu_addr) + if (!watermarks_table) return -EINVAL; - ret = smu_update_table(smu, SMU_TABLE_WATERMARKS, 0, table->cpu_addr, + return smu_update_table(smu, + SMU_TABLE_WATERMARKS, + 0, + watermarks_table, true); - - return ret; } int smu_set_watermarks_for_clock_ranges(struct smu_context *smu, struct dm_pp_wm_sets_with_clock_ranges_soc15 *clock_ranges) { - int ret = 0; - struct smu_table *watermarks = &smu->smu_table.tables[SMU_TABLE_WATERMARKS]; - void *table = watermarks->cpu_addr; + void *table = smu->smu_table.watermarks_table; + + if (!table) + return -EINVAL; mutex_lock(&smu->mutex); @@ -1954,7 +1994,7 @@ int smu_set_watermarks_for_clock_ranges(struct smu_context *smu, mutex_unlock(&smu->mutex); - return ret; + return 0; } const struct amd_ip_funcs smu_ip_funcs = { @@ -2279,13 +2319,9 @@ int smu_set_active_display_count(struct smu_context *smu, uint32_t count) { int ret = 0; - mutex_lock(&smu->mutex); - if (smu->ppt_funcs->set_active_display_count) ret = smu->ppt_funcs->set_active_display_count(smu, count); - mutex_unlock(&smu->mutex); - return ret; } @@ -2432,7 +2468,7 @@ bool smu_baco_is_support(struct smu_context *smu) mutex_lock(&smu->mutex); - if (smu->ppt_funcs->baco_is_support) + if (smu->ppt_funcs && smu->ppt_funcs->baco_is_support) ret = smu->ppt_funcs->baco_is_support(smu); mutex_unlock(&smu->mutex); @@ -2452,14 +2488,28 @@ int smu_baco_get_state(struct smu_context *smu, enum smu_baco_state *state) return 0; } -int smu_baco_reset(struct smu_context *smu) +int smu_baco_enter(struct smu_context *smu) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->baco_enter) + ret = smu->ppt_funcs->baco_enter(smu); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_baco_exit(struct smu_context *smu) { int ret = 0; mutex_lock(&smu->mutex); - if (smu->ppt_funcs->baco_reset) - ret = smu->ppt_funcs->baco_reset(smu); + if (smu->ppt_funcs->baco_exit) + ret = smu->ppt_funcs->baco_exit(smu); mutex_unlock(&smu->mutex); diff --git a/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c b/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c index 472e9fed411a..14ba6aa876e2 100644 --- a/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c +++ b/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c @@ -179,6 +179,7 @@ static struct smu_11_0_cmn2aisc_mapping arcturus_table_map[SMU_TABLE_COUNT] = { TAB_MAP(DRIVER_SMU_CONFIG), TAB_MAP(OVERDRIVE), TAB_MAP(I2C_COMMANDS), + TAB_MAP(ACTIVITY_MONITOR_COEFF), }; static struct smu_11_0_cmn2aisc_mapping arcturus_pwr_src_map[SMU_POWER_SOURCE_COUNT] = { @@ -280,10 +281,8 @@ static int arcturus_get_workload_type(struct smu_context *smu, enum PP_SMC_POWER return -EINVAL; mapping = arcturus_workload_map[profile]; - if (!(mapping.valid_mapping)) { - pr_warn("Unsupported SMU power source: %d\n", profile); + if (!(mapping.valid_mapping)) return -EINVAL; - } return mapping.map_to; } @@ -304,6 +303,10 @@ static int arcturus_tables_init(struct smu_context *smu, struct smu_table *table SMU_TABLE_INIT(tables, SMU_TABLE_I2C_COMMANDS, sizeof(SwI2cRequest_t), PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + SMU_TABLE_INIT(tables, SMU_TABLE_ACTIVITY_MONITOR_COEFF, + sizeof(DpmActivityMonitorCoeffInt_t), PAGE_SIZE, + AMDGPU_GEM_DOMAIN_VRAM); + smu_table->metrics_table = kzalloc(sizeof(SmuMetrics_t), GFP_KERNEL); if (!smu_table->metrics_table) return -ENOMEM; @@ -495,6 +498,7 @@ static int arcturus_store_powerplay_table(struct smu_context *smu) { struct smu_11_0_powerplay_table *powerplay_table = NULL; struct smu_table_context *table_context = &smu->smu_table; + struct smu_baco_context *smu_baco = &smu->smu_baco; int ret = 0; if (!table_context->power_play_table) @@ -507,6 +511,12 @@ static int arcturus_store_powerplay_table(struct smu_context *smu) table_context->thermal_controller_type = powerplay_table->thermal_controller_type; + mutex_lock(&smu_baco->mutex); + if (powerplay_table->platform_caps & SMU_11_0_PP_PLATFORM_CAP_BACO || + powerplay_table->platform_caps & SMU_11_0_PP_PLATFORM_CAP_MACO) + smu_baco->platform_support = true; + mutex_unlock(&smu_baco->mutex); + return ret; } @@ -1308,6 +1318,8 @@ static int arcturus_get_power_limit(struct smu_context *smu, static int arcturus_get_power_profile_mode(struct smu_context *smu, char *buf) { + struct amdgpu_device *adev = smu->adev; + DpmActivityMonitorCoeffInt_t activity_monitor; static const char *profile_name[] = { "BOOTUP_DEFAULT", "3D_FULL_SCREEN", @@ -1317,14 +1329,35 @@ static int arcturus_get_power_profile_mode(struct smu_context *smu, "COMPUTE", "CUSTOM"}; static const char *title[] = { - "PROFILE_INDEX(NAME)"}; + "PROFILE_INDEX(NAME)", + "CLOCK_TYPE(NAME)", + "FPS", + "UseRlcBusy", + "MinActiveFreqType", + "MinActiveFreq", + "BoosterFreqType", + "BoosterFreq", + "PD_Data_limit_c", + "PD_Data_error_coeff", + "PD_Data_error_rate_coeff"}; uint32_t i, size = 0; int16_t workload_type = 0; + int result = 0; + uint32_t smu_version; - if (!smu->pm_enabled || !buf) + if (!buf) return -EINVAL; - size += sprintf(buf + size, "%16s\n", + result = smu_get_smc_version(smu, NULL, &smu_version); + if (result) + return result; + + if (smu_version >= 0x360d00 && !amdgpu_sriov_vf(adev)) + size += sprintf(buf + size, "%16s %s %s %s %s %s %s %s %s %s %s\n", + title[0], title[1], title[2], title[3], title[4], title[5], + title[6], title[7], title[8], title[9], title[10]); + else + size += sprintf(buf + size, "%16s\n", title[0]); for (i = 0; i <= PP_SMC_POWER_PROFILE_CUSTOM; i++) { @@ -1336,8 +1369,50 @@ static int arcturus_get_power_profile_mode(struct smu_context *smu, if (workload_type < 0) continue; + if (smu_version >= 0x360d00 && !amdgpu_sriov_vf(adev)) { + result = smu_update_table(smu, + SMU_TABLE_ACTIVITY_MONITOR_COEFF, + workload_type, + (void *)(&activity_monitor), + false); + if (result) { + pr_err("[%s] Failed to get activity monitor!", __func__); + return result; + } + } + size += sprintf(buf + size, "%2d %14s%s\n", i, profile_name[i], (i == smu->power_profile_mode) ? "*" : " "); + + if (smu_version >= 0x360d00 && !amdgpu_sriov_vf(adev)) { + size += sprintf(buf + size, "%19s %d(%13s) %7d %7d %7d %7d %7d %7d %7d %7d %7d\n", + " ", + 0, + "GFXCLK", + activity_monitor.Gfx_FPS, + activity_monitor.Gfx_UseRlcBusy, + activity_monitor.Gfx_MinActiveFreqType, + activity_monitor.Gfx_MinActiveFreq, + activity_monitor.Gfx_BoosterFreqType, + activity_monitor.Gfx_BoosterFreq, + activity_monitor.Gfx_PD_Data_limit_c, + activity_monitor.Gfx_PD_Data_error_coeff, + activity_monitor.Gfx_PD_Data_error_rate_coeff); + + size += sprintf(buf + size, "%19s %d(%13s) %7d %7d %7d %7d %7d %7d %7d %7d %7d\n", + " ", + 1, + "UCLK", + activity_monitor.Mem_FPS, + activity_monitor.Mem_UseRlcBusy, + activity_monitor.Mem_MinActiveFreqType, + activity_monitor.Mem_MinActiveFreq, + activity_monitor.Mem_BoosterFreqType, + activity_monitor.Mem_BoosterFreq, + activity_monitor.Mem_PD_Data_limit_c, + activity_monitor.Mem_PD_Data_error_coeff, + activity_monitor.Mem_PD_Data_error_rate_coeff); + } } return size; @@ -1347,18 +1422,69 @@ static int arcturus_set_power_profile_mode(struct smu_context *smu, long *input, uint32_t size) { + DpmActivityMonitorCoeffInt_t activity_monitor; int workload_type = 0; uint32_t profile_mode = input[size]; int ret = 0; - - if (!smu->pm_enabled) - return -EINVAL; + uint32_t smu_version; if (profile_mode > PP_SMC_POWER_PROFILE_CUSTOM) { pr_err("Invalid power profile mode %d\n", profile_mode); return -EINVAL; } + ret = smu_get_smc_version(smu, NULL, &smu_version); + if (ret) + return ret; + + if ((profile_mode == PP_SMC_POWER_PROFILE_CUSTOM) && + (smu_version >=0x360d00)) { + ret = smu_update_table(smu, + SMU_TABLE_ACTIVITY_MONITOR_COEFF, + WORKLOAD_PPLIB_CUSTOM_BIT, + (void *)(&activity_monitor), + false); + if (ret) { + pr_err("[%s] Failed to get activity monitor!", __func__); + return ret; + } + + switch (input[0]) { + case 0: /* Gfxclk */ + activity_monitor.Gfx_FPS = input[1]; + activity_monitor.Gfx_UseRlcBusy = input[2]; + activity_monitor.Gfx_MinActiveFreqType = input[3]; + activity_monitor.Gfx_MinActiveFreq = input[4]; + activity_monitor.Gfx_BoosterFreqType = input[5]; + activity_monitor.Gfx_BoosterFreq = input[6]; + activity_monitor.Gfx_PD_Data_limit_c = input[7]; + activity_monitor.Gfx_PD_Data_error_coeff = input[8]; + activity_monitor.Gfx_PD_Data_error_rate_coeff = input[9]; + break; + case 1: /* Uclk */ + activity_monitor.Mem_FPS = input[1]; + activity_monitor.Mem_UseRlcBusy = input[2]; + activity_monitor.Mem_MinActiveFreqType = input[3]; + activity_monitor.Mem_MinActiveFreq = input[4]; + activity_monitor.Mem_BoosterFreqType = input[5]; + activity_monitor.Mem_BoosterFreq = input[6]; + activity_monitor.Mem_PD_Data_limit_c = input[7]; + activity_monitor.Mem_PD_Data_error_coeff = input[8]; + activity_monitor.Mem_PD_Data_error_rate_coeff = input[9]; + break; + } + + ret = smu_update_table(smu, + SMU_TABLE_ACTIVITY_MONITOR_COEFF, + WORKLOAD_PPLIB_CUSTOM_BIT, + (void *)(&activity_monitor), + true); + if (ret) { + pr_err("[%s] Failed to set activity monitor!", __func__); + return ret; + } + } + /* * Conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT * Not all profile modes are supported on arcturus. @@ -1897,7 +2023,7 @@ static int arcturus_i2c_eeprom_read_data(struct i2c_adapter *control, SwI2cRequest_t req; struct amdgpu_device *adev = to_amdgpu_device(control); struct smu_table_context *smu_table = &adev->smu.smu_table; - struct smu_table *table = &smu_table->tables[SMU_TABLE_I2C_COMMANDS]; + struct smu_table *table = &smu_table->driver_table; memset(&req, 0, sizeof(req)); arcturus_fill_eeprom_i2c_req(&req, false, address, numbytes, data); @@ -2051,8 +2177,12 @@ static const struct i2c_algorithm arcturus_i2c_eeprom_i2c_algo = { static int arcturus_i2c_eeprom_control_init(struct i2c_adapter *control) { struct amdgpu_device *adev = to_amdgpu_device(control); + struct smu_context *smu = &adev->smu; int res; + if (!smu->pm_enabled) + return -EOPNOTSUPP; + control->owner = THIS_MODULE; control->class = I2C_CLASS_SPD; control->dev.parent = &adev->pdev->dev; @@ -2068,6 +2198,12 @@ static int arcturus_i2c_eeprom_control_init(struct i2c_adapter *control) static void arcturus_i2c_eeprom_control_fini(struct i2c_adapter *control) { + struct amdgpu_device *adev = to_amdgpu_device(control); + struct smu_context *smu = &adev->smu; + + if (!smu->pm_enabled) + return; + i2c_del_adapter(control); } @@ -2112,6 +2248,7 @@ static const struct pptable_funcs arcturus_ppt_funcs = { .get_profiling_clk_mask = arcturus_get_profiling_clk_mask, .get_power_profile_mode = arcturus_get_power_profile_mode, .set_power_profile_mode = arcturus_set_power_profile_mode, + .set_performance_level = smu_v11_0_set_performance_level, /* debug (internal used) */ .dump_pptable = arcturus_dump_pptable, .get_power_limit = arcturus_get_power_limit, @@ -2135,6 +2272,7 @@ static const struct pptable_funcs arcturus_ppt_funcs = { .check_fw_version = smu_v11_0_check_fw_version, .write_pptable = smu_v11_0_write_pptable, .set_min_dcef_deep_sleep = smu_v11_0_set_min_dcef_deep_sleep, + .set_driver_table_location = smu_v11_0_set_driver_table_location, .set_tool_table_location = smu_v11_0_set_tool_table_location, .notify_memory_pool_location = smu_v11_0_notify_memory_pool_location, .system_features_control = smu_v11_0_system_features_control, @@ -2163,7 +2301,8 @@ static const struct pptable_funcs arcturus_ppt_funcs = { .baco_is_support= smu_v11_0_baco_is_support, .baco_get_state = smu_v11_0_baco_get_state, .baco_set_state = smu_v11_0_baco_set_state, - .baco_reset = smu_v11_0_baco_reset, + .baco_enter = smu_v11_0_baco_enter, + .baco_exit = smu_v11_0_baco_exit, .get_dpm_ultimate_freq = smu_v11_0_get_dpm_ultimate_freq, .set_soft_freq_limited_range = smu_v11_0_set_soft_freq_limited_range, .override_pcie_parameters = smu_v11_0_override_pcie_parameters, diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c b/drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c index cc57fb953e62..9454ab50f9a1 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c @@ -81,8 +81,8 @@ int phm_enable_dynamic_state_management(struct pp_hwmgr *hwmgr) adev = hwmgr->adev; /* Skip for suspend/resume case */ - if (smum_is_dpm_running(hwmgr) && !amdgpu_passthrough(adev) - && adev->in_suspend) { + if (!hwmgr->pp_one_vf && smum_is_dpm_running(hwmgr) + && !amdgpu_passthrough(adev) && adev->in_suspend) { pr_info("dpm has been enabled\n"); return 0; } @@ -99,6 +99,9 @@ int phm_disable_dynamic_state_management(struct pp_hwmgr *hwmgr) PHM_FUNC_CHECK(hwmgr); + if (!hwmgr->not_vf) + return 0; + if (!smum_is_dpm_running(hwmgr)) { pr_info("dpm has been disabled\n"); return 0; @@ -200,6 +203,9 @@ int phm_stop_thermal_controller(struct pp_hwmgr *hwmgr) { PHM_FUNC_CHECK(hwmgr); + if (!hwmgr->not_vf) + return 0; + if (hwmgr->hwmgr_func->stop_thermal_controller == NULL) return -EINVAL; @@ -237,6 +243,9 @@ int phm_start_thermal_controller(struct pp_hwmgr *hwmgr) TEMP_RANGE_MAX}; struct amdgpu_device *adev = hwmgr->adev; + if (!hwmgr->not_vf) + return 0; + if (hwmgr->hwmgr_func->get_thermal_temperature_range) hwmgr->hwmgr_func->get_thermal_temperature_range( hwmgr, &range); @@ -263,6 +272,8 @@ int phm_start_thermal_controller(struct pp_hwmgr *hwmgr) bool phm_check_smc_update_required_for_display_configuration(struct pp_hwmgr *hwmgr) { PHM_FUNC_CHECK(hwmgr); + if (hwmgr->pp_one_vf) + return false; if (hwmgr->hwmgr_func->check_smc_update_required_for_display_configuration == NULL) return false; @@ -482,6 +493,9 @@ int phm_disable_smc_firmware_ctf(struct pp_hwmgr *hwmgr) { PHM_FUNC_CHECK(hwmgr); + if (!hwmgr->not_vf) + return 0; + if (hwmgr->hwmgr_func->disable_smc_firmware_ctf == NULL) return -EINVAL; diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c index d2909c91d65b..f48fdc7f0382 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c @@ -221,6 +221,9 @@ int hwmgr_hw_init(struct pp_hwmgr *hwmgr) { int ret = 0; + hwmgr->pp_one_vf = amdgpu_sriov_is_pp_one_vf((struct amdgpu_device *)hwmgr->adev); + hwmgr->pm_en = (amdgpu_dpm && (hwmgr->not_vf || hwmgr->pp_one_vf)) + ? true : false; if (!hwmgr->pm_en) return 0; @@ -279,7 +282,7 @@ err: int hwmgr_hw_fini(struct pp_hwmgr *hwmgr) { - if (!hwmgr || !hwmgr->pm_en) + if (!hwmgr || !hwmgr->pm_en || !hwmgr->not_vf) return 0; phm_stop_thermal_controller(hwmgr); @@ -299,7 +302,7 @@ int hwmgr_suspend(struct pp_hwmgr *hwmgr) { int ret = 0; - if (!hwmgr || !hwmgr->pm_en) + if (!hwmgr || !hwmgr->pm_en || !hwmgr->not_vf) return 0; phm_disable_smc_firmware_ctf(hwmgr); @@ -321,7 +324,7 @@ int hwmgr_resume(struct pp_hwmgr *hwmgr) if (!hwmgr) return -EINVAL; - if (!hwmgr->pm_en) + if (!hwmgr->not_vf || !hwmgr->pm_en) return 0; ret = phm_setup_asic(hwmgr); @@ -365,6 +368,8 @@ int hwmgr_handle_task(struct pp_hwmgr *hwmgr, enum amd_pp_task task_id, switch (task_id) { case AMD_PP_TASK_DISPLAY_CONFIG_CHANGE: + if (!hwmgr->not_vf) + return ret; ret = phm_pre_display_configuration_changed(hwmgr); if (ret) return ret; @@ -381,6 +386,8 @@ int hwmgr_handle_task(struct pp_hwmgr *hwmgr, enum amd_pp_task task_id, enum PP_StateUILabel requested_ui_label; struct pp_power_state *requested_ps = NULL; + if (!hwmgr->not_vf) + return ret; if (user_state == NULL) { ret = -EINVAL; break; diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/pp_psm.c b/drivers/gpu/drm/amd/powerplay/hwmgr/pp_psm.c index 6bf48934fdc4..31a32a79cfc2 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/pp_psm.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/pp_psm.c @@ -262,20 +262,22 @@ int psm_adjust_power_state_dynamic(struct pp_hwmgr *hwmgr, bool skip_display_set uint32_t index; long workload; - if (!skip_display_settings) - phm_display_configuration_changed(hwmgr); - - if (hwmgr->ps) - power_state_management(hwmgr, new_ps); - else - /* - * for vega12/vega20 which does not support power state manager - * DAL clock limits should also be honoured - */ - phm_apply_clock_adjust_rules(hwmgr); - - if (!skip_display_settings) - phm_notify_smc_display_config_after_ps_adjustment(hwmgr); + if (hwmgr->not_vf) { + if (!skip_display_settings) + phm_display_configuration_changed(hwmgr); + + if (hwmgr->ps) + power_state_management(hwmgr, new_ps); + else + /* + * for vega12/vega20 which does not support power state manager + * DAL clock limits should also be honoured + */ + phm_apply_clock_adjust_rules(hwmgr); + + if (!skip_display_settings) + phm_notify_smc_display_config_after_ps_adjustment(hwmgr); + } if (!phm_force_dpm_levels(hwmgr, hwmgr->request_dpm_level)) hwmgr->dpm_level = hwmgr->request_dpm_level; diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c index 1115761982a7..4e8ab139bb3b 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c @@ -1151,12 +1151,11 @@ static int smu10_set_watermarks_for_clocks_ranges(struct pp_hwmgr *hwmgr, struct smu10_hwmgr *data = hwmgr->backend; struct dm_pp_wm_sets_with_clock_ranges_soc15 *wm_with_clock_ranges = clock_ranges; Watermarks_t *table = &(data->water_marks_table); - int result = 0; smu_set_watermarks_for_clocks_ranges(table,wm_with_clock_ranges); smum_smc_table_manager(hwmgr, (uint8_t *)table, (uint16_t)SMU10_WMTABLE, false); data->water_marks_exist = true; - return result; + return 0; } static int smu10_smus_notify_pwe(struct pp_hwmgr *hwmgr) diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c index f73dff68e799..d70abada66bf 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c @@ -4238,7 +4238,6 @@ static int smu7_check_mc_firmware(struct pp_hwmgr *hwmgr) { struct smu7_hwmgr *data = (struct smu7_hwmgr *)(hwmgr->backend); - uint32_t vbios_version; uint32_t tmp; /* Read MC indirect register offset 0x9F bits [3:0] to see @@ -4247,7 +4246,6 @@ static int smu7_check_mc_firmware(struct pp_hwmgr *hwmgr) */ smu7_get_mc_microcode_version(hwmgr); - vbios_version = hwmgr->microcode_version_info.MC & 0xf; data->need_long_memory_training = false; @@ -4945,7 +4943,7 @@ static int smu7_get_power_profile_mode(struct pp_hwmgr *hwmgr, char *buf) title[0], title[1], title[2], title[3], title[4], title[5], title[6], title[7]); - len = sizeof(smu7_profiling) / sizeof(struct profile_mode_setting); + len = ARRAY_SIZE(smu7_profiling); for (i = 0; i < len; i++) { if (i == hwmgr->power_profile_mode) { @@ -5077,13 +5075,11 @@ static int smu7_get_performance_level(struct pp_hwmgr *hwmgr, const struct pp_hw PHM_PerformanceLevel *level) { const struct smu7_power_state *ps; - struct smu7_hwmgr *data; uint32_t i; if (level == NULL || hwmgr == NULL || state == NULL) return -EINVAL; - data = hwmgr->backend; ps = cast_const_phw_smu7_power_state(state); i = index > ps->performance_level_count - 1 ? @@ -5189,13 +5185,11 @@ uint8_t smu7_get_sleep_divider_id_from_clock(uint32_t clock, int smu7_init_function_pointers(struct pp_hwmgr *hwmgr) { - int ret = 0; - hwmgr->hwmgr_func = &smu7_hwmgr_funcs; if (hwmgr->pp_table_version == PP_TABLE_V0) hwmgr->pptable_func = &pptable_funcs; else if (hwmgr->pp_table_version == PP_TABLE_V1) hwmgr->pptable_func = &pptable_v1_0_funcs; - return ret; + return 0; } diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c index d71a492c87a3..92a65e3daff4 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c @@ -912,6 +912,9 @@ static int vega10_hwmgr_backend_init(struct pp_hwmgr *hwmgr) hwmgr->platform_descriptor.clockStep.memoryClock = 500; data->total_active_cus = adev->gfx.cu_info.number; + if (!hwmgr->not_vf) + return result; + /* Setup default Overdrive Fan control settings */ data->odn_fan_table.target_fan_speed = hwmgr->thermal_controller.advanceFanControlParameters.usMaxFanRPM; @@ -979,6 +982,9 @@ static int vega10_setup_dpm_led_config(struct pp_hwmgr *hwmgr) static int vega10_setup_asic_task(struct pp_hwmgr *hwmgr) { + if (!hwmgr->not_vf) + return 0; + PP_ASSERT_WITH_CODE(!vega10_init_sclk_threshold(hwmgr), "Failed to init sclk threshold!", return -EINVAL); @@ -2503,6 +2509,9 @@ static int vega10_init_smc_table(struct pp_hwmgr *hwmgr) "Failed to setup default DPM tables!", return result); + if (!hwmgr->not_vf) + return 0; + /* initialize ODN table */ if (hwmgr->od_enabled) { if (odn_table->max_vddc) { @@ -2826,6 +2835,8 @@ static int vega10_stop_dpm(struct pp_hwmgr *hwmgr, uint32_t bitmap) struct vega10_hwmgr *data = hwmgr->backend; uint32_t i, feature_mask = 0; + if (!hwmgr->not_vf) + return 0; if(data->smu_features[GNLD_LED_DISPLAY].supported == true){ PP_ASSERT_WITH_CODE(!vega10_enable_smc_features(hwmgr, @@ -2932,61 +2943,73 @@ static int vega10_enable_dpm_tasks(struct pp_hwmgr *hwmgr) struct vega10_hwmgr *data = hwmgr->backend; int tmp_result, result = 0; - vega10_enable_disable_PCC_limit_feature(hwmgr, true); - - smum_send_msg_to_smc_with_parameter(hwmgr, - PPSMC_MSG_ConfigureTelemetry, data->config_telemetry); + if (hwmgr->not_vf) { + vega10_enable_disable_PCC_limit_feature(hwmgr, true); - tmp_result = vega10_construct_voltage_tables(hwmgr); - PP_ASSERT_WITH_CODE(!tmp_result, - "Failed to construct voltage tables!", - result = tmp_result); + smum_send_msg_to_smc_with_parameter(hwmgr, + PPSMC_MSG_ConfigureTelemetry, data->config_telemetry); - tmp_result = vega10_init_smc_table(hwmgr); - PP_ASSERT_WITH_CODE(!tmp_result, - "Failed to initialize SMC table!", - result = tmp_result); + tmp_result = vega10_construct_voltage_tables(hwmgr); + PP_ASSERT_WITH_CODE(!tmp_result, + "Failed to construct voltage tables!", + result = tmp_result); + } - if (PP_CAP(PHM_PlatformCaps_ThermalController)) { - tmp_result = vega10_enable_thermal_protection(hwmgr); + if (hwmgr->not_vf || hwmgr->pp_one_vf) { + tmp_result = vega10_init_smc_table(hwmgr); PP_ASSERT_WITH_CODE(!tmp_result, - "Failed to enable thermal protection!", - result = tmp_result); + "Failed to initialize SMC table!", + result = tmp_result); } - tmp_result = vega10_enable_vrhot_feature(hwmgr); - PP_ASSERT_WITH_CODE(!tmp_result, - "Failed to enable VR hot feature!", - result = tmp_result); + if (hwmgr->not_vf) { + if (PP_CAP(PHM_PlatformCaps_ThermalController)) { + tmp_result = vega10_enable_thermal_protection(hwmgr); + PP_ASSERT_WITH_CODE(!tmp_result, + "Failed to enable thermal protection!", + result = tmp_result); + } - tmp_result = vega10_enable_deep_sleep_master_switch(hwmgr); - PP_ASSERT_WITH_CODE(!tmp_result, - "Failed to enable deep sleep master switch!", - result = tmp_result); + tmp_result = vega10_enable_vrhot_feature(hwmgr); + PP_ASSERT_WITH_CODE(!tmp_result, + "Failed to enable VR hot feature!", + result = tmp_result); - tmp_result = vega10_start_dpm(hwmgr, SMC_DPM_FEATURES); - PP_ASSERT_WITH_CODE(!tmp_result, - "Failed to start DPM!", result = tmp_result); + tmp_result = vega10_enable_deep_sleep_master_switch(hwmgr); + PP_ASSERT_WITH_CODE(!tmp_result, + "Failed to enable deep sleep master switch!", + result = tmp_result); + } - /* enable didt, do not abort if failed didt */ - tmp_result = vega10_enable_didt_config(hwmgr); - PP_ASSERT(!tmp_result, - "Failed to enable didt config!"); + if (hwmgr->not_vf) { + tmp_result = vega10_start_dpm(hwmgr, SMC_DPM_FEATURES); + PP_ASSERT_WITH_CODE(!tmp_result, + "Failed to start DPM!", result = tmp_result); + } + + if (hwmgr->not_vf) { + /* enable didt, do not abort if failed didt */ + tmp_result = vega10_enable_didt_config(hwmgr); + PP_ASSERT(!tmp_result, + "Failed to enable didt config!"); + } tmp_result = vega10_enable_power_containment(hwmgr); PP_ASSERT_WITH_CODE(!tmp_result, - "Failed to enable power containment!", - result = tmp_result); + "Failed to enable power containment!", + result = tmp_result); - tmp_result = vega10_power_control_set_level(hwmgr); - PP_ASSERT_WITH_CODE(!tmp_result, - "Failed to power control set level!", - result = tmp_result); + if (hwmgr->not_vf) { + tmp_result = vega10_power_control_set_level(hwmgr); + PP_ASSERT_WITH_CODE(!tmp_result, + "Failed to power control set level!", + result = tmp_result); - tmp_result = vega10_enable_ulv(hwmgr); - PP_ASSERT_WITH_CODE(!tmp_result, - "Failed to enable ULV!", - result = tmp_result); + tmp_result = vega10_enable_ulv(hwmgr); + PP_ASSERT_WITH_CODE(!tmp_result, + "Failed to enable ULV!", + result = tmp_result); + } return result; } @@ -3080,11 +3103,22 @@ static int vega10_get_pp_table_entry_callback_func(struct pp_hwmgr *hwmgr, performance_level->soc_clock = socclk_dep_table->entries [state_entry->ucSocClockIndexHigh].ulClk; if (gfxclk_dep_table->ucRevId == 0) { - performance_level->gfx_clock = gfxclk_dep_table->entries - [state_entry->ucGfxClockIndexHigh].ulClk; + /* under vega10 pp one vf mode, the gfx clk dpm need be lower + * to level-4 due to the limited 110w-power + */ + if (hwmgr->pp_one_vf && (state_entry->ucGfxClockIndexHigh > 0)) + performance_level->gfx_clock = + gfxclk_dep_table->entries[4].ulClk; + else + performance_level->gfx_clock = gfxclk_dep_table->entries + [state_entry->ucGfxClockIndexHigh].ulClk; } else if (gfxclk_dep_table->ucRevId == 1) { patom_record_V2 = (ATOM_Vega10_GFXCLK_Dependency_Record_V2 *)gfxclk_dep_table->entries; - performance_level->gfx_clock = patom_record_V2[state_entry->ucGfxClockIndexHigh].ulClk; + if (hwmgr->pp_one_vf && (state_entry->ucGfxClockIndexHigh > 0)) + performance_level->gfx_clock = patom_record_V2[4].ulClk; + else + performance_level->gfx_clock = + patom_record_V2[state_entry->ucGfxClockIndexHigh].ulClk; } performance_level->mem_clock = mclk_dep_table->entries @@ -3495,6 +3529,7 @@ static int vega10_upload_dpm_bootup_level(struct pp_hwmgr *hwmgr) smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetSoftMinGfxclkByIndex, data->smc_state_table.gfx_boot_level); + data->dpm_table.gfx_table.dpm_state.soft_min_level = data->smc_state_table.gfx_boot_level; } @@ -3503,7 +3538,8 @@ static int vega10_upload_dpm_bootup_level(struct pp_hwmgr *hwmgr) if (!data->registry_data.mclk_dpm_key_disabled) { if (data->smc_state_table.mem_boot_level != data->dpm_table.mem_table.dpm_state.soft_min_level) { - if (data->smc_state_table.mem_boot_level == NUM_UCLK_DPM_LEVELS - 1) { + if ((data->smc_state_table.mem_boot_level == NUM_UCLK_DPM_LEVELS - 1) + && hwmgr->not_vf) { socclk_idx = vega10_get_soc_index_for_max_uclk(hwmgr); smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetSoftMinSocclkByIndex, @@ -3518,6 +3554,9 @@ static int vega10_upload_dpm_bootup_level(struct pp_hwmgr *hwmgr) } } + if (!hwmgr->not_vf) + return 0; + if (!data->registry_data.socclk_dpm_key_disabled) { if (data->smc_state_table.soc_boot_level != data->dpm_table.soc_table.dpm_state.soft_min_level) { @@ -3560,6 +3599,9 @@ static int vega10_upload_dpm_max_level(struct pp_hwmgr *hwmgr) } } + if (!hwmgr->not_vf) + return 0; + if (!data->registry_data.socclk_dpm_key_disabled) { if (data->smc_state_table.soc_max_level != data->dpm_table.soc_table.dpm_state.soft_max_level) { @@ -4054,15 +4096,25 @@ static int vega10_get_profiling_clk_mask(struct pp_hwmgr *hwmgr, enum amd_dpm_fo } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK) { *mclk_mask = 0; } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) { - *sclk_mask = table_info->vdd_dep_on_sclk->count - 1; + /* under vega10 pp one vf mode, the gfx clk dpm need be lower + * to level-4 due to the limited power + */ + if (hwmgr->pp_one_vf) + *sclk_mask = 4; + else + *sclk_mask = table_info->vdd_dep_on_sclk->count - 1; *soc_mask = table_info->vdd_dep_on_socclk->count - 1; *mclk_mask = table_info->vdd_dep_on_mclk->count - 1; } + return 0; } static void vega10_set_fan_control_mode(struct pp_hwmgr *hwmgr, uint32_t mode) { + if (!hwmgr->not_vf) + return; + switch (mode) { case AMD_FAN_CTRL_NONE: vega10_fan_ctrl_set_fan_speed_percent(hwmgr, 100); @@ -4176,6 +4228,9 @@ static int vega10_dpm_force_dpm_level(struct pp_hwmgr *hwmgr, break; } + if (!hwmgr->not_vf) + return ret; + if (!ret) { if (level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK && hwmgr->dpm_level != AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) vega10_set_fan_control_mode(hwmgr, AMD_FAN_CTRL_NONE); @@ -4360,14 +4415,13 @@ static int vega10_set_watermarks_for_clocks_ranges(struct pp_hwmgr *hwmgr, struct vega10_hwmgr *data = hwmgr->backend; struct dm_pp_wm_sets_with_clock_ranges_soc15 *wm_with_clock_ranges = clock_range; Watermarks_t *table = &(data->smc_state_table.water_marks_table); - int result = 0; if (!data->registry_data.disable_water_mark) { smu_set_watermarks_for_clocks_ranges(table, wm_with_clock_ranges); data->water_marks_bitmap = WaterMarksExist; } - return result; + return 0; } static int vega10_get_ppfeature_status(struct pp_hwmgr *hwmgr, char *buf) @@ -4480,7 +4534,7 @@ static int vega10_print_clock_levels(struct pp_hwmgr *hwmgr, struct vega10_pcie_table *pcie_table = &(data->dpm_table.pcie_table); struct vega10_odn_clock_voltage_dependency_table *podn_vdd_dep = NULL; - int i, now, size = 0; + int i, now, size = 0, count = 0; switch (type) { case PP_SCLK: @@ -4490,7 +4544,12 @@ static int vega10_print_clock_levels(struct pp_hwmgr *hwmgr, smum_send_msg_to_smc(hwmgr, PPSMC_MSG_GetCurrentGfxclkIndex); now = smum_get_argument(hwmgr); - for (i = 0; i < sclk_table->count; i++) + if (hwmgr->pp_one_vf && + (hwmgr->dpm_level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK)) + count = 5; + else + count = sclk_table->count; + for (i = 0; i < count; i++) size += sprintf(buf + size, "%d: %uMhz %s\n", i, sclk_table->dpm_levels[i].value / 100, (i == now) ? "*" : ""); @@ -4701,6 +4760,9 @@ static int vega10_disable_dpm_tasks(struct pp_hwmgr *hwmgr) { int tmp_result, result = 0; + if (!hwmgr->not_vf) + return 0; + if (PP_CAP(PHM_PlatformCaps_ThermalController)) vega10_disable_thermal_protection(hwmgr); @@ -5252,13 +5314,11 @@ static int vega10_get_performance_level(struct pp_hwmgr *hwmgr, const struct pp_ PHM_PerformanceLevel *level) { const struct vega10_power_state *ps; - struct vega10_hwmgr *data; uint32_t i; if (level == NULL || hwmgr == NULL || state == NULL) return -EINVAL; - data = hwmgr->backend; ps = cast_const_phw_vega10_power_state(state); i = index > ps->performance_level_count - 1 ? diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_powertune.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_powertune.c index 6f26cb241ecc..0a677d4bc87b 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_powertune.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_powertune.c @@ -1343,6 +1343,9 @@ int vega10_enable_power_containment(struct pp_hwmgr *hwmgr) hwmgr->default_power_limit = hwmgr->power_limit = (uint32_t)(tdp_table->usMaximumPowerDeliveryLimit); + if (!hwmgr->not_vf) + return 0; + if (PP_CAP(PHM_PlatformCaps_PowerContainment)) { if (data->smu_features[GNLD_PPT].supported) PP_ASSERT_WITH_CODE(!vega10_enable_smc_features(hwmgr, diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c index 7af9ad450ac4..aca61d1ff3c2 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c @@ -499,8 +499,6 @@ static int vega12_get_number_of_dpm_level(struct pp_hwmgr *hwmgr, static int vega12_get_dpm_frequency_by_index(struct pp_hwmgr *hwmgr, PPCLK_e clkID, uint32_t index, uint32_t *clock) { - int result = 0; - /* *SMU expects the Clock ID to be in the top 16 bits. *Lower 16 bits specify the level @@ -512,7 +510,7 @@ static int vega12_get_dpm_frequency_by_index(struct pp_hwmgr *hwmgr, *clock = smum_get_argument(hwmgr); - return result; + return 0; } static int vega12_setup_single_dpm_table(struct pp_hwmgr *hwmgr, diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c index 5bcf0d684151..3b3ec5666051 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c @@ -872,7 +872,7 @@ static int vega20_override_pcie_parameters(struct pp_hwmgr *hwmgr) "[OverridePcieParameters] Attempt to override pcie params failed!", return ret); - data->pcie_parameters_override = 1; + data->pcie_parameters_override = true; data->pcie_gen_level1 = pcie_gen; data->pcie_width_level1 = pcie_width; diff --git a/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h b/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h index 41fce75b263f..b0591a8dda41 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +++ b/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h @@ -254,11 +254,21 @@ struct smu_table_context unsigned long metrics_time; void *metrics_table; void *clocks_table; + void *watermarks_table; void *max_sustainable_clocks; struct smu_bios_boot_up_values boot_values; void *driver_pptable; struct smu_table *tables; + /* + * The driver table is just a staging buffer for + * uploading/downloading content from the SMU. + * + * And the table_id for SMU_MSG_TransferTableSmu2Dram/ + * SMU_MSG_TransferTableDram2Smu instructs SMU + * which content driver is interested. + */ + struct smu_table driver_table; struct smu_table memory_pool; uint8_t thermal_controller_type; @@ -282,6 +292,7 @@ struct smu_power_gate { bool uvd_gated; bool vce_gated; bool vcn_gated; + bool jpeg_gated; }; struct smu_power_context { @@ -437,12 +448,13 @@ struct pptable_funcs { int (*set_power_profile_mode)(struct smu_context *smu, long *input, uint32_t size); int (*dpm_set_uvd_enable)(struct smu_context *smu, bool enable); int (*dpm_set_vce_enable)(struct smu_context *smu, bool enable); + int (*dpm_set_jpeg_enable)(struct smu_context *smu, bool enable); int (*read_sensor)(struct smu_context *smu, enum amd_pp_sensors sensor, void *data, uint32_t *size); int (*pre_display_config_changed)(struct smu_context *smu); int (*display_config_changed)(struct smu_context *smu); int (*apply_clocks_adjust_rules)(struct smu_context *smu); - int (*notify_smc_dispaly_config)(struct smu_context *smu); + int (*notify_smc_display_config)(struct smu_context *smu); int (*force_dpm_limit_value)(struct smu_context *smu, bool highest); int (*unforce_dpm_levels)(struct smu_context *smu); int (*get_profiling_clk_mask)(struct smu_context *smu, @@ -491,9 +503,11 @@ struct pptable_funcs { int (*check_fw_version)(struct smu_context *smu); int (*powergate_sdma)(struct smu_context *smu, bool gate); int (*powergate_vcn)(struct smu_context *smu, bool gate); + int (*powergate_jpeg)(struct smu_context *smu, bool gate); int (*set_gfx_cgpg)(struct smu_context *smu, bool enable); int (*write_pptable)(struct smu_context *smu); int (*set_min_dcef_deep_sleep)(struct smu_context *smu); + int (*set_driver_table_location)(struct smu_context *smu); int (*set_tool_table_location)(struct smu_context *smu); int (*notify_memory_pool_location)(struct smu_context *smu); int (*set_last_dcef_min_deep_sleep_clk)(struct smu_context *smu); @@ -544,7 +558,8 @@ struct pptable_funcs { bool (*baco_is_support)(struct smu_context *smu); enum smu_baco_state (*baco_get_state)(struct smu_context *smu); int (*baco_set_state)(struct smu_context *smu, enum smu_baco_state state); - int (*baco_reset)(struct smu_context *smu); + int (*baco_enter)(struct smu_context *smu); + int (*baco_exit)(struct smu_context *smu); int (*mode2_reset)(struct smu_context *smu); int (*get_dpm_ultimate_freq)(struct smu_context *smu, enum smu_clk_type clk_type, uint32_t *min, uint32_t *max); int (*set_soft_freq_limited_range)(struct smu_context *smu, enum smu_clk_type clk_type, uint32_t min, uint32_t max); @@ -626,7 +641,8 @@ bool smu_baco_is_support(struct smu_context *smu); int smu_baco_get_state(struct smu_context *smu, enum smu_baco_state *state); -int smu_baco_reset(struct smu_context *smu); +int smu_baco_enter(struct smu_context *smu); +int smu_baco_exit(struct smu_context *smu); int smu_mode2_reset(struct smu_context *smu); @@ -692,6 +708,8 @@ int smu_set_soft_freq_range(struct smu_context *smu, enum smu_clk_type clk_type, uint32_t min, uint32_t max); int smu_set_hard_freq_range(struct smu_context *smu, enum smu_clk_type clk_type, uint32_t min, uint32_t max); +int smu_get_dpm_level_range(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t *min_value, uint32_t *max_value); enum amd_dpm_forced_level smu_get_performance_level(struct smu_context *smu); int smu_force_performance_level(struct smu_context *smu, enum amd_dpm_forced_level level); int smu_set_display_count(struct smu_context *smu, uint32_t count); diff --git a/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h b/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h index af977675fd33..2ffb666b97e6 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h +++ b/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h @@ -741,6 +741,7 @@ struct pp_hwmgr { uint32_t smu_version; bool not_vf; bool pm_en; + bool pp_one_vf; struct mutex smu_lock; uint32_t pp_table_version; diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu11_driver_if_arcturus.h b/drivers/gpu/drm/amd/powerplay/inc/smu11_driver_if_arcturus.h index a886f0644d24..ce5b5011c122 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/smu11_driver_if_arcturus.h +++ b/drivers/gpu/drm/amd/powerplay/inc/smu11_driver_if_arcturus.h @@ -622,8 +622,14 @@ typedef struct { uint16_t PccThresholdHigh; uint32_t PaddingAPCC[6]; //FIXME pending SPEC + // OOB Settings + uint16_t BasePerformanceCardPower; + uint16_t MaxPerformanceCardPower; + uint16_t BasePerformanceFrequencyCap; //In Mhz + uint16_t MaxPerformanceFrequencyCap; //In Mhz + // SECTION: Reserved - uint32_t Reserved[11]; + uint32_t Reserved[9]; // SECTION: BOARD PARAMETERS @@ -823,7 +829,6 @@ typedef struct { uint32_t MmHubPadding[8]; // SMU internal use } AvfsFuseOverride_t; -/* NOT CURRENTLY USED typedef struct { uint8_t Gfx_ActiveHystLimit; uint8_t Gfx_IdleHystLimit; @@ -866,7 +871,6 @@ typedef struct { uint32_t MmHubPadding[8]; // SMU internal use } DpmActivityMonitorCoeffInt_t; -*/ // These defines are used with the following messages: // SMC_MSG_TransferTableDram2Smu @@ -878,11 +882,11 @@ typedef struct { #define TABLE_PMSTATUSLOG 4 #define TABLE_SMU_METRICS 5 #define TABLE_DRIVER_SMU_CONFIG 6 -//#define TABLE_ACTIVITY_MONITOR_COEFF 7 #define TABLE_OVERDRIVE 7 #define TABLE_WAFL_XGMI_TOPOLOGY 8 #define TABLE_I2C_COMMANDS 9 -#define TABLE_COUNT 10 +#define TABLE_ACTIVITY_MONITOR_COEFF 10 +#define TABLE_COUNT 11 // These defines are used with the SMC_MSG_SetUclkFastSwitch message. typedef enum { diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu12_driver_if.h b/drivers/gpu/drm/amd/powerplay/inc/smu12_driver_if.h index c27c82851468..2f85a34c0591 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/smu12_driver_if.h +++ b/drivers/gpu/drm/amd/powerplay/inc/smu12_driver_if.h @@ -27,7 +27,7 @@ // *** IMPORTANT *** // SMU TEAM: Always increment the interface version if // any structure is changed in this file -#define SMU12_DRIVER_IF_VERSION 10 +#define SMU12_DRIVER_IF_VERSION 11 typedef struct { int32_t value; @@ -192,6 +192,11 @@ typedef struct { uint16_t SocTemperature; //[centi-Celsius] uint16_t ThrottlerStatus; uint16_t spare; + + uint16_t StapmOriginalLimit; //[mW] + uint16_t StapmCurrentLimit; //[mW] + uint16_t ApuPower; //[mW] + uint16_t dGpuPower; //[mW] } SmuMetrics_t; diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h b/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h index 719844257713..d5314d12628a 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h +++ b/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h @@ -27,7 +27,7 @@ #define SMU11_DRIVER_IF_VERSION_INV 0xFFFFFFFF #define SMU11_DRIVER_IF_VERSION_VG20 0x13 -#define SMU11_DRIVER_IF_VERSION_ARCT 0x10 +#define SMU11_DRIVER_IF_VERSION_ARCT 0x12 #define SMU11_DRIVER_IF_VERSION_NV10 0x33 #define SMU11_DRIVER_IF_VERSION_NV14 0x34 @@ -170,6 +170,8 @@ int smu_v11_0_write_pptable(struct smu_context *smu); int smu_v11_0_set_min_dcef_deep_sleep(struct smu_context *smu); +int smu_v11_0_set_driver_table_location(struct smu_context *smu); + int smu_v11_0_set_tool_table_location(struct smu_context *smu); int smu_v11_0_notify_memory_pool_location(struct smu_context *smu); @@ -247,7 +249,8 @@ enum smu_baco_state smu_v11_0_baco_get_state(struct smu_context *smu); int smu_v11_0_baco_set_state(struct smu_context *smu, enum smu_baco_state state); -int smu_v11_0_baco_reset(struct smu_context *smu); +int smu_v11_0_baco_enter(struct smu_context *smu); +int smu_v11_0_baco_exit(struct smu_context *smu); int smu_v11_0_get_dpm_ultimate_freq(struct smu_context *smu, enum smu_clk_type clk_type, uint32_t *min, uint32_t *max); @@ -261,4 +264,7 @@ int smu_v11_0_set_default_od_settings(struct smu_context *smu, bool initialize, uint32_t smu_v11_0_get_max_power_limit(struct smu_context *smu); +int smu_v11_0_set_performance_level(struct smu_context *smu, + enum amd_dpm_forced_level level); + #endif diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu_v12_0.h b/drivers/gpu/drm/amd/powerplay/inc/smu_v12_0.h index 9d81d789c713..d79e54b5ebf6 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/smu_v12_0.h +++ b/drivers/gpu/drm/amd/powerplay/inc/smu_v12_0.h @@ -57,8 +57,14 @@ int smu_v12_0_powergate_sdma(struct smu_context *smu, bool gate); int smu_v12_0_powergate_vcn(struct smu_context *smu, bool gate); +int smu_v12_0_powergate_jpeg(struct smu_context *smu, bool gate); + int smu_v12_0_set_gfx_cgpg(struct smu_context *smu, bool enable); +int smu_v12_0_read_sensor(struct smu_context *smu, + enum amd_pp_sensors sensor, + void *data, uint32_t *size); + uint32_t smu_v12_0_get_gfxoff_status(struct smu_context *smu); int smu_v12_0_gfx_off_control(struct smu_context *smu, bool enable); @@ -69,6 +75,13 @@ int smu_v12_0_fini_smc_tables(struct smu_context *smu); int smu_v12_0_populate_smc_tables(struct smu_context *smu); +int smu_v12_0_get_enabled_mask(struct smu_context *smu, + uint32_t *feature_mask, uint32_t num); + +int smu_v12_0_get_current_clk_freq(struct smu_context *smu, + enum smu_clk_type clk_id, + uint32_t *value); + int smu_v12_0_get_dpm_ultimate_freq(struct smu_context *smu, enum smu_clk_type clk_type, uint32_t *min, uint32_t *max); @@ -77,4 +90,6 @@ int smu_v12_0_mode2_reset(struct smu_context *smu); int smu_v12_0_set_soft_freq_limited_range(struct smu_context *smu, enum smu_clk_type clk_type, uint32_t min, uint32_t max); +int smu_v12_0_set_driver_table_location(struct smu_context *smu); + #endif diff --git a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c index ca62e92e5a4f..93c66c69ca28 100644 --- a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c +++ b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c @@ -384,8 +384,10 @@ navi10_get_allowed_feature_mask(struct smu_context *smu, *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_ATHUB_PG_BIT); if (smu->adev->pg_flags & AMD_PG_SUPPORT_VCN) - *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_VCN_PG_BIT) - | FEATURE_MASK(FEATURE_JPEG_PG_BIT); + *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_VCN_PG_BIT); + + if (smu->adev->pg_flags & AMD_PG_SUPPORT_JPEG) + *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_JPEG_PG_BIT); /* disable DPM UCLK and DS SOCCLK on navi10 A0 secure board */ if (is_asic_secure(smu)) { @@ -553,6 +555,10 @@ static int navi10_tables_init(struct smu_context *smu, struct smu_table *tables) return -ENOMEM; smu_table->metrics_time = 0; + smu_table->watermarks_table = kzalloc(sizeof(Watermarks_t), GFP_KERNEL); + if (!smu_table->watermarks_table) + return -ENOMEM; + return 0; } @@ -668,6 +674,31 @@ static int navi10_dpm_set_uvd_enable(struct smu_context *smu, bool enable) return ret; } +static int navi10_dpm_set_jpeg_enable(struct smu_context *smu, bool enable) +{ + struct smu_power_context *smu_power = &smu->smu_power; + struct smu_power_gate *power_gate = &smu_power->power_gate; + int ret = 0; + + if (enable) { + if (smu_feature_is_enabled(smu, SMU_FEATURE_JPEG_PG_BIT)) { + ret = smu_send_smc_msg(smu, SMU_MSG_PowerUpJpeg); + if (ret) + return ret; + } + power_gate->jpeg_gated = false; + } else { + if (smu_feature_is_enabled(smu, SMU_FEATURE_JPEG_PG_BIT)) { + ret = smu_send_smc_msg(smu, SMU_MSG_PowerDownJpeg); + if (ret) + return ret; + } + power_gate->jpeg_gated = true; + } + + return ret; +} + static int navi10_get_current_clk_freq_by_table(struct smu_context *smu, enum smu_clk_type clk_type, uint32_t *value) @@ -1350,7 +1381,7 @@ static int navi10_get_profiling_clk_mask(struct smu_context *smu, return ret; } -static int navi10_notify_smc_dispaly_config(struct smu_context *smu) +static int navi10_notify_smc_display_config(struct smu_context *smu) { struct smu_clocks min_clocks = {0}; struct pp_display_clock_request clock_req; @@ -1555,12 +1586,44 @@ static int navi10_get_uclk_dpm_states(struct smu_context *smu, uint32_t *clocks_ return 0; } -static int navi10_set_peak_clock_by_device(struct smu_context *smu) +static int navi10_set_performance_level(struct smu_context *smu, + enum amd_dpm_forced_level level); + +static int navi10_set_standard_performance_level(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + int ret = 0; + uint32_t sclk_freq = 0, uclk_freq = 0; + + switch (adev->asic_type) { + case CHIP_NAVI10: + sclk_freq = NAVI10_UMD_PSTATE_PROFILING_GFXCLK; + uclk_freq = NAVI10_UMD_PSTATE_PROFILING_MEMCLK; + break; + case CHIP_NAVI14: + sclk_freq = NAVI14_UMD_PSTATE_PROFILING_GFXCLK; + uclk_freq = NAVI14_UMD_PSTATE_PROFILING_MEMCLK; + break; + default: + /* by default, this is same as auto performance level */ + return navi10_set_performance_level(smu, AMD_DPM_FORCED_LEVEL_AUTO); + } + + ret = smu_set_soft_freq_range(smu, SMU_SCLK, sclk_freq, sclk_freq); + if (ret) + return ret; + ret = smu_set_soft_freq_range(smu, SMU_UCLK, uclk_freq, uclk_freq); + if (ret) + return ret; + + return ret; +} + +static int navi10_set_peak_performance_level(struct smu_context *smu) { struct amdgpu_device *adev = smu->adev; int ret = 0; uint32_t sclk_freq = 0, uclk_freq = 0; - uint32_t uclk_level = 0; switch (adev->asic_type) { case CHIP_NAVI10: @@ -1601,14 +1664,16 @@ static int navi10_set_peak_clock_by_device(struct smu_context *smu) break; } break; + case CHIP_NAVI12: + sclk_freq = NAVI12_UMD_PSTATE_PEAK_GFXCLK; + break; default: - return -EINVAL; + ret = smu_get_dpm_level_range(smu, SMU_SCLK, NULL, &sclk_freq); + if (ret) + return ret; } - ret = smu_get_dpm_level_count(smu, SMU_UCLK, &uclk_level); - if (ret) - return ret; - ret = smu_get_dpm_freq_by_index(smu, SMU_UCLK, uclk_level - 1, &uclk_freq); + ret = smu_get_dpm_level_range(smu, SMU_UCLK, NULL, &uclk_freq); if (ret) return ret; @@ -1622,19 +1687,45 @@ static int navi10_set_peak_clock_by_device(struct smu_context *smu) return ret; } -static int navi10_set_performance_level(struct smu_context *smu, enum amd_dpm_forced_level level) +static int navi10_set_performance_level(struct smu_context *smu, + enum amd_dpm_forced_level level) { int ret = 0; + uint32_t sclk_mask, mclk_mask, soc_mask; switch (level) { + case AMD_DPM_FORCED_LEVEL_HIGH: + ret = smu_force_dpm_limit_value(smu, true); + break; + case AMD_DPM_FORCED_LEVEL_LOW: + ret = smu_force_dpm_limit_value(smu, false); + break; + case AMD_DPM_FORCED_LEVEL_AUTO: + ret = smu_unforce_dpm_levels(smu); + break; + case AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD: + ret = navi10_set_standard_performance_level(smu); + break; + case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK: + case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK: + ret = smu_get_profiling_clk_mask(smu, level, + &sclk_mask, + &mclk_mask, + &soc_mask); + if (ret) + return ret; + smu_force_clk_levels(smu, SMU_SCLK, 1 << sclk_mask, false); + smu_force_clk_levels(smu, SMU_MCLK, 1 << mclk_mask, false); + smu_force_clk_levels(smu, SMU_SOCCLK, 1 << soc_mask, false); + break; case AMD_DPM_FORCED_LEVEL_PROFILE_PEAK: - ret = navi10_set_peak_clock_by_device(smu); + ret = navi10_set_peak_performance_level(smu); break; + case AMD_DPM_FORCED_LEVEL_MANUAL: + case AMD_DPM_FORCED_LEVEL_PROFILE_EXIT: default: - ret = -EINVAL; break; } - return ret; } @@ -2015,6 +2106,7 @@ static const struct pptable_funcs navi10_ppt_funcs = { .get_allowed_feature_mask = navi10_get_allowed_feature_mask, .set_default_dpm_table = navi10_set_default_dpm_table, .dpm_set_uvd_enable = navi10_dpm_set_uvd_enable, + .dpm_set_jpeg_enable = navi10_dpm_set_jpeg_enable, .get_current_clk_freq_by_table = navi10_get_current_clk_freq_by_table, .print_clk_levels = navi10_print_clk_levels, .force_clk_levels = navi10_force_clk_levels, @@ -2022,7 +2114,7 @@ static const struct pptable_funcs navi10_ppt_funcs = { .get_clock_by_type_with_latency = navi10_get_clock_by_type_with_latency, .pre_display_config_changed = navi10_pre_display_config_changed, .display_config_changed = navi10_display_config_changed, - .notify_smc_dispaly_config = navi10_notify_smc_dispaly_config, + .notify_smc_display_config = navi10_notify_smc_display_config, .force_dpm_limit_value = navi10_force_dpm_limit_value, .unforce_dpm_levels = navi10_unforce_dpm_levels, .is_dpm_running = navi10_is_dpm_running, @@ -2055,6 +2147,7 @@ static const struct pptable_funcs navi10_ppt_funcs = { .check_fw_version = smu_v11_0_check_fw_version, .write_pptable = smu_v11_0_write_pptable, .set_min_dcef_deep_sleep = smu_v11_0_set_min_dcef_deep_sleep, + .set_driver_table_location = smu_v11_0_set_driver_table_location, .set_tool_table_location = smu_v11_0_set_tool_table_location, .notify_memory_pool_location = smu_v11_0_notify_memory_pool_location, .system_features_control = smu_v11_0_system_features_control, @@ -2083,7 +2176,8 @@ static const struct pptable_funcs navi10_ppt_funcs = { .baco_is_support= smu_v11_0_baco_is_support, .baco_get_state = smu_v11_0_baco_get_state, .baco_set_state = smu_v11_0_baco_set_state, - .baco_reset = smu_v11_0_baco_reset, + .baco_enter = smu_v11_0_baco_enter, + .baco_exit = smu_v11_0_baco_exit, .get_dpm_ultimate_freq = smu_v11_0_get_dpm_ultimate_freq, .set_soft_freq_limited_range = smu_v11_0_set_soft_freq_limited_range, .override_pcie_parameters = smu_v11_0_override_pcie_parameters, diff --git a/drivers/gpu/drm/amd/powerplay/navi10_ppt.h b/drivers/gpu/drm/amd/powerplay/navi10_ppt.h index ec03c7992f6d..2abb4ba01db1 100644 --- a/drivers/gpu/drm/amd/powerplay/navi10_ppt.h +++ b/drivers/gpu/drm/amd/powerplay/navi10_ppt.h @@ -27,12 +27,26 @@ #define NAVI10_PEAK_SCLK_XT (1755) #define NAVI10_PEAK_SCLK_XL (1625) +#define NAVI10_UMD_PSTATE_PROFILING_GFXCLK (1300) +#define NAVI10_UMD_PSTATE_PROFILING_SOCCLK (980) +#define NAVI10_UMD_PSTATE_PROFILING_MEMCLK (625) +#define NAVI10_UMD_PSTATE_PROFILING_VCLK (980) +#define NAVI10_UMD_PSTATE_PROFILING_DCLK (850) + #define NAVI14_UMD_PSTATE_PEAK_XT_GFXCLK (1670) #define NAVI14_UMD_PSTATE_PEAK_XTM_GFXCLK (1448) #define NAVI14_UMD_PSTATE_PEAK_XLM_GFXCLK (1181) #define NAVI14_UMD_PSTATE_PEAK_XTX_GFXCLK (1717) #define NAVI14_UMD_PSTATE_PEAK_XL_GFXCLK (1448) +#define NAVI14_UMD_PSTATE_PROFILING_GFXCLK (1200) +#define NAVI14_UMD_PSTATE_PROFILING_SOCCLK (900) +#define NAVI14_UMD_PSTATE_PROFILING_MEMCLK (600) +#define NAVI14_UMD_PSTATE_PROFILING_VCLK (900) +#define NAVI14_UMD_PSTATE_PROFILING_DCLK (800) + +#define NAVI12_UMD_PSTATE_PEAK_GFXCLK (1100) + #define NAVI10_VOLTAGE_SCALE (4) #define smnPCIE_LC_SPEED_CNTL 0x11140290 diff --git a/drivers/gpu/drm/amd/powerplay/renoir_ppt.c b/drivers/gpu/drm/amd/powerplay/renoir_ppt.c index 977bdd962e98..861e6410363b 100644 --- a/drivers/gpu/drm/amd/powerplay/renoir_ppt.c +++ b/drivers/gpu/drm/amd/powerplay/renoir_ppt.c @@ -31,6 +31,9 @@ #include "renoir_ppt.h" +#define CLK_MAP(clk, index) \ + [SMU_##clk] = {1, (index)} + #define MSG_MAP(msg, index) \ [SMU_MSG_##msg] = {1, (index)} @@ -104,6 +107,14 @@ static struct smu_12_0_cmn2aisc_mapping renoir_message_map[SMU_MSG_MAX_COUNT] = MSG_MAP(SetHardMinFclkByFreq, PPSMC_MSG_SetHardMinFclkByFreq), }; +static struct smu_12_0_cmn2aisc_mapping renoir_clk_map[SMU_CLK_COUNT] = { + CLK_MAP(GFXCLK, CLOCK_GFXCLK), + CLK_MAP(SCLK, CLOCK_GFXCLK), + CLK_MAP(SOCCLK, CLOCK_SOCCLK), + CLK_MAP(UCLK, CLOCK_UMCCLK), + CLK_MAP(MCLK, CLOCK_UMCCLK), +}; + static struct smu_12_0_cmn2aisc_mapping renoir_table_map[SMU_TABLE_COUNT] = { TAB_MAP_VALID(WATERMARKS), TAB_MAP_INVALID(CUSTOM_DPM), @@ -125,6 +136,21 @@ static int renoir_get_smu_msg_index(struct smu_context *smc, uint32_t index) return mapping.map_to; } +static int renoir_get_smu_clk_index(struct smu_context *smc, uint32_t index) +{ + struct smu_12_0_cmn2aisc_mapping mapping; + + if (index >= SMU_CLK_COUNT) + return -EINVAL; + + mapping = renoir_clk_map[index]; + if (!(mapping.valid_mapping)) { + return -EINVAL; + } + + return mapping.map_to; +} + static int renoir_get_smu_table_index(struct smu_context *smc, uint32_t index) { struct smu_12_0_cmn2aisc_mapping mapping; @@ -139,6 +165,30 @@ static int renoir_get_smu_table_index(struct smu_context *smc, uint32_t index) return mapping.map_to; } +static int renoir_get_metrics_table(struct smu_context *smu, + SmuMetrics_t *metrics_table) +{ + struct smu_table_context *smu_table= &smu->smu_table; + int ret = 0; + + mutex_lock(&smu->metrics_lock); + if (!smu_table->metrics_time || time_after(jiffies, smu_table->metrics_time + msecs_to_jiffies(100))) { + ret = smu_update_table(smu, SMU_TABLE_SMU_METRICS, 0, + (void *)smu_table->metrics_table, false); + if (ret) { + pr_info("Failed to export SMU metrics table!\n"); + mutex_unlock(&smu->metrics_lock); + return ret; + } + smu_table->metrics_time = jiffies; + } + + memcpy(metrics_table, smu_table->metrics_table, sizeof(SmuMetrics_t)); + mutex_unlock(&smu->metrics_lock); + + return ret; +} + static int renoir_tables_init(struct smu_context *smu, struct smu_table *tables) { struct smu_table_context *smu_table = &smu->smu_table; @@ -154,6 +204,15 @@ static int renoir_tables_init(struct smu_context *smu, struct smu_table *tables) if (!smu_table->clocks_table) return -ENOMEM; + smu_table->metrics_table = kzalloc(sizeof(SmuMetrics_t), GFP_KERNEL); + if (!smu_table->metrics_table) + return -ENOMEM; + smu_table->metrics_time = 0; + + smu_table->watermarks_table = kzalloc(sizeof(Watermarks_t), GFP_KERNEL); + if (!smu_table->watermarks_table) + return -ENOMEM; + return 0; } @@ -187,8 +246,7 @@ static int renoir_print_clk_levels(struct smu_context *smu, memset(&metrics, 0, sizeof(metrics)); - ret = smu_update_table(smu, SMU_TABLE_SMU_METRICS, 0, - (void *)&metrics, false); + ret = renoir_get_metrics_table(smu, &metrics); if (ret) return ret; @@ -301,6 +359,51 @@ static int renoir_dpm_set_uvd_enable(struct smu_context *smu, bool enable) return ret; } +static int renoir_dpm_set_jpeg_enable(struct smu_context *smu, bool enable) +{ + struct smu_power_context *smu_power = &smu->smu_power; + struct smu_power_gate *power_gate = &smu_power->power_gate; + int ret = 0; + + if (enable) { + if (smu_feature_is_enabled(smu, SMU_FEATURE_JPEG_PG_BIT)) { + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_PowerUpJpeg, 0); + if (ret) + return ret; + } + power_gate->jpeg_gated = false; + } else { + if (smu_feature_is_enabled(smu, SMU_FEATURE_JPEG_PG_BIT)) { + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_PowerDownJpeg, 0); + if (ret) + return ret; + } + power_gate->jpeg_gated = true; + } + + return ret; +} + +static int renoir_get_current_clk_freq_by_table(struct smu_context *smu, + enum smu_clk_type clk_type, + uint32_t *value) +{ + int ret = 0, clk_id = 0; + SmuMetrics_t metrics; + + ret = renoir_get_metrics_table(smu, &metrics); + if (ret) + return ret; + + clk_id = smu_clk_get_index(smu, clk_type); + if (clk_id < 0) + return clk_id; + + *value = metrics.ClockFrequency[clk_id]; + + return ret; +} + static int renoir_force_dpm_limit_value(struct smu_context *smu, bool highest) { int ret = 0, i = 0; @@ -361,6 +464,50 @@ static int renoir_unforce_dpm_levels(struct smu_context *smu) { return ret; } +static int renoir_get_gpu_temperature(struct smu_context *smu, uint32_t *value) +{ + int ret = 0; + SmuMetrics_t metrics; + + if (!value) + return -EINVAL; + + ret = renoir_get_metrics_table(smu, &metrics); + if (ret) + return ret; + + *value = (metrics.GfxTemperature / 100) * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + + return 0; +} + +static int renoir_get_current_activity_percent(struct smu_context *smu, + enum amd_pp_sensors sensor, + uint32_t *value) +{ + int ret = 0; + SmuMetrics_t metrics; + + if (!value) + return -EINVAL; + + ret = renoir_get_metrics_table(smu, &metrics); + if (ret) + return ret; + + switch (sensor) { + case AMDGPU_PP_SENSOR_GPU_LOAD: + *value = metrics.AverageGfxActivity / 100; + break; + default: + pr_err("Invalid sensor for retrieving clock activity\n"); + return -EINVAL; + } + + return 0; +} + static int renoir_get_workload_type(struct smu_context *smu, uint32_t profile) { @@ -565,19 +712,43 @@ static int renoir_set_peak_clock_by_device(struct smu_context *smu) return ret; } -static int renoir_set_performance_level(struct smu_context *smu, enum amd_dpm_forced_level level) +static int renoir_set_performance_level(struct smu_context *smu, + enum amd_dpm_forced_level level) { int ret = 0; + uint32_t sclk_mask, mclk_mask, soc_mask; switch (level) { + case AMD_DPM_FORCED_LEVEL_HIGH: + ret = smu_force_dpm_limit_value(smu, true); + break; + case AMD_DPM_FORCED_LEVEL_LOW: + ret = smu_force_dpm_limit_value(smu, false); + break; + case AMD_DPM_FORCED_LEVEL_AUTO: + case AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD: + ret = smu_unforce_dpm_levels(smu); + break; + case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK: + case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK: + ret = smu_get_profiling_clk_mask(smu, level, + &sclk_mask, + &mclk_mask, + &soc_mask); + if (ret) + return ret; + smu_force_clk_levels(smu, SMU_SCLK, 1 << sclk_mask, false); + smu_force_clk_levels(smu, SMU_MCLK, 1 << mclk_mask, false); + smu_force_clk_levels(smu, SMU_SOCCLK, 1 << soc_mask, false); + break; case AMD_DPM_FORCED_LEVEL_PROFILE_PEAK: ret = renoir_set_peak_clock_by_device(smu); break; + case AMD_DPM_FORCED_LEVEL_MANUAL: + case AMD_DPM_FORCED_LEVEL_PROFILE_EXIT: default: - ret = -EINVAL; break; } - return ret; } @@ -636,9 +807,17 @@ static int renoir_set_watermarks_table( } /* pass data to smu controller */ - ret = smu_write_watermarks_table(smu); + if ((smu->watermarks_bitmap & WATERMARKS_EXIST) && + !(smu->watermarks_bitmap & WATERMARKS_LOADED)) { + ret = smu_write_watermarks_table(smu); + if (ret) { + pr_err("Failed to update WMTABLE!"); + return ret; + } + smu->watermarks_bitmap |= WATERMARKS_LOADED; + } - return ret; + return 0; } static int renoir_get_power_profile_mode(struct smu_context *smu, @@ -674,8 +853,36 @@ static int renoir_get_power_profile_mode(struct smu_context *smu, return size; } +static int renoir_read_sensor(struct smu_context *smu, + enum amd_pp_sensors sensor, + void *data, uint32_t *size) +{ + int ret = 0; + + if (!data || !size) + return -EINVAL; + + mutex_lock(&smu->sensor_lock); + switch (sensor) { + case AMDGPU_PP_SENSOR_GPU_LOAD: + ret = renoir_get_current_activity_percent(smu, sensor, (uint32_t *)data); + *size = 4; + break; + case AMDGPU_PP_SENSOR_GPU_TEMP: + ret = renoir_get_gpu_temperature(smu, (uint32_t *)data); + *size = 4; + break; + default: + ret = smu_v12_0_read_sensor(smu, sensor, data, size); + } + mutex_unlock(&smu->sensor_lock); + + return ret; +} + static const struct pptable_funcs renoir_ppt_funcs = { .get_smu_msg_index = renoir_get_smu_msg_index, + .get_smu_clk_index = renoir_get_smu_clk_index, .get_smu_table_index = renoir_get_smu_table_index, .tables_init = renoir_tables_init, .set_power_state = NULL, @@ -683,6 +890,8 @@ static const struct pptable_funcs renoir_ppt_funcs = { .print_clk_levels = renoir_print_clk_levels, .get_current_power_state = renoir_get_current_power_state, .dpm_set_uvd_enable = renoir_dpm_set_uvd_enable, + .dpm_set_jpeg_enable = renoir_dpm_set_jpeg_enable, + .get_current_clk_freq_by_table = renoir_get_current_clk_freq_by_table, .force_dpm_limit_value = renoir_force_dpm_limit_value, .unforce_dpm_levels = renoir_unforce_dpm_levels, .get_workload_type = renoir_get_workload_type, @@ -693,10 +902,12 @@ static const struct pptable_funcs renoir_ppt_funcs = { .get_dpm_clock_table = renoir_get_dpm_clock_table, .set_watermarks_table = renoir_set_watermarks_table, .get_power_profile_mode = renoir_get_power_profile_mode, + .read_sensor = renoir_read_sensor, .check_fw_status = smu_v12_0_check_fw_status, .check_fw_version = smu_v12_0_check_fw_version, .powergate_sdma = smu_v12_0_powergate_sdma, .powergate_vcn = smu_v12_0_powergate_vcn, + .powergate_jpeg = smu_v12_0_powergate_jpeg, .send_smc_msg_with_param = smu_v12_0_send_msg_with_param, .read_smc_arg = smu_v12_0_read_arg, .set_gfx_cgpg = smu_v12_0_set_gfx_cgpg, @@ -704,9 +915,12 @@ static const struct pptable_funcs renoir_ppt_funcs = { .init_smc_tables = smu_v12_0_init_smc_tables, .fini_smc_tables = smu_v12_0_fini_smc_tables, .populate_smc_tables = smu_v12_0_populate_smc_tables, + .get_enabled_mask = smu_v12_0_get_enabled_mask, + .get_current_clk_freq = smu_v12_0_get_current_clk_freq, .get_dpm_ultimate_freq = smu_v12_0_get_dpm_ultimate_freq, .mode2_reset = smu_v12_0_mode2_reset, .set_soft_freq_limited_range = smu_v12_0_set_soft_freq_limited_range, + .set_driver_table_location = smu_v12_0_set_driver_table_location, }; void renoir_set_ppt_funcs(struct smu_context *smu) diff --git a/drivers/gpu/drm/amd/powerplay/smu_internal.h b/drivers/gpu/drm/amd/powerplay/smu_internal.h index 8872f8b2d502..783319ec8bf9 100644 --- a/drivers/gpu/drm/amd/powerplay/smu_internal.h +++ b/drivers/gpu/drm/amd/powerplay/smu_internal.h @@ -42,6 +42,8 @@ ((smu)->ppt_funcs->powergate_sdma ? (smu)->ppt_funcs->powergate_sdma((smu), (gate)) : 0) #define smu_powergate_vcn(smu, gate) \ ((smu)->ppt_funcs->powergate_vcn ? (smu)->ppt_funcs->powergate_vcn((smu), (gate)) : 0) +#define smu_powergate_jpeg(smu, gate) \ + ((smu)->ppt_funcs->powergate_jpeg ? (smu)->ppt_funcs->powergate_jpeg((smu), (gate)) : 0) #define smu_get_vbios_bootup_values(smu) \ ((smu)->ppt_funcs->get_vbios_bootup_values ? (smu)->ppt_funcs->get_vbios_bootup_values((smu)) : 0) @@ -59,6 +61,8 @@ ((smu)->ppt_funcs->write_pptable ? (smu)->ppt_funcs->write_pptable((smu)) : 0) #define smu_set_min_dcef_deep_sleep(smu) \ ((smu)->ppt_funcs->set_min_dcef_deep_sleep ? (smu)->ppt_funcs->set_min_dcef_deep_sleep((smu)) : 0) +#define smu_set_driver_table_location(smu) \ + ((smu)->ppt_funcs->set_driver_table_location ? (smu)->ppt_funcs->set_driver_table_location((smu)) : 0) #define smu_set_tool_table_location(smu) \ ((smu)->ppt_funcs->set_tool_table_location ? (smu)->ppt_funcs->set_tool_table_location((smu)) : 0) #define smu_notify_memory_pool_location(smu) \ @@ -127,8 +131,8 @@ int smu_send_smc_msg(struct smu_context *smu, enum smu_message_type msg); ((smu)->ppt_funcs->display_config_changed ? (smu)->ppt_funcs->display_config_changed((smu)) : 0) #define smu_apply_clocks_adjust_rules(smu) \ ((smu)->ppt_funcs->apply_clocks_adjust_rules ? (smu)->ppt_funcs->apply_clocks_adjust_rules((smu)) : 0) -#define smu_notify_smc_dispaly_config(smu) \ - ((smu)->ppt_funcs->notify_smc_dispaly_config ? (smu)->ppt_funcs->notify_smc_dispaly_config((smu)) : 0) +#define smu_notify_smc_display_config(smu) \ + ((smu)->ppt_funcs->notify_smc_display_config ? (smu)->ppt_funcs->notify_smc_display_config((smu)) : 0) #define smu_force_dpm_limit_value(smu, highest) \ ((smu)->ppt_funcs->force_dpm_limit_value ? (smu)->ppt_funcs->force_dpm_limit_value((smu), (highest)) : 0) #define smu_unforce_dpm_levels(smu) \ @@ -170,6 +174,8 @@ int smu_send_smc_msg(struct smu_context *smu, enum smu_message_type msg); ((smu)->ppt_funcs->dpm_set_uvd_enable ? (smu)->ppt_funcs->dpm_set_uvd_enable((smu), (enable)) : 0) #define smu_dpm_set_vce_enable(smu, enable) \ ((smu)->ppt_funcs->dpm_set_vce_enable ? (smu)->ppt_funcs->dpm_set_vce_enable((smu), (enable)) : 0) +#define smu_dpm_set_jpeg_enable(smu, enable) \ + ((smu)->ppt_funcs->dpm_set_jpeg_enable ? (smu)->ppt_funcs->dpm_set_jpeg_enable((smu), (enable)) : 0) #define smu_set_watermarks_table(smu, tab, clock_ranges) \ ((smu)->ppt_funcs->set_watermarks_table ? (smu)->ppt_funcs->set_watermarks_table((smu), (tab), (clock_ranges)) : 0) diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c index e4268a627eff..02f8c9cb89d9 100644 --- a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c +++ b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c @@ -37,6 +37,7 @@ #include "soc15_common.h" #include "atom.h" #include "amd_pcie.h" +#include "amdgpu_ras.h" #include "asic_reg/thm/thm_11_0_2_offset.h" #include "asic_reg/thm/thm_11_0_2_sh_mask.h" @@ -79,15 +80,13 @@ static int smu_v11_0_wait_for_response(struct smu_context *smu) for (i = 0; i < timeout; i++) { cur_value = RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90); if ((cur_value & MP1_C2PMSG_90__CONTENT_MASK) != 0) - break; + return cur_value == 0x1 ? 0 : -EIO; + udelay(1); } /* timeout means wrong logic */ - if (i == timeout) - return -ETIME; - - return RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90) == 0x1 ? 0 : -EIO; + return -ETIME; } int @@ -103,9 +102,11 @@ smu_v11_0_send_msg_with_param(struct smu_context *smu, return index; ret = smu_v11_0_wait_for_response(smu); - if (ret) - pr_err("failed send message: %10s (%d) \tparam: 0x%08x response %#x\n", - smu_get_message_name(smu, msg), index, param, ret); + if (ret) { + pr_err("Msg issuing pre-check failed and " + "SMU may be not in the right state!\n"); + return ret; + } WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90, 0); @@ -449,8 +450,10 @@ int smu_v11_0_fini_smc_tables(struct smu_context *smu) kfree(smu_table->tables); kfree(smu_table->metrics_table); + kfree(smu_table->watermarks_table); smu_table->tables = NULL; smu_table->metrics_table = NULL; + smu_table->watermarks_table = NULL; smu_table->metrics_time = 0; ret = smu_v11_0_fini_dpm_context(smu); @@ -773,6 +776,24 @@ int smu_v11_0_set_min_dcef_deep_sleep(struct smu_context *smu) return smu_v11_0_set_deep_sleep_dcefclk(smu, table_context->boot_values.dcefclk / 100); } +int smu_v11_0_set_driver_table_location(struct smu_context *smu) +{ + struct smu_table *driver_table = &smu->smu_table.driver_table; + int ret = 0; + + if (driver_table->mc_address) { + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetDriverDramAddrHigh, + upper_32_bits(driver_table->mc_address)); + if (!ret) + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetDriverDramAddrLow, + lower_32_bits(driver_table->mc_address)); + } + + return ret; +} + int smu_v11_0_set_tool_table_location(struct smu_context *smu) { int ret = 0; @@ -834,27 +855,33 @@ int smu_v11_0_get_enabled_mask(struct smu_context *smu, uint32_t *feature_mask, uint32_t num) { uint32_t feature_mask_high = 0, feature_mask_low = 0; + struct smu_feature *feature = &smu->smu_feature; int ret = 0; if (!feature_mask || num < 2) return -EINVAL; - ret = smu_send_smc_msg(smu, SMU_MSG_GetEnabledSmuFeaturesHigh); - if (ret) - return ret; - ret = smu_read_smc_arg(smu, &feature_mask_high); - if (ret) - return ret; + if (bitmap_empty(feature->enabled, feature->feature_num)) { + ret = smu_send_smc_msg(smu, SMU_MSG_GetEnabledSmuFeaturesHigh); + if (ret) + return ret; + ret = smu_read_smc_arg(smu, &feature_mask_high); + if (ret) + return ret; - ret = smu_send_smc_msg(smu, SMU_MSG_GetEnabledSmuFeaturesLow); - if (ret) - return ret; - ret = smu_read_smc_arg(smu, &feature_mask_low); - if (ret) - return ret; + ret = smu_send_smc_msg(smu, SMU_MSG_GetEnabledSmuFeaturesLow); + if (ret) + return ret; + ret = smu_read_smc_arg(smu, &feature_mask_low); + if (ret) + return ret; - feature_mask[0] = feature_mask_low; - feature_mask[1] = feature_mask_high; + feature_mask[0] = feature_mask_low; + feature_mask[1] = feature_mask_high; + } else { + bitmap_copy((unsigned long *)feature_mask, feature->enabled, + feature->feature_num); + } return ret; } @@ -866,21 +893,24 @@ int smu_v11_0_system_features_control(struct smu_context *smu, uint32_t feature_mask[2]; int ret = 0; - if (smu->pm_enabled) { - ret = smu_send_smc_msg(smu, (en ? SMU_MSG_EnableAllSmuFeatures : - SMU_MSG_DisableAllSmuFeatures)); - if (ret) - return ret; - } - - ret = smu_feature_get_enabled_mask(smu, feature_mask, 2); + ret = smu_send_smc_msg(smu, (en ? SMU_MSG_EnableAllSmuFeatures : + SMU_MSG_DisableAllSmuFeatures)); if (ret) return ret; - bitmap_copy(feature->enabled, (unsigned long *)&feature_mask, - feature->feature_num); - bitmap_copy(feature->supported, (unsigned long *)&feature_mask, - feature->feature_num); + if (en) { + ret = smu_feature_get_enabled_mask(smu, feature_mask, 2); + if (ret) + return ret; + + bitmap_copy(feature->enabled, (unsigned long *)&feature_mask, + feature->feature_num); + bitmap_copy(feature->supported, (unsigned long *)&feature_mask, + feature->feature_num); + } else { + bitmap_zero(feature->enabled, feature->feature_num); + bitmap_zero(feature->supported, feature->feature_num); + } return ret; } @@ -1124,11 +1154,12 @@ static int smu_v11_0_set_thermal_range(struct smu_context *smu, int low = SMU_THERMAL_MINIMUM_ALERT_TEMP; int high = SMU_THERMAL_MAXIMUM_ALERT_TEMP; uint32_t val; + struct smu_table_context *table_context = &smu->smu_table; + struct smu_11_0_powerplay_table *powerplay_table = table_context->power_play_table; low = max(SMU_THERMAL_MINIMUM_ALERT_TEMP, range.min / SMU_TEMPERATURE_UNITS_PER_CENTIGRADES); - high = min(SMU_THERMAL_MAXIMUM_ALERT_TEMP, - range.max / SMU_TEMPERATURE_UNITS_PER_CENTIGRADES); + high = min((uint16_t)SMU_THERMAL_MAXIMUM_ALERT_TEMP, powerplay_table->software_shutdown_temp); if (low > high) return -EINVAL; @@ -1617,7 +1648,9 @@ bool smu_v11_0_baco_is_support(struct smu_context *smu) if (!baco_support) return false; - if (!smu_feature_is_enabled(smu, SMU_FEATURE_BACO_BIT)) + /* Arcturus does not support this bit mask */ + if (smu_feature_is_supported(smu, SMU_FEATURE_BACO_BIT) && + !smu_feature_is_enabled(smu, SMU_FEATURE_BACO_BIT)) return false; val = RREG32_SOC15(NBIO, 0, mmRCC_BIF_STRAP0); @@ -1643,6 +1676,10 @@ int smu_v11_0_baco_set_state(struct smu_context *smu, enum smu_baco_state state) { struct smu_baco_context *smu_baco = &smu->smu_baco; + struct amdgpu_device *adev = smu->adev; + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + uint32_t bif_doorbell_intr_cntl; + uint32_t data; int ret = 0; if (smu_v11_0_baco_get_state(smu) == state) @@ -1650,10 +1687,37 @@ int smu_v11_0_baco_set_state(struct smu_context *smu, enum smu_baco_state state) mutex_lock(&smu_baco->mutex); - if (state == SMU_BACO_STATE_ENTER) - ret = smu_send_smc_msg_with_param(smu, SMU_MSG_EnterBaco, BACO_SEQ_BACO); - else + bif_doorbell_intr_cntl = RREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL); + + if (state == SMU_BACO_STATE_ENTER) { + bif_doorbell_intr_cntl = REG_SET_FIELD(bif_doorbell_intr_cntl, + BIF_DOORBELL_INT_CNTL, + DOORBELL_INTERRUPT_DISABLE, 1); + WREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL, bif_doorbell_intr_cntl); + + if (!ras || !ras->supported) { + data = RREG32_SOC15(THM, 0, mmTHM_BACO_CNTL); + data |= 0x80000000; + WREG32_SOC15(THM, 0, mmTHM_BACO_CNTL, data); + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_EnterBaco, 0); + } else { + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_EnterBaco, 1); + } + } else { ret = smu_send_smc_msg(smu, SMU_MSG_ExitBaco); + if (ret) + goto out; + + bif_doorbell_intr_cntl = REG_SET_FIELD(bif_doorbell_intr_cntl, + BIF_DOORBELL_INT_CNTL, + DOORBELL_INTERRUPT_DISABLE, 0); + WREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL, bif_doorbell_intr_cntl); + + /* clear vbios scratch 6 and 7 for coming asic reinit */ + WREG32(adev->bios_scratch_reg_offset + 6, 0); + WREG32(adev->bios_scratch_reg_offset + 7, 0); + } if (ret) goto out; @@ -1663,13 +1727,17 @@ out: return ret; } -int smu_v11_0_baco_reset(struct smu_context *smu) +int smu_v11_0_baco_enter(struct smu_context *smu) { + struct amdgpu_device *adev = smu->adev; int ret = 0; - ret = smu_v11_0_baco_set_armd3_sequence(smu, BACO_SEQ_BACO); - if (ret) - return ret; + /* Arcturus does not need this audio workaround */ + if (adev->asic_type != CHIP_ARCTURUS) { + ret = smu_v11_0_baco_set_armd3_sequence(smu, BACO_SEQ_BACO); + if (ret) + return ret; + } ret = smu_v11_0_baco_set_state(smu, SMU_BACO_STATE_ENTER); if (ret) @@ -1677,6 +1745,13 @@ int smu_v11_0_baco_reset(struct smu_context *smu) msleep(10); + return ret; +} + +int smu_v11_0_baco_exit(struct smu_context *smu) +{ + int ret = 0; + ret = smu_v11_0_baco_set_state(smu, SMU_BACO_STATE_EXIT); if (ret) return ret; @@ -1815,3 +1890,42 @@ int smu_v11_0_set_default_od_settings(struct smu_context *smu, bool initialize, } return ret; } + +int smu_v11_0_set_performance_level(struct smu_context *smu, + enum amd_dpm_forced_level level) +{ + int ret = 0; + uint32_t sclk_mask, mclk_mask, soc_mask; + + switch (level) { + case AMD_DPM_FORCED_LEVEL_HIGH: + ret = smu_force_dpm_limit_value(smu, true); + break; + case AMD_DPM_FORCED_LEVEL_LOW: + ret = smu_force_dpm_limit_value(smu, false); + break; + case AMD_DPM_FORCED_LEVEL_AUTO: + case AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD: + ret = smu_unforce_dpm_levels(smu); + break; + case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK: + case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK: + case AMD_DPM_FORCED_LEVEL_PROFILE_PEAK: + ret = smu_get_profiling_clk_mask(smu, level, + &sclk_mask, + &mclk_mask, + &soc_mask); + if (ret) + return ret; + smu_force_clk_levels(smu, SMU_SCLK, 1 << sclk_mask, false); + smu_force_clk_levels(smu, SMU_MCLK, 1 << mclk_mask, false); + smu_force_clk_levels(smu, SMU_SOCCLK, 1 << soc_mask, false); + break; + case AMD_DPM_FORCED_LEVEL_MANUAL: + case AMD_DPM_FORCED_LEVEL_PROFILE_EXIT: + default: + break; + } + return ret; +} + diff --git a/drivers/gpu/drm/amd/powerplay/smu_v12_0.c b/drivers/gpu/drm/amd/powerplay/smu_v12_0.c index 094cfc46adac..870e6db2907e 100644 --- a/drivers/gpu/drm/amd/powerplay/smu_v12_0.c +++ b/drivers/gpu/drm/amd/powerplay/smu_v12_0.c @@ -66,15 +66,13 @@ int smu_v12_0_wait_for_response(struct smu_context *smu) for (i = 0; i < adev->usec_timeout; i++) { cur_value = RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90); if ((cur_value & MP1_C2PMSG_90__CONTENT_MASK) != 0) - break; + return cur_value == 0x1 ? 0 : -EIO; + udelay(1); } /* timeout means wrong logic */ - if (i == adev->usec_timeout) - return -ETIME; - - return RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90) == 0x1 ? 0 : -EIO; + return -ETIME; } int @@ -90,9 +88,11 @@ smu_v12_0_send_msg_with_param(struct smu_context *smu, return index; ret = smu_v12_0_wait_for_response(smu); - if (ret) - pr_err("Failed to send message 0x%x, response 0x%x, param 0x%x\n", - index, ret, param); + if (ret) { + pr_err("Msg issuing pre-check failed and " + "SMU may be not in the right state!\n"); + return ret; + } WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90, 0); @@ -159,7 +159,7 @@ int smu_v12_0_check_fw_version(struct smu_context *smu) int smu_v12_0_powergate_sdma(struct smu_context *smu, bool gate) { - if (!(smu->adev->flags & AMD_IS_APU)) + if (!smu->is_apu) return 0; if (gate) @@ -170,7 +170,7 @@ int smu_v12_0_powergate_sdma(struct smu_context *smu, bool gate) int smu_v12_0_powergate_vcn(struct smu_context *smu, bool gate) { - if (!(smu->adev->flags & AMD_IS_APU)) + if (!smu->is_apu) return 0; if (gate) @@ -179,6 +179,17 @@ int smu_v12_0_powergate_vcn(struct smu_context *smu, bool gate) return smu_send_smc_msg(smu, SMU_MSG_PowerUpVcn); } +int smu_v12_0_powergate_jpeg(struct smu_context *smu, bool gate) +{ + if (!smu->is_apu) + return 0; + + if (gate) + return smu_send_smc_msg_with_param(smu, SMU_MSG_PowerDownJpeg, 0); + else + return smu_send_smc_msg_with_param(smu, SMU_MSG_PowerUpJpeg, 0); +} + int smu_v12_0_set_gfx_cgpg(struct smu_context *smu, bool enable) { if (!(smu->adev->pg_flags & AMD_PG_SUPPORT_GFX_PG)) @@ -188,6 +199,39 @@ int smu_v12_0_set_gfx_cgpg(struct smu_context *smu, bool enable) SMU_MSG_SetGfxCGPG, enable ? 1 : 0); } +int smu_v12_0_read_sensor(struct smu_context *smu, + enum amd_pp_sensors sensor, + void *data, uint32_t *size) +{ + int ret = 0; + + if(!data || !size) + return -EINVAL; + + switch (sensor) { + case AMDGPU_PP_SENSOR_GFX_MCLK: + ret = smu_get_current_clk_freq(smu, SMU_UCLK, (uint32_t *)data); + *size = 4; + break; + case AMDGPU_PP_SENSOR_GFX_SCLK: + ret = smu_get_current_clk_freq(smu, SMU_GFXCLK, (uint32_t *)data); + *size = 4; + break; + case AMDGPU_PP_SENSOR_MIN_FAN_RPM: + *(uint32_t *)data = 0; + *size = 4; + break; + default: + ret = smu_common_read_sensor(smu, sensor, data, size); + break; + } + + if (ret) + *size = 0; + + return ret; +} + /** * smu_v12_0_get_gfxoff_status - get gfxoff status * @@ -274,16 +318,57 @@ int smu_v12_0_fini_smc_tables(struct smu_context *smu) int smu_v12_0_populate_smc_tables(struct smu_context *smu) { struct smu_table_context *smu_table = &smu->smu_table; - struct smu_table *table = NULL; - table = &smu_table->tables[SMU_TABLE_DPMCLOCKS]; - if (!table) + return smu_update_table(smu, SMU_TABLE_DPMCLOCKS, 0, smu_table->clocks_table, false); +} + +int smu_v12_0_get_enabled_mask(struct smu_context *smu, + uint32_t *feature_mask, uint32_t num) +{ + uint32_t feature_mask_high = 0, feature_mask_low = 0; + int ret = 0; + + if (!feature_mask || num < 2) return -EINVAL; - if (!table->cpu_addr) + ret = smu_send_smc_msg(smu, SMU_MSG_GetEnabledSmuFeaturesHigh); + if (ret) + return ret; + ret = smu_read_smc_arg(smu, &feature_mask_high); + if (ret) + return ret; + + ret = smu_send_smc_msg(smu, SMU_MSG_GetEnabledSmuFeaturesLow); + if (ret) + return ret; + ret = smu_read_smc_arg(smu, &feature_mask_low); + if (ret) + return ret; + + feature_mask[0] = feature_mask_low; + feature_mask[1] = feature_mask_high; + + return ret; +} + +int smu_v12_0_get_current_clk_freq(struct smu_context *smu, + enum smu_clk_type clk_id, + uint32_t *value) +{ + int ret = 0; + uint32_t freq = 0; + + if (clk_id >= SMU_CLK_COUNT || !value) return -EINVAL; - return smu_update_table(smu, SMU_TABLE_DPMCLOCKS, 0, smu_table->clocks_table, false); + ret = smu_get_current_clk_freq_by_table(smu, clk_id, &freq); + if (ret) + return ret; + + freq *= 100; + *value = freq; + + return ret; } int smu_v12_0_get_dpm_ultimate_freq(struct smu_context *smu, enum smu_clk_type clk_type, @@ -421,3 +506,21 @@ int smu_v12_0_set_soft_freq_limited_range(struct smu_context *smu, enum smu_clk_ return ret; } + +int smu_v12_0_set_driver_table_location(struct smu_context *smu) +{ + struct smu_table *driver_table = &smu->smu_table.driver_table; + int ret = 0; + + if (driver_table->mc_address) { + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetDriverDramAddrHigh, + upper_32_bits(driver_table->mc_address)); + if (!ret) + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetDriverDramAddrLow, + lower_32_bits(driver_table->mc_address)); + } + + return ret; +} diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/ci_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/ci_smumgr.c index 15590fd86ef4..868e2d5f6e62 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/ci_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/ci_smumgr.c @@ -653,8 +653,8 @@ static int ci_min_max_v_gnbl_pm_lid_from_bapm_vddc(struct pp_hwmgr *hwmgr) static int ci_populate_bapm_vddc_base_leakage_sidd(struct pp_hwmgr *hwmgr) { struct ci_smumgr *smu_data = (struct ci_smumgr *)(hwmgr->smu_backend); - uint16_t HiSidd = smu_data->power_tune_table.BapmVddCBaseLeakageHiSidd; - uint16_t LoSidd = smu_data->power_tune_table.BapmVddCBaseLeakageLoSidd; + uint16_t HiSidd; + uint16_t LoSidd; struct phm_cac_tdp_table *cac_table = hwmgr->dyn_state.cac_dtp_table; HiSidd = (uint16_t)(cac_table->usHighCACLeakage / 100 * 256); diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/fiji_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/fiji_smumgr.c index da025b1d302d..32ebb383c456 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/fiji_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/fiji_smumgr.c @@ -940,7 +940,7 @@ static int fiji_populate_single_graphic_level(struct pp_hwmgr *hwmgr, { int result; /* PP_Clocks minClocks; */ - uint32_t threshold, mvdd; + uint32_t mvdd; struct smu7_hwmgr *data = (struct smu7_hwmgr *)(hwmgr->backend); struct phm_ppt_v1_information *table_info = (struct phm_ppt_v1_information *)(hwmgr->pptable); @@ -973,8 +973,6 @@ static int fiji_populate_single_graphic_level(struct pp_hwmgr *hwmgr, level->VoltageDownHyst = 0; level->PowerThrottle = 0; - threshold = clock * data->fast_watermark_threshold / 100; - data->display_timing.min_clock_in_sr = hwmgr->display_config->min_core_set_clock_in_sr; if (phm_cap_enabled(hwmgr->platform_descriptor.platformCaps, PHM_PlatformCaps_SclkDeepSleep)) @@ -1501,7 +1499,7 @@ static int fiji_populate_memory_timing_parameters(struct pp_hwmgr *hwmgr, uint32_t dram_timing; uint32_t dram_timing2; uint32_t burstTime; - ULONG state, trrds, trrdl; + ULONG trrds, trrdl; int result; result = atomctrl_set_engine_dram_timings_rv770(hwmgr, @@ -1513,7 +1511,6 @@ static int fiji_populate_memory_timing_parameters(struct pp_hwmgr *hwmgr, dram_timing2 = cgs_read_register(hwmgr->device, mmMC_ARB_DRAM_TIMING2); burstTime = cgs_read_register(hwmgr->device, mmMC_ARB_BURST_TIME); - state = PHM_GET_FIELD(burstTime, MC_ARB_BURST_TIME, STATE0); trrds = PHM_GET_FIELD(burstTime, MC_ARB_BURST_TIME, TRRDS0); trrdl = PHM_GET_FIELD(burstTime, MC_ARB_BURST_TIME, TRRDL0); diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/smu10_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/smu10_smumgr.c index aa0ee2b46135..2319400a3fcb 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/smu10_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/smu10_smumgr.c @@ -137,7 +137,7 @@ static int smu10_copy_table_from_smc(struct pp_hwmgr *hwmgr, priv->smu_tables.entry[table_id].table_id); /* flush hdp cache */ - adev->nbio.funcs->hdp_flush(adev, NULL); + amdgpu_asic_flush_hdp(adev, NULL); memcpy(table, (uint8_t *)priv->smu_tables.entry[table_id].table, priv->smu_tables.entry[table_id].size); @@ -150,6 +150,7 @@ static int smu10_copy_table_to_smc(struct pp_hwmgr *hwmgr, { struct smu10_smumgr *priv = (struct smu10_smumgr *)(hwmgr->smu_backend); + struct amdgpu_device *adev = hwmgr->adev; PP_ASSERT_WITH_CODE(table_id < MAX_SMU_TABLE, "Invalid SMU Table ID!", return -EINVAL;); @@ -161,6 +162,8 @@ static int smu10_copy_table_to_smc(struct pp_hwmgr *hwmgr, memcpy(priv->smu_tables.entry[table_id].table, table, priv->smu_tables.entry[table_id].size); + amdgpu_asic_flush_hdp(adev, NULL); + smu10_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetDriverDramAddrHigh, upper_32_bits(priv->smu_tables.entry[table_id].mc_addr)); diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/smu9_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/smu9_smumgr.c index 742b3dc1f6cb..adfbcbe5d113 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/smu9_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/smu9_smumgr.c @@ -61,15 +61,29 @@ static uint32_t smu9_wait_for_response(struct pp_hwmgr *hwmgr) uint32_t reg; uint32_t ret; - reg = SOC15_REG_OFFSET(MP1, 0, mmMP1_SMN_C2PMSG_90); + /* Due to the L1 policy problem under SRIOV, we have to use + * mmMP1_SMN_C2PMSG_103 as the driver response register + */ + if (hwmgr->pp_one_vf) { + reg = SOC15_REG_OFFSET(MP1, 0, mmMP1_SMN_C2PMSG_103); - ret = phm_wait_for_register_unequal(hwmgr, reg, - 0, MP1_C2PMSG_90__CONTENT_MASK); + ret = phm_wait_for_register_unequal(hwmgr, reg, + 0, MP1_C2PMSG_103__CONTENT_MASK); - if (ret) - pr_err("No response from smu\n"); + if (ret) + pr_err("No response from smu\n"); - return RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90); + return RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_103); + } else { + reg = SOC15_REG_OFFSET(MP1, 0, mmMP1_SMN_C2PMSG_90); + + ret = phm_wait_for_register_unequal(hwmgr, reg, + 0, MP1_C2PMSG_90__CONTENT_MASK); + + if (ret) + pr_err("No response from smu\n"); + return RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90); + } } /* @@ -83,7 +97,11 @@ static int smu9_send_msg_to_smc_without_waiting(struct pp_hwmgr *hwmgr, { struct amdgpu_device *adev = hwmgr->adev; - WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_66, msg); + if (hwmgr->pp_one_vf) { + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_101, msg); + } else { + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_66, msg); + } return 0; } @@ -101,7 +119,10 @@ int smu9_send_msg_to_smc(struct pp_hwmgr *hwmgr, uint16_t msg) smu9_wait_for_response(hwmgr); - WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90, 0); + if (hwmgr->pp_one_vf) + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_103, 0); + else + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90, 0); smu9_send_msg_to_smc_without_waiting(hwmgr, msg); @@ -127,9 +148,17 @@ int smu9_send_msg_to_smc_with_parameter(struct pp_hwmgr *hwmgr, smu9_wait_for_response(hwmgr); - WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90, 0); - - WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_82, parameter); + /* Due to the L1 policy problem under SRIOV, we have to use + * mmMP1_SMN_C2PMSG_101 as the driver message register and + * mmMP1_SMN_C2PMSG_102 as the driver parameter register. + */ + if (hwmgr->pp_one_vf) { + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_103, 0); + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_102, parameter); + } else { + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90, 0); + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_82, parameter); + } smu9_send_msg_to_smc_without_waiting(hwmgr, msg); @@ -144,5 +173,8 @@ uint32_t smu9_get_argument(struct pp_hwmgr *hwmgr) { struct amdgpu_device *adev = hwmgr->adev; - return RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_82); + if (hwmgr->pp_one_vf) + return RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_102); + else + return RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_82); } diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.c index 0f3836fd9666..715564009089 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.c @@ -58,7 +58,7 @@ static int vega10_copy_table_from_smc(struct pp_hwmgr *hwmgr, priv->smu_tables.entry[table_id].table_id); /* flush hdp cache */ - adev->nbio.funcs->hdp_flush(adev, NULL); + amdgpu_asic_flush_hdp(adev, NULL); memcpy(table, priv->smu_tables.entry[table_id].table, priv->smu_tables.entry[table_id].size); @@ -70,6 +70,13 @@ static int vega10_copy_table_to_smc(struct pp_hwmgr *hwmgr, uint8_t *table, int16_t table_id) { struct vega10_smumgr *priv = hwmgr->smu_backend; + struct amdgpu_device *adev = hwmgr->adev; + + /* under sriov, vbios or hypervisor driver + * has already copy table to smc so here only skip it + */ + if (!hwmgr->not_vf) + return 0; PP_ASSERT_WITH_CODE(table_id < MAX_SMU_TABLE, "Invalid SMU Table ID!", return -EINVAL); @@ -81,6 +88,8 @@ static int vega10_copy_table_to_smc(struct pp_hwmgr *hwmgr, memcpy(priv->smu_tables.entry[table_id].table, table, priv->smu_tables.entry[table_id].size); + amdgpu_asic_flush_hdp(adev, NULL); + smu9_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetDriverDramAddrHigh, upper_32_bits(priv->smu_tables.entry[table_id].mc_addr)); @@ -100,6 +109,14 @@ int vega10_enable_smc_features(struct pp_hwmgr *hwmgr, int msg = enable ? PPSMC_MSG_EnableSmuFeatures : PPSMC_MSG_DisableSmuFeatures; + /* VF has no permission to change smu feature due + * to security concern even under pp one vf mode + * it still can't do it. For vega10, the smu in + * vbios will enable the appropriate features. + * */ + if (!hwmgr->not_vf) + return 0; + return smum_send_msg_to_smc_with_parameter(hwmgr, msg, feature_mask); } diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.c index 90c782c132d2..a3915bfcce81 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.c @@ -66,7 +66,7 @@ static int vega12_copy_table_from_smc(struct pp_hwmgr *hwmgr, return -EINVAL); /* flush hdp cache */ - adev->nbio.funcs->hdp_flush(adev, NULL); + amdgpu_asic_flush_hdp(adev, NULL); memcpy(table, priv->smu_tables.entry[table_id].table, priv->smu_tables.entry[table_id].size); @@ -84,6 +84,7 @@ static int vega12_copy_table_to_smc(struct pp_hwmgr *hwmgr, { struct vega12_smumgr *priv = (struct vega12_smumgr *)(hwmgr->smu_backend); + struct amdgpu_device *adev = hwmgr->adev; PP_ASSERT_WITH_CODE(table_id < TABLE_COUNT, "Invalid SMU Table ID!", return -EINVAL); @@ -95,6 +96,8 @@ static int vega12_copy_table_to_smc(struct pp_hwmgr *hwmgr, memcpy(priv->smu_tables.entry[table_id].table, table, priv->smu_tables.entry[table_id].size); + amdgpu_asic_flush_hdp(adev, NULL); + PP_ASSERT_WITH_CODE(smu9_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetDriverDramAddrHigh, upper_32_bits(priv->smu_tables.entry[table_id].mc_addr)) == 0, diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.c index f604612f411f..0db57fb83d30 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.c @@ -189,7 +189,7 @@ static int vega20_copy_table_from_smc(struct pp_hwmgr *hwmgr, return ret); /* flush hdp cache */ - adev->nbio.funcs->hdp_flush(adev, NULL); + amdgpu_asic_flush_hdp(adev, NULL); memcpy(table, priv->smu_tables.entry[table_id].table, priv->smu_tables.entry[table_id].size); @@ -207,6 +207,7 @@ static int vega20_copy_table_to_smc(struct pp_hwmgr *hwmgr, { struct vega20_smumgr *priv = (struct vega20_smumgr *)(hwmgr->smu_backend); + struct amdgpu_device *adev = hwmgr->adev; int ret = 0; PP_ASSERT_WITH_CODE(table_id < TABLE_COUNT, @@ -219,6 +220,8 @@ static int vega20_copy_table_to_smc(struct pp_hwmgr *hwmgr, memcpy(priv->smu_tables.entry[table_id].table, table, priv->smu_tables.entry[table_id].size); + amdgpu_asic_flush_hdp(adev, NULL); + PP_ASSERT_WITH_CODE((ret = vega20_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetDriverDramAddrHigh, upper_32_bits(priv->smu_tables.entry[table_id].mc_addr))) == 0, @@ -242,11 +245,14 @@ int vega20_set_activity_monitor_coeff(struct pp_hwmgr *hwmgr, { struct vega20_smumgr *priv = (struct vega20_smumgr *)(hwmgr->smu_backend); + struct amdgpu_device *adev = hwmgr->adev; int ret = 0; memcpy(priv->smu_tables.entry[TABLE_ACTIVITY_MONITOR_COEFF].table, table, priv->smu_tables.entry[TABLE_ACTIVITY_MONITOR_COEFF].size); + amdgpu_asic_flush_hdp(adev, NULL); + PP_ASSERT_WITH_CODE((ret = vega20_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetDriverDramAddrHigh, upper_32_bits(priv->smu_tables.entry[TABLE_ACTIVITY_MONITOR_COEFF].mc_addr))) == 0, @@ -290,7 +296,7 @@ int vega20_get_activity_monitor_coeff(struct pp_hwmgr *hwmgr, return ret); /* flush hdp cache */ - adev->nbio.funcs->hdp_flush(adev, NULL); + amdgpu_asic_flush_hdp(adev, NULL); memcpy(table, priv->smu_tables.entry[TABLE_ACTIVITY_MONITOR_COEFF].table, priv->smu_tables.entry[TABLE_ACTIVITY_MONITOR_COEFF].size); diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c index ae18fbcb26fb..b0e0d67cd54b 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c @@ -1114,7 +1114,6 @@ static int vegam_populate_smc_acpi_level(struct pp_hwmgr *hwmgr, (struct phm_ppt_v1_information *)(hwmgr->pptable); SMIO_Pattern vol_level; uint32_t mvdd; - uint16_t us_mvdd; table->ACPILevel.Flags &= ~PPSMC_SWSTATE_FLAG_DC; @@ -1168,17 +1167,6 @@ static int vegam_populate_smc_acpi_level(struct pp_hwmgr *hwmgr, "in Clock Dependency Table", ); - us_mvdd = 0; - if ((SMU7_VOLTAGE_CONTROL_NONE == data->mvdd_control) || - (data->mclk_dpm_key_disabled)) - us_mvdd = data->vbios_boot_state.mvdd_bootup_value; - else { - if (!vegam_populate_mvdd_value(hwmgr, - data->dpm_table.mclk_table.dpm_levels[0].value, - &vol_level)) - us_mvdd = vol_level.Voltage; - } - if (!vegam_populate_mvdd_value(hwmgr, 0, &vol_level)) table->MemoryACPILevel.MinMvdd = PP_HOST_TO_SMC_UL(vol_level.Voltage); else @@ -1383,11 +1371,16 @@ static int vegam_populate_smc_boot_level(struct pp_hwmgr *hwmgr, result = phm_find_boot_level(&(data->dpm_table.sclk_table), data->vbios_boot_state.sclk_bootup_value, (uint32_t *)&(table->GraphicsBootLevel)); + if (result) + return result; result = phm_find_boot_level(&(data->dpm_table.mclk_table), data->vbios_boot_state.mclk_bootup_value, (uint32_t *)&(table->MemoryBootLevel)); + if (result) + return result; + table->BootVddc = data->vbios_boot_state.vddc_bootup_value * VOLTAGE_SCALE; table->BootVddci = data->vbios_boot_state.vddci_bootup_value * @@ -1493,7 +1486,7 @@ static int vegam_populate_clock_stretcher_data_table(struct pp_hwmgr *hwmgr) struct vegam_smumgr *smu_data = (struct vegam_smumgr *)(hwmgr->smu_backend); - uint8_t i, stretch_amount, stretch_amount2, volt_offset = 0; + uint8_t i, stretch_amount, volt_offset = 0; struct phm_ppt_v1_information *table_info = (struct phm_ppt_v1_information *)(hwmgr->pptable); struct phm_ppt_v1_clock_voltage_dependency_table *sclk_table = @@ -1532,11 +1525,9 @@ static int vegam_populate_clock_stretcher_data_table(struct pp_hwmgr *hwmgr) (table_info->cac_dtp_table->ucCKS_LDO_REFSEL != 0) ? table_info->cac_dtp_table->ucCKS_LDO_REFSEL : 5; /* Populate CKS Lookup Table */ - if (stretch_amount == 1 || stretch_amount == 2 || stretch_amount == 5) - stretch_amount2 = 0; - else if (stretch_amount == 3 || stretch_amount == 4) - stretch_amount2 = 1; - else { + if (!(stretch_amount == 1 || stretch_amount == 2 || + stretch_amount == 5 || stretch_amount == 3 || + stretch_amount == 4)) { phm_cap_unset(hwmgr->platform_descriptor.platformCaps, PHM_PlatformCaps_ClockStretcher); PP_ASSERT_WITH_CODE(false, diff --git a/drivers/gpu/drm/amd/powerplay/vega20_ppt.c b/drivers/gpu/drm/amd/powerplay/vega20_ppt.c index 0d3a3b0a934e..38febd5ca4da 100644 --- a/drivers/gpu/drm/amd/powerplay/vega20_ppt.c +++ b/drivers/gpu/drm/amd/powerplay/vega20_ppt.c @@ -338,6 +338,10 @@ static int vega20_tables_init(struct smu_context *smu, struct smu_table *tables) return -ENOMEM; smu_table->metrics_time = 0; + smu_table->watermarks_table = kzalloc(sizeof(Watermarks_t), GFP_KERNEL); + if (!smu_table->watermarks_table) + return -ENOMEM; + return 0; } @@ -2235,7 +2239,7 @@ static int vega20_apply_clocks_adjust_rules(struct smu_context *smu) } static int -vega20_notify_smc_dispaly_config(struct smu_context *smu) +vega20_notify_smc_display_config(struct smu_context *smu) { struct vega20_dpm_table *dpm_table = smu->smu_dpm.dpm_context; struct vega20_single_dpm_table *memtable = &dpm_table->mem_table; @@ -3194,6 +3198,7 @@ static const struct pptable_funcs vega20_ppt_funcs = { .get_od_percentage = vega20_get_od_percentage, .get_power_profile_mode = vega20_get_power_profile_mode, .set_power_profile_mode = vega20_set_power_profile_mode, + .set_performance_level = smu_v11_0_set_performance_level, .set_od_percentage = vega20_set_od_percentage, .set_default_od_settings = vega20_set_default_od_settings, .od_edit_dpm_table = vega20_odn_edit_dpm_table, @@ -3203,7 +3208,7 @@ static const struct pptable_funcs vega20_ppt_funcs = { .pre_display_config_changed = vega20_pre_display_config_changed, .display_config_changed = vega20_display_config_changed, .apply_clocks_adjust_rules = vega20_apply_clocks_adjust_rules, - .notify_smc_dispaly_config = vega20_notify_smc_dispaly_config, + .notify_smc_display_config = vega20_notify_smc_display_config, .force_dpm_limit_value = vega20_force_dpm_limit_value, .unforce_dpm_levels = vega20_unforce_dpm_levels, .get_profiling_clk_mask = vega20_get_profiling_clk_mask, @@ -3231,6 +3236,7 @@ static const struct pptable_funcs vega20_ppt_funcs = { .check_fw_version = smu_v11_0_check_fw_version, .write_pptable = smu_v11_0_write_pptable, .set_min_dcef_deep_sleep = smu_v11_0_set_min_dcef_deep_sleep, + .set_driver_table_location = smu_v11_0_set_driver_table_location, .set_tool_table_location = smu_v11_0_set_tool_table_location, .notify_memory_pool_location = smu_v11_0_notify_memory_pool_location, .system_features_control = smu_v11_0_system_features_control, @@ -3259,7 +3265,8 @@ static const struct pptable_funcs vega20_ppt_funcs = { .baco_is_support= smu_v11_0_baco_is_support, .baco_get_state = smu_v11_0_baco_get_state, .baco_set_state = smu_v11_0_baco_set_state, - .baco_reset = smu_v11_0_baco_reset, + .baco_enter = smu_v11_0_baco_enter, + .baco_exit = smu_v11_0_baco_exit, .get_dpm_ultimate_freq = smu_v11_0_get_dpm_ultimate_freq, .set_soft_freq_limited_range = smu_v11_0_set_soft_freq_limited_range, .override_pcie_parameters = smu_v11_0_override_pcie_parameters, |