diff options
Diffstat (limited to 'drivers/gpu/drm/amd/powerplay')
80 files changed, 20280 insertions, 232 deletions
diff --git a/drivers/gpu/drm/amd/powerplay/Makefile b/drivers/gpu/drm/amd/powerplay/Makefile index 231785a9e24c..390345f2d601 100644 --- a/drivers/gpu/drm/amd/powerplay/Makefile +++ b/drivers/gpu/drm/amd/powerplay/Makefile @@ -35,7 +35,7 @@ AMD_POWERPLAY = $(addsuffix /Makefile,$(addprefix $(FULL_AMD_PATH)/powerplay/,$( include $(AMD_POWERPLAY) -POWER_MGR = amd_powerplay.o +POWER_MGR = amd_powerplay.o amdgpu_smu.o smu_v11_0.o smu_v12_0.o vega20_ppt.o arcturus_ppt.o navi10_ppt.o renoir_ppt.o AMD_PP_POWER = $(addprefix $(AMD_PP_PATH)/,$(POWER_MGR)) diff --git a/drivers/gpu/drm/amd/powerplay/amd_powerplay.c b/drivers/gpu/drm/amd/powerplay/amd_powerplay.c index 3f73f7cd18b9..7932eb163a00 100644 --- a/drivers/gpu/drm/amd/powerplay/amd_powerplay.c +++ b/drivers/gpu/drm/amd/powerplay/amd_powerplay.c @@ -53,7 +53,7 @@ static int amd_powerplay_create(struct amdgpu_device *adev) mutex_init(&hwmgr->smu_lock); hwmgr->chip_family = adev->family; hwmgr->chip_id = adev->asic_type; - hwmgr->feature_mask = adev->powerplay.pp_feature; + hwmgr->feature_mask = adev->pm.pp_feature; hwmgr->display_config = &adev->pm.pm_display_cfg; adev->powerplay.pp_handle = hwmgr; adev->powerplay.pp_funcs = &pp_dpm_funcs; @@ -924,6 +924,19 @@ static int pp_odn_edit_dpm_table(void *handle, uint32_t type, long *input, uint3 return hwmgr->hwmgr_func->odn_edit_dpm_table(hwmgr, type, input, size); } +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) + return -EINVAL; + + if (hwmgr->hwmgr_func->set_mp1_state) + return hwmgr->hwmgr_func->set_mp1_state(hwmgr, mp1_state); + + return 0; +} + static int pp_dpm_switch_power_profile(void *handle, enum PP_SMC_POWER_PROFILE type, bool en) { @@ -956,6 +969,14 @@ static int pp_dpm_switch_power_profile(void *handle, workload = hwmgr->workload_setting[index]; } + if (type == PP_SMC_POWER_PROFILE_COMPUTE && + hwmgr->hwmgr_func->disable_power_features_for_compute_performance) { + if (hwmgr->hwmgr_func->disable_power_features_for_compute_performance(hwmgr, en)) { + mutex_unlock(&hwmgr->smu_lock); + return -EINVAL; + } + } + if (hwmgr->dpm_level != AMD_DPM_FORCED_LEVEL_MANUAL) hwmgr->hwmgr_func->set_power_profile_mode(hwmgr, &workload, 0); mutex_unlock(&hwmgr->smu_lock); @@ -1304,7 +1325,7 @@ static int pp_notify_smu_enable_pwe(void *handle) if (hwmgr->hwmgr_func->smus_notify_pwe == NULL) { pr_info_ratelimited("%s was not implemented.\n", __func__); - return -EINVAL;; + return -EINVAL; } mutex_lock(&hwmgr->smu_lock); @@ -1341,7 +1362,7 @@ static int pp_set_min_deep_sleep_dcefclk(void *handle, uint32_t clock) if (hwmgr->hwmgr_func->set_min_deep_sleep_dcefclk == NULL) { pr_debug("%s was not implemented.\n", __func__); - return -EINVAL;; + return -EINVAL; } mutex_lock(&hwmgr->smu_lock); @@ -1360,7 +1381,7 @@ static int pp_set_hard_min_dcefclk_by_freq(void *handle, uint32_t clock) if (hwmgr->hwmgr_func->set_hard_min_dcefclk_by_freq == NULL) { pr_debug("%s was not implemented.\n", __func__); - return -EINVAL;; + return -EINVAL; } mutex_lock(&hwmgr->smu_lock); @@ -1379,7 +1400,7 @@ static int pp_set_hard_min_fclk_by_freq(void *handle, uint32_t clock) if (hwmgr->hwmgr_func->set_hard_min_fclk_by_freq == NULL) { pr_debug("%s was not implemented.\n", __func__); - return -EINVAL;; + return -EINVAL; } mutex_lock(&hwmgr->smu_lock); @@ -1408,6 +1429,7 @@ static int pp_get_asic_baco_capability(void *handle, bool *cap) { struct pp_hwmgr *hwmgr = handle; + *cap = false; if (!hwmgr) return -EINVAL; @@ -1495,6 +1517,80 @@ static int pp_set_ppfeature_status(void *handle, uint64_t ppfeature_masks) return ret; } +static int pp_asic_reset_mode_2(void *handle) +{ + struct pp_hwmgr *hwmgr = handle; + int ret = 0; + + if (!hwmgr || !hwmgr->pm_en) + return -EINVAL; + + if (hwmgr->hwmgr_func->asic_reset == NULL) { + pr_info_ratelimited("%s was not implemented.\n", __func__); + return -EINVAL; + } + + mutex_lock(&hwmgr->smu_lock); + ret = hwmgr->hwmgr_func->asic_reset(hwmgr, SMU_ASIC_RESET_MODE_2); + mutex_unlock(&hwmgr->smu_lock); + + return ret; +} + +static int pp_smu_i2c_bus_access(void *handle, bool acquire) +{ + struct pp_hwmgr *hwmgr = handle; + int ret = 0; + + if (!hwmgr || !hwmgr->pm_en) + return -EINVAL; + + if (hwmgr->hwmgr_func->smu_i2c_bus_access == NULL) { + pr_info_ratelimited("%s was not implemented.\n", __func__); + return -EINVAL; + } + + mutex_lock(&hwmgr->smu_lock); + ret = hwmgr->hwmgr_func->smu_i2c_bus_access(hwmgr, acquire); + mutex_unlock(&hwmgr->smu_lock); + + return ret; +} + +static int pp_set_df_cstate(void *handle, enum pp_df_cstate state) +{ + struct pp_hwmgr *hwmgr = handle; + + if (!hwmgr) + return -EINVAL; + + if (!hwmgr->pm_en || !hwmgr->hwmgr_func->set_df_cstate) + return 0; + + mutex_lock(&hwmgr->smu_lock); + hwmgr->hwmgr_func->set_df_cstate(hwmgr, state); + mutex_unlock(&hwmgr->smu_lock); + + return 0; +} + +static int pp_set_xgmi_pstate(void *handle, uint32_t pstate) +{ + struct pp_hwmgr *hwmgr = handle; + + if (!hwmgr) + return -EINVAL; + + if (!hwmgr->pm_en || !hwmgr->hwmgr_func->set_xgmi_pstate) + return 0; + + mutex_lock(&hwmgr->smu_lock); + hwmgr->hwmgr_func->set_xgmi_pstate(hwmgr, pstate); + mutex_unlock(&hwmgr->smu_lock); + + return 0; +} + static const struct amd_pm_funcs pp_dpm_funcs = { .load_firmware = pp_dpm_load_fw, .wait_for_fw_loading_complete = pp_dpm_fw_loading_complete, @@ -1525,6 +1621,7 @@ static const struct amd_pm_funcs pp_dpm_funcs = { .get_power_profile_mode = pp_get_power_profile_mode, .set_power_profile_mode = pp_set_power_profile_mode, .odn_edit_dpm_table = pp_odn_edit_dpm_table, + .set_mp1_state = pp_dpm_set_mp1_state, .set_power_limit = pp_set_power_limit, .get_power_limit = pp_get_power_limit, /* export to DC */ @@ -1550,4 +1647,8 @@ static const struct amd_pm_funcs pp_dpm_funcs = { .set_asic_baco_state = pp_set_asic_baco_state, .get_ppfeature_status = pp_get_ppfeature_status, .set_ppfeature_status = pp_set_ppfeature_status, + .asic_reset_mode_2 = pp_asic_reset_mode_2, + .smu_i2c_bus_access = pp_smu_i2c_bus_access, + .set_df_cstate = pp_set_df_cstate, + .set_xgmi_pstate = pp_set_xgmi_pstate, }; diff --git a/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c b/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c new file mode 100644 index 000000000000..40b546c75fc2 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c @@ -0,0 +1,2550 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include <linux/firmware.h> + +#include "pp_debug.h" +#include "amdgpu.h" +#include "amdgpu_smu.h" +#include "smu_internal.h" +#include "soc15_common.h" +#include "smu_v11_0.h" +#include "smu_v12_0.h" +#include "atom.h" +#include "amd_pcie.h" +#include "vega20_ppt.h" +#include "arcturus_ppt.h" +#include "navi10_ppt.h" +#include "renoir_ppt.h" + +#undef __SMU_DUMMY_MAP +#define __SMU_DUMMY_MAP(type) #type +static const char* __smu_message_names[] = { + SMU_MESSAGE_TYPES +}; + +const char *smu_get_message_name(struct smu_context *smu, enum smu_message_type type) +{ + if (type < 0 || type >= SMU_MSG_MAX_COUNT) + return "unknown smu message"; + return __smu_message_names[type]; +} + +#undef __SMU_DUMMY_MAP +#define __SMU_DUMMY_MAP(fea) #fea +static const char* __smu_feature_names[] = { + SMU_FEATURE_MASKS +}; + +const char *smu_get_feature_name(struct smu_context *smu, enum smu_feature_mask feature) +{ + if (feature < 0 || feature >= SMU_FEATURE_COUNT) + return "unknown smu feature"; + return __smu_feature_names[feature]; +} + +size_t smu_sys_get_pp_feature_mask(struct smu_context *smu, char *buf) +{ + size_t size = 0; + int ret = 0, i = 0; + uint32_t feature_mask[2] = { 0 }; + int32_t feature_index = 0; + uint32_t count = 0; + uint32_t sort_feature[SMU_FEATURE_COUNT]; + uint64_t hw_feature_count = 0; + + mutex_lock(&smu->mutex); + + ret = smu_feature_get_enabled_mask(smu, feature_mask, 2); + if (ret) + goto failed; + + size = sprintf(buf + size, "features high: 0x%08x low: 0x%08x\n", + feature_mask[1], feature_mask[0]); + + for (i = 0; i < SMU_FEATURE_COUNT; i++) { + feature_index = smu_feature_get_index(smu, i); + if (feature_index < 0) + continue; + sort_feature[feature_index] = i; + hw_feature_count++; + } + + for (i = 0; i < hw_feature_count; i++) { + size += sprintf(buf + size, "%02d. %-20s (%2d) : %s\n", + count++, + smu_get_feature_name(smu, sort_feature[i]), + i, + !!smu_feature_is_enabled(smu, sort_feature[i]) ? + "enabled" : "disabled"); + } + +failed: + mutex_unlock(&smu->mutex); + + return size; +} + +static int smu_feature_update_enable_state(struct smu_context *smu, + uint64_t feature_mask, + bool enabled) +{ + struct smu_feature *feature = &smu->smu_feature; + uint32_t feature_low = 0, feature_high = 0; + int ret = 0; + + if (!smu->pm_enabled) + return ret; + + feature_low = (feature_mask >> 0 ) & 0xffffffff; + feature_high = (feature_mask >> 32) & 0xffffffff; + + if (enabled) { + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_EnableSmuFeaturesLow, + feature_low); + if (ret) + return ret; + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_EnableSmuFeaturesHigh, + feature_high); + if (ret) + return ret; + } else { + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_DisableSmuFeaturesLow, + feature_low); + if (ret) + return ret; + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_DisableSmuFeaturesHigh, + feature_high); + if (ret) + return ret; + } + + mutex_lock(&feature->mutex); + if (enabled) + bitmap_or(feature->enabled, feature->enabled, + (unsigned long *)(&feature_mask), SMU_FEATURE_MAX); + else + bitmap_andnot(feature->enabled, feature->enabled, + (unsigned long *)(&feature_mask), SMU_FEATURE_MAX); + mutex_unlock(&feature->mutex); + + return ret; +} + +int smu_sys_set_pp_feature_mask(struct smu_context *smu, uint64_t new_mask) +{ + int ret = 0; + uint32_t feature_mask[2] = { 0 }; + uint64_t feature_2_enabled = 0; + uint64_t feature_2_disabled = 0; + uint64_t feature_enables = 0; + + mutex_lock(&smu->mutex); + + ret = smu_feature_get_enabled_mask(smu, feature_mask, 2); + if (ret) + goto out; + + feature_enables = ((uint64_t)feature_mask[1] << 32 | (uint64_t)feature_mask[0]); + + feature_2_enabled = ~feature_enables & new_mask; + feature_2_disabled = feature_enables & ~new_mask; + + if (feature_2_enabled) { + ret = smu_feature_update_enable_state(smu, feature_2_enabled, true); + if (ret) + goto out; + } + if (feature_2_disabled) { + ret = smu_feature_update_enable_state(smu, feature_2_disabled, false); + if (ret) + goto out; + } + +out: + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_get_smc_version(struct smu_context *smu, uint32_t *if_version, uint32_t *smu_version) +{ + int ret = 0; + + if (!if_version && !smu_version) + return -EINVAL; + + if (if_version) { + ret = smu_send_smc_msg(smu, SMU_MSG_GetDriverIfVersion); + if (ret) + return ret; + + ret = smu_read_smc_arg(smu, if_version); + if (ret) + return ret; + } + + if (smu_version) { + ret = smu_send_smc_msg(smu, SMU_MSG_GetSmuVersion); + if (ret) + return ret; + + ret = smu_read_smc_arg(smu, smu_version); + if (ret) + return ret; + } + + return ret; +} + +int smu_set_soft_freq_range(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t min, uint32_t max) +{ + int ret = 0; + + if (min <= 0 && max <= 0) + return -EINVAL; + + if (!smu_clk_dpm_is_enabled(smu, clk_type)) + return 0; + + ret = smu_set_soft_freq_limited_range(smu, clk_type, min, max); + return ret; +} + +int smu_set_hard_freq_range(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t min, uint32_t max) +{ + int ret = 0, clk_id = 0; + uint32_t param; + + if (min <= 0 && max <= 0) + return -EINVAL; + + if (!smu_clk_dpm_is_enabled(smu, clk_type)) + return 0; + + clk_id = smu_clk_get_index(smu, clk_type); + if (clk_id < 0) + return clk_id; + + if (max > 0) { + param = (uint32_t)((clk_id << 16) | (max & 0xffff)); + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetHardMaxByFreq, + param); + if (ret) + return ret; + } + + if (min > 0) { + param = (uint32_t)((clk_id << 16) | (min & 0xffff)); + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetHardMinByFreq, + param); + if (ret) + return ret; + } + + + return ret; +} + +int smu_get_dpm_freq_range(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t *min, uint32_t *max, bool lock_needed) +{ + uint32_t clock_limit; + int ret = 0; + + if (!min && !max) + return -EINVAL; + + if (lock_needed) + mutex_lock(&smu->mutex); + + if (!smu_clk_dpm_is_enabled(smu, clk_type)) { + switch (clk_type) { + case SMU_MCLK: + case SMU_UCLK: + clock_limit = smu->smu_table.boot_values.uclk; + break; + case SMU_GFXCLK: + case SMU_SCLK: + clock_limit = smu->smu_table.boot_values.gfxclk; + break; + case SMU_SOCCLK: + clock_limit = smu->smu_table.boot_values.socclk; + break; + default: + clock_limit = 0; + break; + } + + /* clock in Mhz unit */ + if (min) + *min = clock_limit / 100; + if (max) + *max = clock_limit / 100; + } else { + /* + * Todo: Use each asic(ASIC_ppt funcs) control the callbacks exposed to the + * core driver and then have helpers for stuff that is common(SMU_v11_x | SMU_v12_x funcs). + */ + ret = smu_get_dpm_ultimate_freq(smu, clk_type, min, max); + } + + if (lock_needed) + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_get_dpm_freq_by_index(struct smu_context *smu, enum smu_clk_type clk_type, + uint16_t level, uint32_t *value) +{ + int ret = 0, clk_id = 0; + uint32_t param; + + if (!value) + return -EINVAL; + + if (!smu_clk_dpm_is_enabled(smu, clk_type)) + return 0; + + clk_id = smu_clk_get_index(smu, clk_type); + if (clk_id < 0) + return clk_id; + + param = (uint32_t)(((clk_id & 0xffff) << 16) | (level & 0xffff)); + + ret = smu_send_smc_msg_with_param(smu,SMU_MSG_GetDpmFreqByIndex, + param); + if (ret) + return ret; + + ret = smu_read_smc_arg(smu, ¶m); + if (ret) + return ret; + + /* BIT31: 0 - Fine grained DPM, 1 - Dicrete DPM + * now, we un-support it */ + *value = param & 0x7fffffff; + + return ret; +} + +int smu_get_dpm_level_count(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t *value) +{ + return smu_get_dpm_freq_by_index(smu, clk_type, 0xff, value); +} + +bool smu_clk_dpm_is_enabled(struct smu_context *smu, enum smu_clk_type clk_type) +{ + enum smu_feature_mask feature_id = 0; + + switch (clk_type) { + case SMU_MCLK: + case SMU_UCLK: + feature_id = SMU_FEATURE_DPM_UCLK_BIT; + break; + case SMU_GFXCLK: + case SMU_SCLK: + feature_id = SMU_FEATURE_DPM_GFXCLK_BIT; + break; + case SMU_SOCCLK: + feature_id = SMU_FEATURE_DPM_SOCCLK_BIT; + break; + default: + return true; + } + + if(!smu_feature_is_enabled(smu, feature_id)) { + return false; + } + + return true; +} + +/** + * smu_dpm_set_power_gate - power gate/ungate the specific IP block + * + * @smu: smu_context pointer + * @block_type: the IP block to power gate/ungate + * @gate: to power gate if true, ungate otherwise + * + * This API uses no smu->mutex lock protection due to: + * 1. It is either called by other IP block(gfx/sdma/vcn/uvd/vce). + * This is guarded to be race condition free by the caller. + * 2. Or get called on user setting request of power_dpm_force_performance_level. + * Under this case, the smu->mutex lock protection is already enforced on + * the parent API smu_force_performance_level of the call path. + */ +int smu_dpm_set_power_gate(struct smu_context *smu, uint32_t block_type, + bool gate) +{ + int ret = 0; + + switch (block_type) { + case AMD_IP_BLOCK_TYPE_UVD: + ret = smu_dpm_set_uvd_enable(smu, gate); + break; + case AMD_IP_BLOCK_TYPE_VCE: + ret = smu_dpm_set_vce_enable(smu, gate); + break; + case AMD_IP_BLOCK_TYPE_GFX: + ret = smu_gfx_off_control(smu, gate); + break; + case AMD_IP_BLOCK_TYPE_SDMA: + ret = smu_powergate_sdma(smu, gate); + break; + default: + break; + } + + return ret; +} + +int smu_get_power_num_states(struct smu_context *smu, + struct pp_states_info *state_info) +{ + if (!state_info) + return -EINVAL; + + /* not support power state */ + memset(state_info, 0, sizeof(struct pp_states_info)); + state_info->nums = 1; + state_info->states[0] = POWER_STATE_TYPE_DEFAULT; + + return 0; +} + +int smu_common_read_sensor(struct smu_context *smu, enum amd_pp_sensors sensor, + void *data, uint32_t *size) +{ + struct smu_power_context *smu_power = &smu->smu_power; + struct smu_power_gate *power_gate = &smu_power->power_gate; + int ret = 0; + + if(!data || !size) + return -EINVAL; + + switch (sensor) { + case AMDGPU_PP_SENSOR_STABLE_PSTATE_SCLK: + *((uint32_t *)data) = smu->pstate_sclk; + *size = 4; + break; + case AMDGPU_PP_SENSOR_STABLE_PSTATE_MCLK: + *((uint32_t *)data) = smu->pstate_mclk; + *size = 4; + break; + case AMDGPU_PP_SENSOR_ENABLED_SMC_FEATURES_MASK: + ret = smu_feature_get_enabled_mask(smu, (uint32_t *)data, 2); + *size = 8; + break; + case AMDGPU_PP_SENSOR_UVD_POWER: + *(uint32_t *)data = smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UVD_BIT) ? 1 : 0; + *size = 4; + break; + case AMDGPU_PP_SENSOR_VCE_POWER: + *(uint32_t *)data = smu_feature_is_enabled(smu, SMU_FEATURE_DPM_VCE_BIT) ? 1 : 0; + *size = 4; + break; + case AMDGPU_PP_SENSOR_VCN_POWER_STATE: + *(uint32_t *)data = power_gate->vcn_gated ? 0 : 1; + *size = 4; + break; + default: + ret = -EINVAL; + break; + } + + if (ret) + *size = 0; + + return ret; +} + +int smu_update_table(struct smu_context *smu, enum smu_table_id table_index, int argument, + void *table_data, bool drv2smu) +{ + struct smu_table_context *smu_table = &smu->smu_table; + struct amdgpu_device *adev = smu->adev; + struct smu_table *table = NULL; + int ret = 0; + int table_id = smu_table_get_index(smu, table_index); + + if (!table_data || table_id >= SMU_TABLE_COUNT || table_id < 0) + return -EINVAL; + + table = &smu_table->tables[table_index]; + + if (drv2smu) + memcpy(table->cpu_addr, table_data, table->size); + + 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, + table_id | ((argument & 0xFFFF) << 16)); + if (ret) + return ret; + + /* flush hdp cache */ + adev->nbio.funcs->hdp_flush(adev, NULL); + + if (!drv2smu) + memcpy(table_data, table->cpu_addr, table->size); + + return ret; +} + +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 + return false; +} + +bool is_support_sw_smu_xgmi(struct amdgpu_device *adev) +{ + if (!is_support_sw_smu(adev)) + return false; + + if (adev->asic_type == CHIP_VEGA20) + return true; + + return false; +} + +int smu_sys_get_pp_table(struct smu_context *smu, void **table) +{ + struct smu_table_context *smu_table = &smu->smu_table; + uint32_t powerplay_table_size; + + if (!smu_table->power_play_table && !smu_table->hardcode_pptable) + return -EINVAL; + + mutex_lock(&smu->mutex); + + if (smu_table->hardcode_pptable) + *table = smu_table->hardcode_pptable; + else + *table = smu_table->power_play_table; + + powerplay_table_size = smu_table->power_play_table_size; + + mutex_unlock(&smu->mutex); + + return powerplay_table_size; +} + +int smu_sys_set_pp_table(struct smu_context *smu, void *buf, size_t size) +{ + struct smu_table_context *smu_table = &smu->smu_table; + ATOM_COMMON_TABLE_HEADER *header = (ATOM_COMMON_TABLE_HEADER *)buf; + int ret = 0; + + if (!smu->pm_enabled) + return -EINVAL; + if (header->usStructureSize != size) { + pr_err("pp table size not matched !\n"); + return -EIO; + } + + mutex_lock(&smu->mutex); + if (!smu_table->hardcode_pptable) + smu_table->hardcode_pptable = kzalloc(size, GFP_KERNEL); + if (!smu_table->hardcode_pptable) { + ret = -ENOMEM; + goto failed; + } + + memcpy(smu_table->hardcode_pptable, buf, size); + smu_table->power_play_table = smu_table->hardcode_pptable; + smu_table->power_play_table_size = size; + + /* + * Special hw_fini action(for Navi1x, the DPMs disablement will be + * skipped) may be needed for custom pptable uploading. + */ + smu->uploading_custom_pp_table = true; + + ret = smu_reset(smu); + if (ret) + pr_info("smu reset failed, ret = %d\n", ret); + + smu->uploading_custom_pp_table = false; + +failed: + mutex_unlock(&smu->mutex); + return ret; +} + +int smu_feature_init_dpm(struct smu_context *smu) +{ + struct smu_feature *feature = &smu->smu_feature; + int ret = 0; + uint32_t allowed_feature_mask[SMU_FEATURE_MAX/32]; + + if (!smu->pm_enabled) + return ret; + mutex_lock(&feature->mutex); + bitmap_zero(feature->allowed, SMU_FEATURE_MAX); + mutex_unlock(&feature->mutex); + + ret = smu_get_allowed_feature_mask(smu, allowed_feature_mask, + SMU_FEATURE_MAX/32); + if (ret) + return ret; + + mutex_lock(&feature->mutex); + bitmap_or(feature->allowed, feature->allowed, + (unsigned long *)allowed_feature_mask, + feature->feature_num); + mutex_unlock(&feature->mutex); + + return ret; +} + + +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) + return 1; + + feature_id = smu_feature_get_index(smu, mask); + if (feature_id < 0) + return 0; + + WARN_ON(feature_id > feature->feature_num); + + mutex_lock(&feature->mutex); + ret = test_bit(feature_id, feature->enabled); + mutex_unlock(&feature->mutex); + + return ret; +} + +int smu_feature_set_enabled(struct smu_context *smu, enum smu_feature_mask mask, + bool enable) +{ + struct smu_feature *feature = &smu->smu_feature; + int feature_id; + + feature_id = smu_feature_get_index(smu, mask); + if (feature_id < 0) + return -EINVAL; + + WARN_ON(feature_id > feature->feature_num); + + return smu_feature_update_enable_state(smu, + 1ULL << feature_id, + enable); +} + +int smu_feature_is_supported(struct smu_context *smu, enum smu_feature_mask mask) +{ + struct smu_feature *feature = &smu->smu_feature; + int feature_id; + int ret = 0; + + feature_id = smu_feature_get_index(smu, mask); + if (feature_id < 0) + return 0; + + WARN_ON(feature_id > feature->feature_num); + + mutex_lock(&feature->mutex); + ret = test_bit(feature_id, feature->supported); + mutex_unlock(&feature->mutex); + + return ret; +} + +int smu_feature_set_supported(struct smu_context *smu, + enum smu_feature_mask mask, + bool enable) +{ + struct smu_feature *feature = &smu->smu_feature; + int feature_id; + int ret = 0; + + feature_id = smu_feature_get_index(smu, mask); + if (feature_id < 0) + return -EINVAL; + + WARN_ON(feature_id > feature->feature_num); + + mutex_lock(&feature->mutex); + if (enable) + test_and_set_bit(feature_id, feature->supported); + else + test_and_clear_bit(feature_id, feature->supported); + mutex_unlock(&feature->mutex); + + return ret; +} + +static int smu_set_funcs(struct amdgpu_device *adev) +{ + struct smu_context *smu = &adev->smu; + + if (adev->pm.pp_feature & PP_OVERDRIVE_MASK) + smu->od_enabled = true; + + switch (adev->asic_type) { + case CHIP_VEGA20: + adev->pm.pp_feature &= ~PP_GFXOFF_MASK; + vega20_set_ppt_funcs(smu); + break; + case CHIP_NAVI10: + case CHIP_NAVI14: + case CHIP_NAVI12: + navi10_set_ppt_funcs(smu); + break; + case CHIP_ARCTURUS: + adev->pm.pp_feature &= ~PP_GFXOFF_MASK; + arcturus_set_ppt_funcs(smu); + /* OD is not supported on Arcturus */ + smu->od_enabled =false; + break; + case CHIP_RENOIR: + renoir_set_ppt_funcs(smu); + break; + default: + return -EINVAL; + } + + return 0; +} + +static int smu_early_init(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct smu_context *smu = &adev->smu; + + smu->adev = adev; + smu->pm_enabled = !!amdgpu_dpm; + smu->is_apu = false; + mutex_init(&smu->mutex); + + return smu_set_funcs(adev); +} + +static int smu_late_init(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct smu_context *smu = &adev->smu; + + if (!smu->pm_enabled) + return 0; + + smu_handle_task(&adev->smu, + smu->smu_dpm.dpm_level, + AMD_PP_TASK_COMPLETE_INIT, + false); + + return 0; +} + +int smu_get_atom_data_table(struct smu_context *smu, uint32_t table, + uint16_t *size, uint8_t *frev, uint8_t *crev, + uint8_t **addr) +{ + struct amdgpu_device *adev = smu->adev; + uint16_t data_start; + + if (!amdgpu_atom_parse_data_header(adev->mode_info.atom_context, table, + size, frev, crev, &data_start)) + return -EINVAL; + + *addr = (uint8_t *)adev->mode_info.atom_context->bios + data_start; + + return 0; +} + +static int smu_initialize_pptable(struct smu_context *smu) +{ + /* TODO */ + return 0; +} + +static int smu_smc_table_sw_init(struct smu_context *smu) +{ + int ret; + + ret = smu_initialize_pptable(smu); + if (ret) { + pr_err("Failed to init smu_initialize_pptable!\n"); + return ret; + } + + /** + * Create smu_table structure, and init smc tables such as + * TABLE_PPTABLE, TABLE_WATERMARKS, TABLE_SMU_METRICS, and etc. + */ + ret = smu_init_smc_tables(smu); + if (ret) { + pr_err("Failed to init smc tables!\n"); + return ret; + } + + /** + * Create smu_power_context structure, and allocate smu_dpm_context and + * context size to fill the smu_power_context data. + */ + ret = smu_init_power(smu); + if (ret) { + pr_err("Failed to init smu_init_power!\n"); + return ret; + } + + return 0; +} + +static int smu_smc_table_sw_fini(struct smu_context *smu) +{ + int ret; + + ret = smu_fini_smc_tables(smu); + if (ret) { + pr_err("Failed to smu_fini_smc_tables!\n"); + return ret; + } + + return 0; +} + +static int smu_sw_init(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct smu_context *smu = &adev->smu; + int ret; + + smu->pool_size = adev->pm.smu_prv_buffer_size; + smu->smu_feature.feature_num = SMU_FEATURE_MAX; + mutex_init(&smu->smu_feature.mutex); + bitmap_zero(smu->smu_feature.supported, SMU_FEATURE_MAX); + bitmap_zero(smu->smu_feature.enabled, SMU_FEATURE_MAX); + bitmap_zero(smu->smu_feature.allowed, SMU_FEATURE_MAX); + + mutex_init(&smu->smu_baco.mutex); + smu->smu_baco.state = SMU_BACO_STATE_EXIT; + smu->smu_baco.platform_support = false; + + mutex_init(&smu->sensor_lock); + + smu->watermarks_bitmap = 0; + smu->power_profile_mode = PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT; + smu->default_power_profile_mode = PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT; + + smu->workload_mask = 1 << smu->workload_prority[PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT]; + smu->workload_prority[PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT] = 0; + smu->workload_prority[PP_SMC_POWER_PROFILE_FULLSCREEN3D] = 1; + smu->workload_prority[PP_SMC_POWER_PROFILE_POWERSAVING] = 2; + smu->workload_prority[PP_SMC_POWER_PROFILE_VIDEO] = 3; + smu->workload_prority[PP_SMC_POWER_PROFILE_VR] = 4; + smu->workload_prority[PP_SMC_POWER_PROFILE_COMPUTE] = 5; + smu->workload_prority[PP_SMC_POWER_PROFILE_CUSTOM] = 6; + + smu->workload_setting[0] = PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT; + smu->workload_setting[1] = PP_SMC_POWER_PROFILE_FULLSCREEN3D; + smu->workload_setting[2] = PP_SMC_POWER_PROFILE_POWERSAVING; + smu->workload_setting[3] = PP_SMC_POWER_PROFILE_VIDEO; + smu->workload_setting[4] = PP_SMC_POWER_PROFILE_VR; + smu->workload_setting[5] = PP_SMC_POWER_PROFILE_COMPUTE; + smu->workload_setting[6] = PP_SMC_POWER_PROFILE_CUSTOM; + smu->display_config = &adev->pm.pm_display_cfg; + + smu->smu_dpm.dpm_level = AMD_DPM_FORCED_LEVEL_AUTO; + smu->smu_dpm.requested_dpm_level = AMD_DPM_FORCED_LEVEL_AUTO; + ret = smu_init_microcode(smu); + if (ret) { + pr_err("Failed to load smu firmware!\n"); + return ret; + } + + ret = smu_smc_table_sw_init(smu); + if (ret) { + pr_err("Failed to sw init smc table!\n"); + return ret; + } + + ret = smu_register_irq_handler(smu); + if (ret) { + pr_err("Failed to register smc irq handler!\n"); + return ret; + } + + return 0; +} + +static int smu_sw_fini(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct smu_context *smu = &adev->smu; + int ret; + + kfree(smu->irq_source); + smu->irq_source = NULL; + + ret = smu_smc_table_sw_fini(smu); + if (ret) { + pr_err("Failed to sw fini smc table!\n"); + return ret; + } + + ret = smu_fini_power(smu); + if (ret) { + pr_err("Failed to init smu_fini_power!\n"); + return ret; + } + + return 0; +} + +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; + int ret, i; + + for (i = 0; i < SMU_TABLE_COUNT; i++) { + if (tables[i].size == 0) + continue; + 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; + } + + return 0; +failed: + while (--i >= 0) { + if (tables[i].size == 0) + continue; + amdgpu_bo_free_kernel(&tables[i].bo, + &tables[i].mc_address, + &tables[i].cpu_addr); + + } + return ret; +} + +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; + + 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); + } + + return 0; +} + +static int smu_smc_table_hw_init(struct smu_context *smu, + bool initialize) +{ + struct amdgpu_device *adev = smu->adev; + int ret; + + if (smu_is_dpm_running(smu) && adev->in_suspend) { + pr_info("dpm has been enabled\n"); + return 0; + } + + if (adev->asic_type != CHIP_ARCTURUS) { + ret = smu_init_display_count(smu, 0); + if (ret) + return ret; + } + + if (initialize) { + /* get boot_values from vbios to set revision, gfxclk, and etc. */ + ret = smu_get_vbios_bootup_values(smu); + if (ret) + return ret; + + ret = smu_setup_pptable(smu); + if (ret) + return ret; + + ret = smu_get_clk_info_from_vbios(smu); + if (ret) + return ret; + + /* + * check if the format_revision in vbios is up to pptable header + * version, and the structure size is not 0. + */ + ret = smu_check_pptable(smu); + if (ret) + return ret; + + /* + * allocate vram bos to store smc table contents. + */ + ret = smu_init_fb_allocations(smu); + if (ret) + return ret; + + /* + * Parse pptable format and fill PPTable_t smc_pptable to + * smu_table_context structure. And read the smc_dpm_table from vbios, + * then fill it into smc_pptable. + */ + ret = smu_parse_pptable(smu); + if (ret) + return ret; + + /* + * Send msg GetDriverIfVersion to check if the return value is equal + * with DRIVER_IF_VERSION of smc header. + */ + ret = smu_check_fw_version(smu); + if (ret) + return ret; + } + + /* smu_dump_pptable(smu); */ + + /* + * 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; + + 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) + return ret; + + /* + * Set min deep sleep dce fclk with bootup value from vbios via + * SetMinDeepSleepDcefclk MSG. + */ + ret = smu_set_min_dcef_deep_sleep(smu); + if (ret) + return ret; + } + + /* + * Set initialized values (get from vbios) to dpm tables context such as + * gfxclk, memclk, dcefclk, and etc. And enable the DPM feature for each + * type of clks. + */ + if (initialize) { + ret = smu_populate_smc_tables(smu); + if (ret) + return ret; + + ret = smu_init_max_sustainable_clocks(smu); + if (ret) + return ret; + } + + if (adev->asic_type != CHIP_ARCTURUS) { + ret = smu_override_pcie_parameters(smu); + if (ret) + return ret; + } + + ret = smu_set_default_od_settings(smu, initialize); + if (ret) + return ret; + + if (initialize) { + ret = smu_populate_umd_state_clk(smu); + if (ret) + return ret; + + ret = smu_get_power_limit(smu, &smu->default_power_limit, false, false); + if (ret) + return ret; + } + + /* + * Set PMSTATUSLOG table bo address with SetToolsDramAddr MSG for tools. + */ + ret = smu_set_tool_table_location(smu); + + if (!smu_is_dpm_running(smu)) + pr_info("dpm has been disabled\n"); + + return ret; +} + +/** + * smu_alloc_memory_pool - allocate memory pool in the system memory + * + * @smu: amdgpu_device pointer + * + * This memory pool will be used for SMC use and msg SetSystemVirtualDramAddr + * and DramLogSetDramAddr can notify it changed. + * + * Returns 0 on success, error on failure. + */ +static int smu_alloc_memory_pool(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + struct smu_table_context *smu_table = &smu->smu_table; + struct smu_table *memory_pool = &smu_table->memory_pool; + uint64_t pool_size = smu->pool_size; + int ret = 0; + + if (pool_size == SMU_MEMORY_POOL_SIZE_ZERO) + return ret; + + memory_pool->size = pool_size; + memory_pool->align = PAGE_SIZE; + memory_pool->domain = AMDGPU_GEM_DOMAIN_GTT; + + switch (pool_size) { + case SMU_MEMORY_POOL_SIZE_256_MB: + case SMU_MEMORY_POOL_SIZE_512_MB: + case SMU_MEMORY_POOL_SIZE_1_GB: + case SMU_MEMORY_POOL_SIZE_2_GB: + ret = amdgpu_bo_create_kernel(adev, + memory_pool->size, + memory_pool->align, + memory_pool->domain, + &memory_pool->bo, + &memory_pool->mc_address, + &memory_pool->cpu_addr); + break; + default: + break; + } + + return ret; +} + +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; + + amdgpu_bo_free_kernel(&memory_pool->bo, + &memory_pool->mc_address, + &memory_pool->cpu_addr); + + memset(memory_pool, 0, sizeof(struct smu_table)); + + return ret; +} + +static int smu_start_smc_engine(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + int ret = 0; + + if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) { + if (adev->asic_type < CHIP_NAVI10) { + if (smu->ppt_funcs->load_microcode) { + ret = smu->ppt_funcs->load_microcode(smu); + if (ret) + return ret; + } + } + } + + if (smu->ppt_funcs->check_fw_status) { + ret = smu->ppt_funcs->check_fw_status(smu); + if (ret) + pr_err("SMC is not ready\n"); + } + + return ret; +} + +static int smu_hw_init(void *handle) +{ + int ret; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct smu_context *smu = &adev->smu; + + ret = smu_start_smc_engine(smu); + if (ret) { + pr_err("SMU is not ready yet!\n"); + return ret; + } + + if (adev->flags & AMD_IS_APU) { + smu_powergate_sdma(&adev->smu, false); + smu_powergate_vcn(&adev->smu, false); + smu_set_gfx_cgpg(&adev->smu, true); + } + + if (!smu->pm_enabled) + return 0; + + ret = smu_feature_init_dpm(smu); + if (ret) + goto failed; + + ret = smu_smc_table_hw_init(smu, true); + if (ret) + goto failed; + + ret = smu_alloc_memory_pool(smu); + if (ret) + goto failed; + + /* + * Use msg SetSystemVirtualDramAddr and DramLogSetDramAddr can notify + * pool location. + */ + ret = smu_notify_memory_pool_location(smu); + if (ret) + goto failed; + + ret = smu_start_thermal_control(smu); + if (ret) + goto failed; + + if (!smu->pm_enabled) + adev->pm.dpm_enabled = false; + else + adev->pm.dpm_enabled = true; /* TODO: will set dpm_enabled flag while VCN and DAL DPM is workable */ + + pr_info("SMU is initialized successfully!\n"); + + return 0; + +failed: + return ret; +} + +static int smu_stop_dpms(struct smu_context *smu) +{ + return smu_send_smc_msg(smu, SMU_MSG_DisableAllSmuFeatures); +} + +static int smu_hw_fini(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct smu_context *smu = &adev->smu; + struct smu_table_context *table_context = &smu->smu_table; + int ret = 0; + + if (adev->flags & AMD_IS_APU) { + smu_powergate_sdma(&adev->smu, true); + smu_powergate_vcn(&adev->smu, true); + } + + ret = smu_stop_thermal_control(smu); + if (ret) { + 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); + table_context->driver_pptable = NULL; + + kfree(table_context->max_sustainable_clocks); + table_context->max_sustainable_clocks = NULL; + + kfree(table_context->overdrive_table); + table_context->overdrive_table = NULL; + + ret = smu_fini_fb_allocations(smu); + if (ret) + return ret; + + ret = smu_free_memory_pool(smu); + if (ret) + return ret; + + return 0; +} + +int smu_reset(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + int ret = 0; + + ret = smu_hw_fini(adev); + if (ret) + return ret; + + ret = smu_hw_init(adev); + if (ret) + return ret; + + return ret; +} + +static int smu_suspend(void *handle) +{ + int ret; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct smu_context *smu = &adev->smu; + bool baco_feature_is_enabled = false; + + if(!(adev->flags & AMD_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) { + ret = smu_feature_set_enabled(smu, SMU_FEATURE_BACO_BIT, true); + if (ret) { + pr_warn("set BACO feature enabled failed, return %d\n", ret); + return ret; + } + } + + smu->watermarks_bitmap &= ~(WATERMARKS_LOADED); + + if (adev->asic_type >= CHIP_NAVI10 && + adev->gfx.rlc.funcs->stop) + adev->gfx.rlc.funcs->stop(adev); + if (smu->is_apu) + smu_set_gfx_cgpg(&adev->smu, false); + + return 0; +} + +static int smu_resume(void *handle) +{ + int ret; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct smu_context *smu = &adev->smu; + + pr_info("SMU is resuming...\n"); + + ret = smu_start_smc_engine(smu); + if (ret) { + pr_err("SMU is not ready yet!\n"); + goto failed; + } + + ret = smu_smc_table_hw_init(smu, false); + if (ret) + goto failed; + + ret = smu_start_thermal_control(smu); + if (ret) + goto failed; + + if (smu->is_apu) + smu_set_gfx_cgpg(&adev->smu, true); + + smu->disable_uclk_switch = 0; + + pr_info("SMU is resumed successfully!\n"); + + return 0; + +failed: + return ret; +} + +int smu_display_configuration_change(struct smu_context *smu, + const struct amd_pp_display_configuration *display_config) +{ + int index = 0; + int num_of_active_display = 0; + + if (!smu->pm_enabled || !is_support_sw_smu(smu->adev)) + return -EINVAL; + + if (!display_config) + return -EINVAL; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->set_deep_sleep_dcefclk) + smu->ppt_funcs->set_deep_sleep_dcefclk(smu, + display_config->min_dcef_deep_sleep_set_clk / 100); + + for (index = 0; index < display_config->num_path_including_non_display; index++) { + if (display_config->displays[index].controller_id != 0) + num_of_active_display++; + } + + smu_set_active_display_count(smu, num_of_active_display); + + smu_store_cc6_data(smu, display_config->cpu_pstate_separation_time, + display_config->cpu_cc6_disable, + display_config->cpu_pstate_disable, + display_config->nb_pstate_switch_disable); + + mutex_unlock(&smu->mutex); + + return 0; +} + +static int smu_get_clock_info(struct smu_context *smu, + struct smu_clock_info *clk_info, + enum smu_perf_level_designation designation) +{ + int ret; + struct smu_performance_level level = {0}; + + if (!clk_info) + return -EINVAL; + + ret = smu_get_perf_level(smu, PERF_LEVEL_ACTIVITY, &level); + if (ret) + return -EINVAL; + + clk_info->min_mem_clk = level.memory_clock; + clk_info->min_eng_clk = level.core_clock; + clk_info->min_bus_bandwidth = level.non_local_mem_freq * level.non_local_mem_width; + + ret = smu_get_perf_level(smu, designation, &level); + if (ret) + return -EINVAL; + + clk_info->min_mem_clk = level.memory_clock; + clk_info->min_eng_clk = level.core_clock; + clk_info->min_bus_bandwidth = level.non_local_mem_freq * level.non_local_mem_width; + + return 0; +} + +int smu_get_current_clocks(struct smu_context *smu, + struct amd_pp_clock_info *clocks) +{ + struct amd_pp_simple_clock_info simple_clocks = {0}; + struct smu_clock_info hw_clocks; + int ret = 0; + + if (!is_support_sw_smu(smu->adev)) + return -EINVAL; + + mutex_lock(&smu->mutex); + + smu_get_dal_power_level(smu, &simple_clocks); + + if (smu->support_power_containment) + ret = smu_get_clock_info(smu, &hw_clocks, + PERF_LEVEL_POWER_CONTAINMENT); + else + ret = smu_get_clock_info(smu, &hw_clocks, PERF_LEVEL_ACTIVITY); + + if (ret) { + pr_err("Error in smu_get_clock_info\n"); + goto failed; + } + + clocks->min_engine_clock = hw_clocks.min_eng_clk; + clocks->max_engine_clock = hw_clocks.max_eng_clk; + clocks->min_memory_clock = hw_clocks.min_mem_clk; + clocks->max_memory_clock = hw_clocks.max_mem_clk; + clocks->min_bus_bandwidth = hw_clocks.min_bus_bandwidth; + clocks->max_bus_bandwidth = hw_clocks.max_bus_bandwidth; + clocks->max_engine_clock_in_sr = hw_clocks.max_eng_clk; + clocks->min_engine_clock_in_sr = hw_clocks.min_eng_clk; + + if (simple_clocks.level == 0) + clocks->max_clocks_state = PP_DAL_POWERLEVEL_7; + else + clocks->max_clocks_state = simple_clocks.level; + + if (!smu_get_current_shallow_sleep_clocks(smu, &hw_clocks)) { + clocks->max_engine_clock_in_sr = hw_clocks.max_eng_clk; + clocks->min_engine_clock_in_sr = hw_clocks.min_eng_clk; + } + +failed: + mutex_unlock(&smu->mutex); + return ret; +} + +static int smu_set_clockgating_state(void *handle, + enum amd_clockgating_state state) +{ + return 0; +} + +static int smu_set_powergating_state(void *handle, + enum amd_powergating_state state) +{ + return 0; +} + +static int smu_enable_umd_pstate(void *handle, + enum amd_dpm_forced_level *level) +{ + uint32_t profile_mode_mask = AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD | + AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK | + AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK | + AMD_DPM_FORCED_LEVEL_PROFILE_PEAK; + + struct smu_context *smu = (struct smu_context*)(handle); + struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm); + + if (!smu->is_apu && (!smu->pm_enabled || !smu_dpm_ctx->dpm_context)) + return -EINVAL; + + if (!(smu_dpm_ctx->dpm_level & profile_mode_mask)) { + /* enter umd pstate, save current level, disable gfx cg*/ + if (*level & profile_mode_mask) { + smu_dpm_ctx->saved_dpm_level = smu_dpm_ctx->dpm_level; + smu_dpm_ctx->enable_umd_pstate = true; + amdgpu_device_ip_set_clockgating_state(smu->adev, + AMD_IP_BLOCK_TYPE_GFX, + AMD_CG_STATE_UNGATE); + amdgpu_device_ip_set_powergating_state(smu->adev, + AMD_IP_BLOCK_TYPE_GFX, + AMD_PG_STATE_UNGATE); + } + } else { + /* exit umd pstate, restore level, enable gfx cg*/ + if (!(*level & profile_mode_mask)) { + if (*level == AMD_DPM_FORCED_LEVEL_PROFILE_EXIT) + *level = smu_dpm_ctx->saved_dpm_level; + smu_dpm_ctx->enable_umd_pstate = false; + amdgpu_device_ip_set_clockgating_state(smu->adev, + AMD_IP_BLOCK_TYPE_GFX, + AMD_CG_STATE_GATE); + amdgpu_device_ip_set_powergating_state(smu->adev, + AMD_IP_BLOCK_TYPE_GFX, + AMD_PG_STATE_GATE); + } + } + + 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) +{ + int ret = 0; + int index = 0; + long workload; + struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm); + + if (!smu->pm_enabled) + return -EINVAL; + + if (!skip_display_settings) { + ret = smu_display_config_changed(smu); + if (ret) { + pr_err("Failed to change display config!"); + return ret; + } + } + + ret = smu_apply_clocks_adjust_rules(smu); + if (ret) { + pr_err("Failed to apply clocks adjust rules!"); + return ret; + } + + if (!skip_display_settings) { + ret = smu_notify_smc_dispaly_config(smu); + if (ret) { + pr_err("Failed to notify smc display config!"); + return ret; + } + } + + 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; + } + } + + /* update the saved copy */ + smu_dpm_ctx->dpm_level = level; + } + + if (smu_dpm_ctx->dpm_level != AMD_DPM_FORCED_LEVEL_MANUAL) { + index = fls(smu->workload_mask); + index = index > 0 && index <= WORKLOAD_POLICY_MAX ? index - 1 : 0; + workload = smu->workload_setting[index]; + + if (smu->power_profile_mode != workload) + smu_set_power_profile_mode(smu, &workload, 0, false); + } + + return ret; +} + +int smu_handle_task(struct smu_context *smu, + enum amd_dpm_forced_level level, + enum amd_pp_task task_id, + bool lock_needed) +{ + int ret = 0; + + if (lock_needed) + mutex_lock(&smu->mutex); + + switch (task_id) { + case AMD_PP_TASK_DISPLAY_CONFIG_CHANGE: + ret = smu_pre_display_config_changed(smu); + if (ret) + goto out; + ret = smu_set_cpu_power_state(smu); + if (ret) + goto out; + ret = smu_adjust_power_state_dynamic(smu, level, false); + break; + case AMD_PP_TASK_COMPLETE_INIT: + case AMD_PP_TASK_READJUST_POWER_STATE: + ret = smu_adjust_power_state_dynamic(smu, level, true); + break; + default: + break; + } + +out: + if (lock_needed) + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_switch_power_profile(struct smu_context *smu, + enum PP_SMC_POWER_PROFILE type, + bool en) +{ + struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm); + long workload; + uint32_t index; + + if (!smu->pm_enabled) + return -EINVAL; + + if (!(type < PP_SMC_POWER_PROFILE_CUSTOM)) + return -EINVAL; + + mutex_lock(&smu->mutex); + + if (!en) { + smu->workload_mask &= ~(1 << smu->workload_prority[type]); + index = fls(smu->workload_mask); + index = index > 0 && index <= WORKLOAD_POLICY_MAX ? index - 1 : 0; + workload = smu->workload_setting[index]; + } else { + smu->workload_mask |= (1 << smu->workload_prority[type]); + index = fls(smu->workload_mask); + index = index <= WORKLOAD_POLICY_MAX ? index - 1 : 0; + workload = smu->workload_setting[index]; + } + + if (smu_dpm_ctx->dpm_level != AMD_DPM_FORCED_LEVEL_MANUAL) + smu_set_power_profile_mode(smu, &workload, 0, false); + + mutex_unlock(&smu->mutex); + + return 0; +} + +enum amd_dpm_forced_level smu_get_performance_level(struct smu_context *smu) +{ + struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm); + enum amd_dpm_forced_level level; + + if (!smu->is_apu && !smu_dpm_ctx->dpm_context) + return -EINVAL; + + mutex_lock(&(smu->mutex)); + level = smu_dpm_ctx->dpm_level; + mutex_unlock(&(smu->mutex)); + + return level; +} + +int smu_force_performance_level(struct smu_context *smu, enum amd_dpm_forced_level level) +{ + struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm); + int ret = 0; + + if (!smu->is_apu && !smu_dpm_ctx->dpm_context) + return -EINVAL; + + mutex_lock(&smu->mutex); + + ret = smu_enable_umd_pstate(smu, &level); + if (ret) { + mutex_unlock(&smu->mutex); + return ret; + } + + ret = smu_handle_task(smu, level, + AMD_PP_TASK_READJUST_POWER_STATE, + false); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_set_display_count(struct smu_context *smu, uint32_t count) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + ret = smu_init_display_count(smu, count); + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_force_clk_levels(struct smu_context *smu, + enum smu_clk_type clk_type, + uint32_t mask, + bool lock_needed) +{ + struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm); + int ret = 0; + + if (smu_dpm_ctx->dpm_level != AMD_DPM_FORCED_LEVEL_MANUAL) { + pr_debug("force clock level is for dpm manual mode only.\n"); + return -EINVAL; + } + + if (lock_needed) + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs && smu->ppt_funcs->force_clk_levels) + ret = smu->ppt_funcs->force_clk_levels(smu, clk_type, mask); + + if (lock_needed) + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_set_mp1_state(struct smu_context *smu, + enum pp_mp1_state mp1_state) +{ + uint16_t msg; + int ret; + + /* + * The SMC is not fully ready. That may be + * expected as the IP may be masked. + * So, just return without error. + */ + if (!smu->pm_enabled) + return 0; + + mutex_lock(&smu->mutex); + + switch (mp1_state) { + case PP_MP1_STATE_SHUTDOWN: + msg = SMU_MSG_PrepareMp1ForShutdown; + break; + case PP_MP1_STATE_UNLOAD: + msg = SMU_MSG_PrepareMp1ForUnload; + break; + case PP_MP1_STATE_RESET: + msg = SMU_MSG_PrepareMp1ForReset; + break; + case PP_MP1_STATE_NONE: + default: + mutex_unlock(&smu->mutex); + return 0; + } + + /* some asics may not support those messages */ + if (smu_msg_get_index(smu, msg) < 0) { + mutex_unlock(&smu->mutex); + return 0; + } + + ret = smu_send_smc_msg(smu, msg); + if (ret) + pr_err("[PrepareMp1] Failed!\n"); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_set_df_cstate(struct smu_context *smu, + enum pp_df_cstate state) +{ + int ret = 0; + + /* + * The SMC is not fully ready. That may be + * expected as the IP may be masked. + * So, just return without error. + */ + if (!smu->pm_enabled) + return 0; + + if (!smu->ppt_funcs || !smu->ppt_funcs->set_df_cstate) + return 0; + + mutex_lock(&smu->mutex); + + ret = smu->ppt_funcs->set_df_cstate(smu, state); + if (ret) + pr_err("[SetDfCstate] failed!\n"); + + mutex_unlock(&smu->mutex); + + return ret; +} + +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]; + + if (!table->cpu_addr) + return -EINVAL; + + ret = smu_update_table(smu, SMU_TABLE_WATERMARKS, 0, table->cpu_addr, + 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; + + mutex_lock(&smu->mutex); + + if (!smu->disable_watermark && + smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT) && + smu_feature_is_enabled(smu, SMU_FEATURE_DPM_SOCCLK_BIT)) { + smu_set_watermarks_table(smu, table, clock_ranges); + smu->watermarks_bitmap |= WATERMARKS_EXIST; + smu->watermarks_bitmap &= ~WATERMARKS_LOADED; + } + + mutex_unlock(&smu->mutex); + + return ret; +} + +const struct amd_ip_funcs smu_ip_funcs = { + .name = "smu", + .early_init = smu_early_init, + .late_init = smu_late_init, + .sw_init = smu_sw_init, + .sw_fini = smu_sw_fini, + .hw_init = smu_hw_init, + .hw_fini = smu_hw_fini, + .suspend = smu_suspend, + .resume = smu_resume, + .is_idle = NULL, + .check_soft_reset = NULL, + .wait_for_idle = NULL, + .soft_reset = NULL, + .set_clockgating_state = smu_set_clockgating_state, + .set_powergating_state = smu_set_powergating_state, + .enable_umd_pstate = smu_enable_umd_pstate, +}; + +const struct amdgpu_ip_block_version smu_v11_0_ip_block = +{ + .type = AMD_IP_BLOCK_TYPE_SMC, + .major = 11, + .minor = 0, + .rev = 0, + .funcs = &smu_ip_funcs, +}; + +const struct amdgpu_ip_block_version smu_v12_0_ip_block = +{ + .type = AMD_IP_BLOCK_TYPE_SMC, + .major = 12, + .minor = 0, + .rev = 0, + .funcs = &smu_ip_funcs, +}; + +int smu_load_microcode(struct smu_context *smu) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->load_microcode) + ret = smu->ppt_funcs->load_microcode(smu); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_check_fw_status(struct smu_context *smu) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->check_fw_status) + ret = smu->ppt_funcs->check_fw_status(smu); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_set_gfx_cgpg(struct smu_context *smu, bool enabled) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->set_gfx_cgpg) + ret = smu->ppt_funcs->set_gfx_cgpg(smu, enabled); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_set_fan_speed_rpm(struct smu_context *smu, uint32_t speed) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->set_fan_speed_rpm) + ret = smu->ppt_funcs->set_fan_speed_rpm(smu, speed); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_get_power_limit(struct smu_context *smu, + uint32_t *limit, + bool def, + bool lock_needed) +{ + int ret = 0; + + if (lock_needed) + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->get_power_limit) + ret = smu->ppt_funcs->get_power_limit(smu, limit, def); + + if (lock_needed) + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_set_power_limit(struct smu_context *smu, uint32_t limit) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->set_power_limit) + ret = smu->ppt_funcs->set_power_limit(smu, limit); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_print_clk_levels(struct smu_context *smu, enum smu_clk_type clk_type, char *buf) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->print_clk_levels) + ret = smu->ppt_funcs->print_clk_levels(smu, clk_type, buf); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_get_od_percentage(struct smu_context *smu, enum smu_clk_type type) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->get_od_percentage) + ret = smu->ppt_funcs->get_od_percentage(smu, type); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_set_od_percentage(struct smu_context *smu, enum smu_clk_type type, uint32_t value) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->set_od_percentage) + ret = smu->ppt_funcs->set_od_percentage(smu, type, value); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_od_edit_dpm_table(struct smu_context *smu, + enum PP_OD_DPM_TABLE_COMMAND type, + long *input, uint32_t size) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->od_edit_dpm_table) + ret = smu->ppt_funcs->od_edit_dpm_table(smu, type, input, size); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_read_sensor(struct smu_context *smu, + enum amd_pp_sensors sensor, + void *data, uint32_t *size) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->read_sensor) + ret = smu->ppt_funcs->read_sensor(smu, sensor, data, size); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_get_power_profile_mode(struct smu_context *smu, char *buf) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->get_power_profile_mode) + ret = smu->ppt_funcs->get_power_profile_mode(smu, buf); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_set_power_profile_mode(struct smu_context *smu, + long *param, + uint32_t param_size, + bool lock_needed) +{ + int ret = 0; + + if (lock_needed) + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->set_power_profile_mode) + ret = smu->ppt_funcs->set_power_profile_mode(smu, param, param_size); + + if (lock_needed) + mutex_unlock(&smu->mutex); + + return ret; +} + + +int smu_get_fan_control_mode(struct smu_context *smu) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->get_fan_control_mode) + ret = smu->ppt_funcs->get_fan_control_mode(smu); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_set_fan_control_mode(struct smu_context *smu, int value) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->set_fan_control_mode) + ret = smu->ppt_funcs->set_fan_control_mode(smu, value); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_get_fan_speed_percent(struct smu_context *smu, uint32_t *speed) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->get_fan_speed_percent) + ret = smu->ppt_funcs->get_fan_speed_percent(smu, speed); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_set_fan_speed_percent(struct smu_context *smu, uint32_t speed) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->set_fan_speed_percent) + ret = smu->ppt_funcs->set_fan_speed_percent(smu, speed); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_get_fan_speed_rpm(struct smu_context *smu, uint32_t *speed) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->get_fan_speed_rpm) + ret = smu->ppt_funcs->get_fan_speed_rpm(smu, speed); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_set_deep_sleep_dcefclk(struct smu_context *smu, int clk) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->set_deep_sleep_dcefclk) + ret = smu->ppt_funcs->set_deep_sleep_dcefclk(smu, clk); + + mutex_unlock(&smu->mutex); + + return ret; +} + +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; +} + +int smu_get_clock_by_type(struct smu_context *smu, + enum amd_pp_clock_type type, + struct amd_pp_clocks *clocks) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->get_clock_by_type) + ret = smu->ppt_funcs->get_clock_by_type(smu, type, clocks); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_get_max_high_clocks(struct smu_context *smu, + struct amd_pp_simple_clock_info *clocks) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->get_max_high_clocks) + ret = smu->ppt_funcs->get_max_high_clocks(smu, clocks); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_get_clock_by_type_with_latency(struct smu_context *smu, + enum smu_clk_type clk_type, + struct pp_clock_levels_with_latency *clocks) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->get_clock_by_type_with_latency) + ret = smu->ppt_funcs->get_clock_by_type_with_latency(smu, clk_type, clocks); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_get_clock_by_type_with_voltage(struct smu_context *smu, + enum amd_pp_clock_type type, + struct pp_clock_levels_with_voltage *clocks) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->get_clock_by_type_with_voltage) + ret = smu->ppt_funcs->get_clock_by_type_with_voltage(smu, type, clocks); + + mutex_unlock(&smu->mutex); + + return ret; +} + + +int smu_display_clock_voltage_request(struct smu_context *smu, + struct pp_display_clock_request *clock_req) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->display_clock_voltage_request) + ret = smu->ppt_funcs->display_clock_voltage_request(smu, clock_req); + + mutex_unlock(&smu->mutex); + + return ret; +} + + +int smu_display_disable_memory_clock_switch(struct smu_context *smu, bool disable_memory_clock_switch) +{ + int ret = -EINVAL; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->display_disable_memory_clock_switch) + ret = smu->ppt_funcs->display_disable_memory_clock_switch(smu, disable_memory_clock_switch); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_notify_smu_enable_pwe(struct smu_context *smu) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->notify_smu_enable_pwe) + ret = smu->ppt_funcs->notify_smu_enable_pwe(smu); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_set_xgmi_pstate(struct smu_context *smu, + uint32_t pstate) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->set_xgmi_pstate) + ret = smu->ppt_funcs->set_xgmi_pstate(smu, pstate); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_set_azalia_d3_pme(struct smu_context *smu) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->set_azalia_d3_pme) + ret = smu->ppt_funcs->set_azalia_d3_pme(smu); + + mutex_unlock(&smu->mutex); + + return ret; +} + +bool smu_baco_is_support(struct smu_context *smu) +{ + bool ret = false; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->baco_is_support) + ret = smu->ppt_funcs->baco_is_support(smu); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_baco_get_state(struct smu_context *smu, enum smu_baco_state *state) +{ + if (smu->ppt_funcs->baco_get_state) + return -EINVAL; + + mutex_lock(&smu->mutex); + *state = smu->ppt_funcs->baco_get_state(smu); + mutex_unlock(&smu->mutex); + + return 0; +} + +int smu_baco_reset(struct smu_context *smu) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->baco_reset) + ret = smu->ppt_funcs->baco_reset(smu); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_mode2_reset(struct smu_context *smu) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->mode2_reset) + ret = smu->ppt_funcs->mode2_reset(smu); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_get_max_sustainable_clocks_by_dc(struct smu_context *smu, + struct pp_smu_nv_clock_table *max_clocks) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->get_max_sustainable_clocks_by_dc) + ret = smu->ppt_funcs->get_max_sustainable_clocks_by_dc(smu, max_clocks); + + mutex_unlock(&smu->mutex); + + return ret; +} + +int smu_get_uclk_dpm_states(struct smu_context *smu, + unsigned int *clock_values_in_khz, + unsigned int *num_states) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->get_uclk_dpm_states) + ret = smu->ppt_funcs->get_uclk_dpm_states(smu, clock_values_in_khz, num_states); + + mutex_unlock(&smu->mutex); + + return ret; +} + +enum amd_pm_state_type smu_get_current_power_state(struct smu_context *smu) +{ + enum amd_pm_state_type pm_state = POWER_STATE_TYPE_DEFAULT; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->get_current_power_state) + pm_state = smu->ppt_funcs->get_current_power_state(smu); + + mutex_unlock(&smu->mutex); + + return pm_state; +} + +int smu_get_dpm_clock_table(struct smu_context *smu, + struct dpm_clocks *clock_table) +{ + int ret = 0; + + mutex_lock(&smu->mutex); + + if (smu->ppt_funcs->get_dpm_clock_table) + ret = smu->ppt_funcs->get_dpm_clock_table(smu, clock_table); + + mutex_unlock(&smu->mutex); + + return ret; +} + +uint32_t smu_get_pptable_power_limit(struct smu_context *smu) +{ + uint32_t ret = 0; + + if (smu->ppt_funcs->get_pptable_power_limit) + ret = smu->ppt_funcs->get_pptable_power_limit(smu); + + return ret; +} diff --git a/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c b/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c new file mode 100644 index 000000000000..58c7c4a3053e --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c @@ -0,0 +1,2169 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "pp_debug.h" +#include <linux/firmware.h> +#include "amdgpu.h" +#include "amdgpu_smu.h" +#include "smu_internal.h" +#include "atomfirmware.h" +#include "amdgpu_atomfirmware.h" +#include "smu_v11_0.h" +#include "smu11_driver_if_arcturus.h" +#include "soc15_common.h" +#include "atom.h" +#include "power_state.h" +#include "arcturus_ppt.h" +#include "smu_v11_0_pptable.h" +#include "arcturus_ppsmc.h" +#include "nbio/nbio_7_4_sh_mask.h" +#include "amdgpu_xgmi.h" +#include <linux/i2c.h> +#include <linux/pci.h> +#include "amdgpu_ras.h" + +#define to_amdgpu_device(x) (container_of(x, struct amdgpu_ras, eeprom_control.eeprom_accessor))->adev + +#define CTF_OFFSET_EDGE 5 +#define CTF_OFFSET_HOTSPOT 5 +#define CTF_OFFSET_HBM 5 + +#define MSG_MAP(msg, index) \ + [SMU_MSG_##msg] = {1, (index)} +#define ARCTURUS_FEA_MAP(smu_feature, arcturus_feature) \ + [smu_feature] = {1, (arcturus_feature)} + +#define SMU_FEATURES_LOW_MASK 0x00000000FFFFFFFF +#define SMU_FEATURES_LOW_SHIFT 0 +#define SMU_FEATURES_HIGH_MASK 0xFFFFFFFF00000000 +#define SMU_FEATURES_HIGH_SHIFT 32 + +#define SMC_DPM_FEATURE ( \ + FEATURE_DPM_PREFETCHER_MASK | \ + FEATURE_DPM_GFXCLK_MASK | \ + FEATURE_DPM_UCLK_MASK | \ + FEATURE_DPM_SOCCLK_MASK | \ + FEATURE_DPM_MP0CLK_MASK | \ + FEATURE_DPM_FCLK_MASK | \ + FEATURE_DPM_XGMI_MASK) + +/* possible frequency drift (1Mhz) */ +#define EPSILON 1 + +static struct smu_11_0_cmn2aisc_mapping arcturus_message_map[SMU_MSG_MAX_COUNT] = { + MSG_MAP(TestMessage, PPSMC_MSG_TestMessage), + MSG_MAP(GetSmuVersion, PPSMC_MSG_GetSmuVersion), + MSG_MAP(GetDriverIfVersion, PPSMC_MSG_GetDriverIfVersion), + MSG_MAP(SetAllowedFeaturesMaskLow, PPSMC_MSG_SetAllowedFeaturesMaskLow), + MSG_MAP(SetAllowedFeaturesMaskHigh, PPSMC_MSG_SetAllowedFeaturesMaskHigh), + MSG_MAP(EnableAllSmuFeatures, PPSMC_MSG_EnableAllSmuFeatures), + MSG_MAP(DisableAllSmuFeatures, PPSMC_MSG_DisableAllSmuFeatures), + MSG_MAP(EnableSmuFeaturesLow, PPSMC_MSG_EnableSmuFeaturesLow), + MSG_MAP(EnableSmuFeaturesHigh, PPSMC_MSG_EnableSmuFeaturesHigh), + MSG_MAP(DisableSmuFeaturesLow, PPSMC_MSG_DisableSmuFeaturesLow), + MSG_MAP(DisableSmuFeaturesHigh, PPSMC_MSG_DisableSmuFeaturesHigh), + MSG_MAP(GetEnabledSmuFeaturesLow, PPSMC_MSG_GetEnabledSmuFeaturesLow), + MSG_MAP(GetEnabledSmuFeaturesHigh, PPSMC_MSG_GetEnabledSmuFeaturesHigh), + MSG_MAP(SetDriverDramAddrHigh, PPSMC_MSG_SetDriverDramAddrHigh), + MSG_MAP(SetDriverDramAddrLow, PPSMC_MSG_SetDriverDramAddrLow), + MSG_MAP(SetToolsDramAddrHigh, PPSMC_MSG_SetToolsDramAddrHigh), + MSG_MAP(SetToolsDramAddrLow, PPSMC_MSG_SetToolsDramAddrLow), + MSG_MAP(TransferTableSmu2Dram, PPSMC_MSG_TransferTableSmu2Dram), + MSG_MAP(TransferTableDram2Smu, PPSMC_MSG_TransferTableDram2Smu), + MSG_MAP(UseDefaultPPTable, PPSMC_MSG_UseDefaultPPTable), + MSG_MAP(UseBackupPPTable, PPSMC_MSG_UseBackupPPTable), + MSG_MAP(SetSystemVirtualDramAddrHigh, PPSMC_MSG_SetSystemVirtualDramAddrHigh), + MSG_MAP(SetSystemVirtualDramAddrLow, PPSMC_MSG_SetSystemVirtualDramAddrLow), + MSG_MAP(EnterBaco, PPSMC_MSG_EnterBaco), + MSG_MAP(ExitBaco, PPSMC_MSG_ExitBaco), + MSG_MAP(ArmD3, PPSMC_MSG_ArmD3), + MSG_MAP(SetSoftMinByFreq, PPSMC_MSG_SetSoftMinByFreq), + MSG_MAP(SetSoftMaxByFreq, PPSMC_MSG_SetSoftMaxByFreq), + MSG_MAP(SetHardMinByFreq, PPSMC_MSG_SetHardMinByFreq), + MSG_MAP(SetHardMaxByFreq, PPSMC_MSG_SetHardMaxByFreq), + MSG_MAP(GetMinDpmFreq, PPSMC_MSG_GetMinDpmFreq), + MSG_MAP(GetMaxDpmFreq, PPSMC_MSG_GetMaxDpmFreq), + MSG_MAP(GetDpmFreqByIndex, PPSMC_MSG_GetDpmFreqByIndex), + MSG_MAP(SetWorkloadMask, PPSMC_MSG_SetWorkloadMask), + MSG_MAP(SetDfSwitchType, PPSMC_MSG_SetDfSwitchType), + MSG_MAP(GetVoltageByDpm, PPSMC_MSG_GetVoltageByDpm), + MSG_MAP(GetVoltageByDpmOverdrive, PPSMC_MSG_GetVoltageByDpmOverdrive), + MSG_MAP(SetPptLimit, PPSMC_MSG_SetPptLimit), + MSG_MAP(GetPptLimit, PPSMC_MSG_GetPptLimit), + MSG_MAP(PowerUpVcn0, PPSMC_MSG_PowerUpVcn0), + MSG_MAP(PowerDownVcn0, PPSMC_MSG_PowerDownVcn0), + MSG_MAP(PowerUpVcn1, PPSMC_MSG_PowerUpVcn1), + MSG_MAP(PowerDownVcn1, PPSMC_MSG_PowerDownVcn1), + MSG_MAP(PrepareMp1ForUnload, PPSMC_MSG_PrepareMp1ForUnload), + MSG_MAP(PrepareMp1ForReset, PPSMC_MSG_PrepareMp1ForReset), + MSG_MAP(PrepareMp1ForShutdown, PPSMC_MSG_PrepareMp1ForShutdown), + MSG_MAP(SoftReset, PPSMC_MSG_SoftReset), + MSG_MAP(RunAfllBtc, PPSMC_MSG_RunAfllBtc), + MSG_MAP(RunDcBtc, PPSMC_MSG_RunDcBtc), + MSG_MAP(DramLogSetDramAddrHigh, PPSMC_MSG_DramLogSetDramAddrHigh), + MSG_MAP(DramLogSetDramAddrLow, PPSMC_MSG_DramLogSetDramAddrLow), + MSG_MAP(DramLogSetDramSize, PPSMC_MSG_DramLogSetDramSize), + MSG_MAP(GetDebugData, PPSMC_MSG_GetDebugData), + MSG_MAP(WaflTest, PPSMC_MSG_WaflTest), + MSG_MAP(SetXgmiMode, PPSMC_MSG_SetXgmiMode), + MSG_MAP(SetMemoryChannelEnable, PPSMC_MSG_SetMemoryChannelEnable), +}; + +static struct smu_11_0_cmn2aisc_mapping arcturus_clk_map[SMU_CLK_COUNT] = { + CLK_MAP(GFXCLK, PPCLK_GFXCLK), + CLK_MAP(SCLK, PPCLK_GFXCLK), + CLK_MAP(SOCCLK, PPCLK_SOCCLK), + CLK_MAP(FCLK, PPCLK_FCLK), + CLK_MAP(UCLK, PPCLK_UCLK), + CLK_MAP(MCLK, PPCLK_UCLK), + CLK_MAP(DCLK, PPCLK_DCLK), + CLK_MAP(VCLK, PPCLK_VCLK), +}; + +static struct smu_11_0_cmn2aisc_mapping arcturus_feature_mask_map[SMU_FEATURE_COUNT] = { + FEA_MAP(DPM_PREFETCHER), + FEA_MAP(DPM_GFXCLK), + FEA_MAP(DPM_UCLK), + FEA_MAP(DPM_SOCCLK), + FEA_MAP(DPM_FCLK), + FEA_MAP(DPM_MP0CLK), + ARCTURUS_FEA_MAP(SMU_FEATURE_XGMI_BIT, FEATURE_DPM_XGMI_BIT), + FEA_MAP(DS_GFXCLK), + FEA_MAP(DS_SOCCLK), + FEA_MAP(DS_LCLK), + FEA_MAP(DS_FCLK), + FEA_MAP(DS_UCLK), + FEA_MAP(GFX_ULV), + ARCTURUS_FEA_MAP(SMU_FEATURE_VCN_PG_BIT, FEATURE_DPM_VCN_BIT), + FEA_MAP(RSMU_SMN_CG), + FEA_MAP(WAFL_CG), + FEA_MAP(PPT), + FEA_MAP(TDC), + FEA_MAP(APCC_PLUS), + FEA_MAP(VR0HOT), + FEA_MAP(VR1HOT), + FEA_MAP(FW_CTF), + FEA_MAP(FAN_CONTROL), + FEA_MAP(THERMAL), + FEA_MAP(OUT_OF_BAND_MONITOR), + FEA_MAP(TEMP_DEPENDENT_VMIN), +}; + +static struct smu_11_0_cmn2aisc_mapping arcturus_table_map[SMU_TABLE_COUNT] = { + TAB_MAP(PPTABLE), + TAB_MAP(AVFS), + TAB_MAP(AVFS_PSM_DEBUG), + TAB_MAP(AVFS_FUSE_OVERRIDE), + TAB_MAP(PMSTATUSLOG), + TAB_MAP(SMU_METRICS), + TAB_MAP(DRIVER_SMU_CONFIG), + TAB_MAP(OVERDRIVE), + TAB_MAP(I2C_COMMANDS), +}; + +static struct smu_11_0_cmn2aisc_mapping arcturus_pwr_src_map[SMU_POWER_SOURCE_COUNT] = { + PWR_MAP(AC), + PWR_MAP(DC), +}; + +static struct smu_11_0_cmn2aisc_mapping arcturus_workload_map[PP_SMC_POWER_PROFILE_COUNT] = { + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT, WORKLOAD_PPLIB_DEFAULT_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_POWERSAVING, WORKLOAD_PPLIB_POWER_SAVING_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_VIDEO, WORKLOAD_PPLIB_VIDEO_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_COMPUTE, WORKLOAD_PPLIB_COMPUTE_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_CUSTOM, WORKLOAD_PPLIB_CUSTOM_BIT), +}; + +static int arcturus_get_smu_msg_index(struct smu_context *smc, uint32_t index) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (index >= SMU_MSG_MAX_COUNT) + return -EINVAL; + + mapping = arcturus_message_map[index]; + if (!(mapping.valid_mapping)) + return -EINVAL; + + return mapping.map_to; +} + +static int arcturus_get_smu_clk_index(struct smu_context *smc, uint32_t index) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (index >= SMU_CLK_COUNT) + return -EINVAL; + + mapping = arcturus_clk_map[index]; + if (!(mapping.valid_mapping)) { + pr_warn("Unsupported SMU clk: %d\n", index); + return -EINVAL; + } + + return mapping.map_to; +} + +static int arcturus_get_smu_feature_index(struct smu_context *smc, uint32_t index) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (index >= SMU_FEATURE_COUNT) + return -EINVAL; + + mapping = arcturus_feature_mask_map[index]; + if (!(mapping.valid_mapping)) { + return -EINVAL; + } + + return mapping.map_to; +} + +static int arcturus_get_smu_table_index(struct smu_context *smc, uint32_t index) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (index >= SMU_TABLE_COUNT) + return -EINVAL; + + mapping = arcturus_table_map[index]; + if (!(mapping.valid_mapping)) { + pr_warn("Unsupported SMU table: %d\n", index); + return -EINVAL; + } + + return mapping.map_to; +} + +static int arcturus_get_pwr_src_index(struct smu_context *smc, uint32_t index) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (index >= SMU_POWER_SOURCE_COUNT) + return -EINVAL; + + mapping = arcturus_pwr_src_map[index]; + if (!(mapping.valid_mapping)) { + pr_warn("Unsupported SMU power source: %d\n", index); + return -EINVAL; + } + + return mapping.map_to; +} + + +static int arcturus_get_workload_type(struct smu_context *smu, enum PP_SMC_POWER_PROFILE profile) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (profile > PP_SMC_POWER_PROFILE_CUSTOM) + return -EINVAL; + + mapping = arcturus_workload_map[profile]; + if (!(mapping.valid_mapping)) { + pr_warn("Unsupported SMU power source: %d\n", profile); + return -EINVAL; + } + + return mapping.map_to; +} + +static int arcturus_tables_init(struct smu_context *smu, struct smu_table *tables) +{ + struct smu_table_context *smu_table = &smu->smu_table; + + SMU_TABLE_INIT(tables, SMU_TABLE_PPTABLE, sizeof(PPTable_t), + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + + SMU_TABLE_INIT(tables, SMU_TABLE_PMSTATUSLOG, SMU11_TOOL_SIZE, + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + + SMU_TABLE_INIT(tables, SMU_TABLE_SMU_METRICS, sizeof(SmuMetrics_t), + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + + SMU_TABLE_INIT(tables, SMU_TABLE_I2C_COMMANDS, sizeof(SwI2cRequest_t), + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + + smu_table->metrics_table = kzalloc(sizeof(SmuMetrics_t), GFP_KERNEL); + if (!smu_table->metrics_table) + return -ENOMEM; + smu_table->metrics_time = 0; + + return 0; +} + +static int arcturus_allocate_dpm_context(struct smu_context *smu) +{ + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + + if (smu_dpm->dpm_context) + return -EINVAL; + + smu_dpm->dpm_context = kzalloc(sizeof(struct arcturus_dpm_table), + GFP_KERNEL); + if (!smu_dpm->dpm_context) + return -ENOMEM; + + if (smu_dpm->golden_dpm_context) + return -EINVAL; + + smu_dpm->golden_dpm_context = kzalloc(sizeof(struct arcturus_dpm_table), + GFP_KERNEL); + if (!smu_dpm->golden_dpm_context) + return -ENOMEM; + + smu_dpm->dpm_context_size = sizeof(struct arcturus_dpm_table); + + smu_dpm->dpm_current_power_state = kzalloc(sizeof(struct smu_power_state), + GFP_KERNEL); + if (!smu_dpm->dpm_current_power_state) + return -ENOMEM; + + smu_dpm->dpm_request_power_state = kzalloc(sizeof(struct smu_power_state), + GFP_KERNEL); + if (!smu_dpm->dpm_request_power_state) + return -ENOMEM; + + return 0; +} + +static int +arcturus_get_allowed_feature_mask(struct smu_context *smu, + uint32_t *feature_mask, uint32_t num) +{ + if (num > 2) + return -EINVAL; + + /* pptable will handle the features to enable */ + memset(feature_mask, 0xFF, sizeof(uint32_t) * num); + + return 0; +} + +static int +arcturus_set_single_dpm_table(struct smu_context *smu, + struct arcturus_single_dpm_table *single_dpm_table, + PPCLK_e clk_id) +{ + int ret = 0; + uint32_t i, num_of_levels = 0, clk; + + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_GetDpmFreqByIndex, + (clk_id << 16 | 0xFF)); + if (ret) { + pr_err("[%s] failed to get dpm levels!\n", __func__); + return ret; + } + + smu_read_smc_arg(smu, &num_of_levels); + if (!num_of_levels) { + pr_err("[%s] number of clk levels is invalid!\n", __func__); + return -EINVAL; + } + + single_dpm_table->count = num_of_levels; + for (i = 0; i < num_of_levels; i++) { + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_GetDpmFreqByIndex, + (clk_id << 16 | i)); + if (ret) { + pr_err("[%s] failed to get dpm freq by index!\n", __func__); + return ret; + } + smu_read_smc_arg(smu, &clk); + if (!clk) { + pr_err("[%s] clk value is invalid!\n", __func__); + return -EINVAL; + } + single_dpm_table->dpm_levels[i].value = clk; + single_dpm_table->dpm_levels[i].enabled = true; + } + return 0; +} + +static void arcturus_init_single_dpm_state(struct arcturus_dpm_state *dpm_state) +{ + dpm_state->soft_min_level = 0x0; + dpm_state->soft_max_level = 0xffff; + dpm_state->hard_min_level = 0x0; + dpm_state->hard_max_level = 0xffff; +} + +static int arcturus_set_default_dpm_table(struct smu_context *smu) +{ + int ret; + + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + struct arcturus_dpm_table *dpm_table = NULL; + struct arcturus_single_dpm_table *single_dpm_table; + + dpm_table = smu_dpm->dpm_context; + + /* socclk */ + single_dpm_table = &(dpm_table->soc_table); + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_SOCCLK_BIT)) { + ret = arcturus_set_single_dpm_table(smu, single_dpm_table, + PPCLK_SOCCLK); + if (ret) { + pr_err("[%s] failed to get socclk dpm levels!\n", __func__); + return ret; + } + } else { + single_dpm_table->count = 1; + single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.socclk / 100; + } + arcturus_init_single_dpm_state(&(single_dpm_table->dpm_state)); + + /* gfxclk */ + single_dpm_table = &(dpm_table->gfx_table); + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_GFXCLK_BIT)) { + ret = arcturus_set_single_dpm_table(smu, single_dpm_table, + PPCLK_GFXCLK); + if (ret) { + pr_err("[SetupDefaultDpmTable] failed to get gfxclk dpm levels!"); + return ret; + } + } else { + single_dpm_table->count = 1; + single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.gfxclk / 100; + } + arcturus_init_single_dpm_state(&(single_dpm_table->dpm_state)); + + /* memclk */ + single_dpm_table = &(dpm_table->mem_table); + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT)) { + ret = arcturus_set_single_dpm_table(smu, single_dpm_table, + PPCLK_UCLK); + if (ret) { + pr_err("[SetupDefaultDpmTable] failed to get memclk dpm levels!"); + return ret; + } + } else { + single_dpm_table->count = 1; + single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.uclk / 100; + } + arcturus_init_single_dpm_state(&(single_dpm_table->dpm_state)); + + /* fclk */ + single_dpm_table = &(dpm_table->fclk_table); + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_FCLK_BIT)) { + ret = arcturus_set_single_dpm_table(smu, single_dpm_table, + PPCLK_FCLK); + if (ret) { + pr_err("[SetupDefaultDpmTable] failed to get fclk dpm levels!"); + return ret; + } + } else { + single_dpm_table->count = 1; + single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.fclk / 100; + } + arcturus_init_single_dpm_state(&(single_dpm_table->dpm_state)); + + memcpy(smu_dpm->golden_dpm_context, dpm_table, + sizeof(struct arcturus_dpm_table)); + + return 0; +} + +static int arcturus_check_powerplay_table(struct smu_context *smu) +{ + return 0; +} + +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; + int ret = 0; + + if (!table_context->power_play_table) + return -EINVAL; + + powerplay_table = table_context->power_play_table; + + memcpy(table_context->driver_pptable, &powerplay_table->smc_pptable, + sizeof(PPTable_t)); + + table_context->thermal_controller_type = powerplay_table->thermal_controller_type; + + return ret; +} + +static int arcturus_append_powerplay_table(struct smu_context *smu) +{ + struct smu_table_context *table_context = &smu->smu_table; + PPTable_t *smc_pptable = table_context->driver_pptable; + struct atom_smc_dpm_info_v4_6 *smc_dpm_table; + int index, ret; + + index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1, + smc_dpm_info); + + ret = smu_get_atom_data_table(smu, index, NULL, NULL, NULL, + (uint8_t **)&smc_dpm_table); + if (ret) + return ret; + + pr_info("smc_dpm_info table revision(format.content): %d.%d\n", + smc_dpm_table->table_header.format_revision, + smc_dpm_table->table_header.content_revision); + + if ((smc_dpm_table->table_header.format_revision == 4) && + (smc_dpm_table->table_header.content_revision == 6)) + memcpy(&smc_pptable->MaxVoltageStepGfx, + &smc_dpm_table->maxvoltagestepgfx, + sizeof(*smc_dpm_table) - offsetof(struct atom_smc_dpm_info_v4_6, maxvoltagestepgfx)); + + return 0; +} + +static int arcturus_run_btc(struct smu_context *smu) +{ + int ret = 0; + + ret = smu_send_smc_msg(smu, SMU_MSG_RunAfllBtc); + if (ret) { + pr_err("RunAfllBtc failed!\n"); + return ret; + } + + return smu_send_smc_msg(smu, SMU_MSG_RunDcBtc); +} + +static int arcturus_populate_umd_state_clk(struct smu_context *smu) +{ + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + struct arcturus_dpm_table *dpm_table = NULL; + struct arcturus_single_dpm_table *gfx_table = NULL; + struct arcturus_single_dpm_table *mem_table = NULL; + + dpm_table = smu_dpm->dpm_context; + gfx_table = &(dpm_table->gfx_table); + mem_table = &(dpm_table->mem_table); + + smu->pstate_sclk = gfx_table->dpm_levels[0].value; + smu->pstate_mclk = mem_table->dpm_levels[0].value; + + if (gfx_table->count > ARCTURUS_UMD_PSTATE_GFXCLK_LEVEL && + mem_table->count > ARCTURUS_UMD_PSTATE_MCLK_LEVEL) { + smu->pstate_sclk = gfx_table->dpm_levels[ARCTURUS_UMD_PSTATE_GFXCLK_LEVEL].value; + smu->pstate_mclk = mem_table->dpm_levels[ARCTURUS_UMD_PSTATE_MCLK_LEVEL].value; + } + + smu->pstate_sclk = smu->pstate_sclk * 100; + smu->pstate_mclk = smu->pstate_mclk * 100; + + return 0; +} + +static int arcturus_get_clk_table(struct smu_context *smu, + struct pp_clock_levels_with_latency *clocks, + struct arcturus_single_dpm_table *dpm_table) +{ + int i, count; + + count = (dpm_table->count > MAX_NUM_CLOCKS) ? MAX_NUM_CLOCKS : dpm_table->count; + clocks->num_levels = count; + + for (i = 0; i < count; i++) { + clocks->data[i].clocks_in_khz = + dpm_table->dpm_levels[i].value * 1000; + clocks->data[i].latency_in_us = 0; + } + + return 0; +} + +static int arcturus_freqs_in_same_level(int32_t frequency1, + int32_t frequency2) +{ + return (abs(frequency1 - frequency2) <= EPSILON); +} + +static int arcturus_print_clk_levels(struct smu_context *smu, + enum smu_clk_type type, char *buf) +{ + int i, now, size = 0; + int ret = 0; + struct pp_clock_levels_with_latency clocks; + struct arcturus_single_dpm_table *single_dpm_table; + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + struct arcturus_dpm_table *dpm_table = NULL; + + dpm_table = smu_dpm->dpm_context; + + switch (type) { + case SMU_SCLK: + ret = smu_get_current_clk_freq(smu, SMU_GFXCLK, &now); + if (ret) { + pr_err("Attempt to get current gfx clk Failed!"); + return ret; + } + + single_dpm_table = &(dpm_table->gfx_table); + ret = arcturus_get_clk_table(smu, &clocks, single_dpm_table); + if (ret) { + pr_err("Attempt to get gfx clk levels Failed!"); + return ret; + } + + /* + * For DPM disabled case, there will be only one clock level. + * And it's safe to assume that is always the current clock. + */ + for (i = 0; i < clocks.num_levels; i++) + size += sprintf(buf + size, "%d: %uMhz %s\n", i, + clocks.data[i].clocks_in_khz / 1000, + (clocks.num_levels == 1) ? "*" : + (arcturus_freqs_in_same_level( + clocks.data[i].clocks_in_khz / 1000, + now / 100) ? "*" : "")); + break; + + case SMU_MCLK: + ret = smu_get_current_clk_freq(smu, SMU_UCLK, &now); + if (ret) { + pr_err("Attempt to get current mclk Failed!"); + return ret; + } + + single_dpm_table = &(dpm_table->mem_table); + ret = arcturus_get_clk_table(smu, &clocks, single_dpm_table); + if (ret) { + pr_err("Attempt to get memory clk levels Failed!"); + return ret; + } + + for (i = 0; i < clocks.num_levels; i++) + size += sprintf(buf + size, "%d: %uMhz %s\n", + i, clocks.data[i].clocks_in_khz / 1000, + (clocks.num_levels == 1) ? "*" : + (arcturus_freqs_in_same_level( + clocks.data[i].clocks_in_khz / 1000, + now / 100) ? "*" : "")); + break; + + case SMU_SOCCLK: + ret = smu_get_current_clk_freq(smu, SMU_SOCCLK, &now); + if (ret) { + pr_err("Attempt to get current socclk Failed!"); + return ret; + } + + single_dpm_table = &(dpm_table->soc_table); + ret = arcturus_get_clk_table(smu, &clocks, single_dpm_table); + if (ret) { + pr_err("Attempt to get socclk levels Failed!"); + return ret; + } + + for (i = 0; i < clocks.num_levels; i++) + size += sprintf(buf + size, "%d: %uMhz %s\n", + i, clocks.data[i].clocks_in_khz / 1000, + (clocks.num_levels == 1) ? "*" : + (arcturus_freqs_in_same_level( + clocks.data[i].clocks_in_khz / 1000, + now / 100) ? "*" : "")); + break; + + case SMU_FCLK: + ret = smu_get_current_clk_freq(smu, SMU_FCLK, &now); + if (ret) { + pr_err("Attempt to get current fclk Failed!"); + return ret; + } + + single_dpm_table = &(dpm_table->fclk_table); + ret = arcturus_get_clk_table(smu, &clocks, single_dpm_table); + if (ret) { + pr_err("Attempt to get fclk levels Failed!"); + return ret; + } + + for (i = 0; i < single_dpm_table->count; i++) + size += sprintf(buf + size, "%d: %uMhz %s\n", + i, single_dpm_table->dpm_levels[i].value, + (clocks.num_levels == 1) ? "*" : + (arcturus_freqs_in_same_level( + clocks.data[i].clocks_in_khz / 1000, + now / 100) ? "*" : "")); + break; + + default: + break; + } + + return size; +} + +static int arcturus_upload_dpm_level(struct smu_context *smu, bool max, + uint32_t feature_mask) +{ + struct arcturus_single_dpm_table *single_dpm_table; + struct arcturus_dpm_table *dpm_table = + smu->smu_dpm.dpm_context; + uint32_t freq; + int ret = 0; + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_GFXCLK_BIT) && + (feature_mask & FEATURE_DPM_GFXCLK_MASK)) { + single_dpm_table = &(dpm_table->gfx_table); + freq = max ? single_dpm_table->dpm_state.soft_max_level : + single_dpm_table->dpm_state.soft_min_level; + ret = smu_send_smc_msg_with_param(smu, + (max ? SMU_MSG_SetSoftMaxByFreq : SMU_MSG_SetSoftMinByFreq), + (PPCLK_GFXCLK << 16) | (freq & 0xffff)); + if (ret) { + pr_err("Failed to set soft %s gfxclk !\n", + max ? "max" : "min"); + return ret; + } + } + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT) && + (feature_mask & FEATURE_DPM_UCLK_MASK)) { + single_dpm_table = &(dpm_table->mem_table); + freq = max ? single_dpm_table->dpm_state.soft_max_level : + single_dpm_table->dpm_state.soft_min_level; + ret = smu_send_smc_msg_with_param(smu, + (max ? SMU_MSG_SetSoftMaxByFreq : SMU_MSG_SetSoftMinByFreq), + (PPCLK_UCLK << 16) | (freq & 0xffff)); + if (ret) { + pr_err("Failed to set soft %s memclk !\n", + max ? "max" : "min"); + return ret; + } + } + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_SOCCLK_BIT) && + (feature_mask & FEATURE_DPM_SOCCLK_MASK)) { + single_dpm_table = &(dpm_table->soc_table); + freq = max ? single_dpm_table->dpm_state.soft_max_level : + single_dpm_table->dpm_state.soft_min_level; + ret = smu_send_smc_msg_with_param(smu, + (max ? SMU_MSG_SetSoftMaxByFreq : SMU_MSG_SetSoftMinByFreq), + (PPCLK_SOCCLK << 16) | (freq & 0xffff)); + if (ret) { + pr_err("Failed to set soft %s socclk !\n", + max ? "max" : "min"); + return ret; + } + } + + return ret; +} + +static int arcturus_force_clk_levels(struct smu_context *smu, + enum smu_clk_type type, uint32_t mask) +{ + struct arcturus_dpm_table *dpm_table; + struct arcturus_single_dpm_table *single_dpm_table; + uint32_t soft_min_level, soft_max_level; + int ret = 0; + + soft_min_level = mask ? (ffs(mask) - 1) : 0; + soft_max_level = mask ? (fls(mask) - 1) : 0; + + dpm_table = smu->smu_dpm.dpm_context; + + switch (type) { + case SMU_SCLK: + single_dpm_table = &(dpm_table->gfx_table); + + if (soft_max_level >= single_dpm_table->count) { + pr_err("Clock level specified %d is over max allowed %d\n", + soft_max_level, single_dpm_table->count - 1); + ret = -EINVAL; + break; + } + + single_dpm_table->dpm_state.soft_min_level = + single_dpm_table->dpm_levels[soft_min_level].value; + single_dpm_table->dpm_state.soft_max_level = + single_dpm_table->dpm_levels[soft_max_level].value; + + ret = arcturus_upload_dpm_level(smu, false, FEATURE_DPM_GFXCLK_MASK); + if (ret) { + pr_err("Failed to upload boot level to lowest!\n"); + break; + } + + ret = arcturus_upload_dpm_level(smu, true, FEATURE_DPM_GFXCLK_MASK); + if (ret) + pr_err("Failed to upload dpm max level to highest!\n"); + + break; + + case SMU_MCLK: + case SMU_SOCCLK: + case SMU_FCLK: + /* + * Should not arrive here since Arcturus does not + * support mclk/socclk/fclk softmin/softmax settings + */ + ret = -EINVAL; + break; + + default: + break; + } + + return ret; +} + +static int arcturus_get_thermal_temperature_range(struct smu_context *smu, + struct smu_temperature_range *range) +{ + PPTable_t *pptable = smu->smu_table.driver_pptable; + + if (!range) + return -EINVAL; + + range->max = pptable->TedgeLimit * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + range->edge_emergency_max = (pptable->TedgeLimit + CTF_OFFSET_EDGE) * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + range->hotspot_crit_max = pptable->ThotspotLimit * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + range->hotspot_emergency_max = (pptable->ThotspotLimit + CTF_OFFSET_HOTSPOT) * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + range->mem_crit_max = pptable->TmemLimit * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + range->mem_emergency_max = (pptable->TmemLimit + CTF_OFFSET_HBM)* + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + + return 0; +} + +static int arcturus_get_metrics_table(struct smu_context *smu, + SmuMetrics_t *metrics_table) +{ + struct smu_table_context *smu_table= &smu->smu_table; + int ret = 0; + + if (!smu_table->metrics_time || + time_after(jiffies, smu_table->metrics_time + HZ / 1000)) { + 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"); + return ret; + } + smu_table->metrics_time = jiffies; + } + + memcpy(metrics_table, smu_table->metrics_table, sizeof(SmuMetrics_t)); + + return ret; +} + +static int arcturus_get_current_activity_percent(struct smu_context *smu, + enum amd_pp_sensors sensor, + uint32_t *value) +{ + SmuMetrics_t metrics; + int ret = 0; + + if (!value) + return -EINVAL; + + ret = arcturus_get_metrics_table(smu, &metrics); + if (ret) + return ret; + + switch (sensor) { + case AMDGPU_PP_SENSOR_GPU_LOAD: + *value = metrics.AverageGfxActivity; + break; + case AMDGPU_PP_SENSOR_MEM_LOAD: + *value = metrics.AverageUclkActivity; + break; + default: + pr_err("Invalid sensor for retrieving clock activity\n"); + return -EINVAL; + } + + return 0; +} + +static int arcturus_get_gpu_power(struct smu_context *smu, uint32_t *value) +{ + SmuMetrics_t metrics; + int ret = 0; + + if (!value) + return -EINVAL; + + ret = arcturus_get_metrics_table(smu, &metrics); + if (ret) + return ret; + + *value = metrics.AverageSocketPower << 8; + + return 0; +} + +static int arcturus_thermal_get_temperature(struct smu_context *smu, + enum amd_pp_sensors sensor, + uint32_t *value) +{ + SmuMetrics_t metrics; + int ret = 0; + + if (!value) + return -EINVAL; + + ret = arcturus_get_metrics_table(smu, &metrics); + if (ret) + return ret; + + switch (sensor) { + case AMDGPU_PP_SENSOR_HOTSPOT_TEMP: + *value = metrics.TemperatureHotspot * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + break; + case AMDGPU_PP_SENSOR_EDGE_TEMP: + *value = metrics.TemperatureEdge * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + break; + case AMDGPU_PP_SENSOR_MEM_TEMP: + *value = metrics.TemperatureHBM * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + break; + default: + pr_err("Invalid sensor for retrieving temp\n"); + return -EINVAL; + } + + return 0; +} + +static int arcturus_read_sensor(struct smu_context *smu, + enum amd_pp_sensors sensor, + void *data, uint32_t *size) +{ + struct smu_table_context *table_context = &smu->smu_table; + PPTable_t *pptable = table_context->driver_pptable; + int ret = 0; + + if (!data || !size) + return -EINVAL; + + mutex_lock(&smu->sensor_lock); + switch (sensor) { + case AMDGPU_PP_SENSOR_MAX_FAN_RPM: + *(uint32_t *)data = pptable->FanMaximumRpm; + *size = 4; + break; + case AMDGPU_PP_SENSOR_MEM_LOAD: + case AMDGPU_PP_SENSOR_GPU_LOAD: + ret = arcturus_get_current_activity_percent(smu, + sensor, + (uint32_t *)data); + *size = 4; + break; + case AMDGPU_PP_SENSOR_GPU_POWER: + ret = arcturus_get_gpu_power(smu, (uint32_t *)data); + *size = 4; + break; + case AMDGPU_PP_SENSOR_HOTSPOT_TEMP: + case AMDGPU_PP_SENSOR_EDGE_TEMP: + case AMDGPU_PP_SENSOR_MEM_TEMP: + ret = arcturus_thermal_get_temperature(smu, sensor, + (uint32_t *)data); + *size = 4; + break; + default: + ret = smu_v11_0_read_sensor(smu, sensor, data, size); + } + mutex_unlock(&smu->sensor_lock); + + return ret; +} + +static int arcturus_get_fan_speed_rpm(struct smu_context *smu, + uint32_t *speed) +{ + SmuMetrics_t metrics; + int ret = 0; + + if (!speed) + return -EINVAL; + + ret = arcturus_get_metrics_table(smu, &metrics); + if (ret) + return ret; + + *speed = metrics.CurrFanSpeed; + + return ret; +} + +static int arcturus_get_fan_speed_percent(struct smu_context *smu, + uint32_t *speed) +{ + PPTable_t *pptable = smu->smu_table.driver_pptable; + uint32_t percent, current_rpm; + int ret = 0; + + if (!speed) + return -EINVAL; + + ret = arcturus_get_fan_speed_rpm(smu, ¤t_rpm); + if (ret) + return ret; + + percent = current_rpm * 100 / pptable->FanMaximumRpm; + *speed = percent > 100 ? 100 : percent; + + return ret; +} + +static int arcturus_get_current_clk_freq_by_table(struct smu_context *smu, + enum smu_clk_type clk_type, + uint32_t *value) +{ + static SmuMetrics_t metrics; + int ret = 0, clk_id = 0; + + if (!value) + return -EINVAL; + + clk_id = smu_clk_get_index(smu, clk_type); + if (clk_id < 0) + return -EINVAL; + + ret = arcturus_get_metrics_table(smu, &metrics); + if (ret) + return ret; + + switch (clk_id) { + case PPCLK_GFXCLK: + /* + * CurrClock[clk_id] can provide accurate + * output only when the dpm feature is enabled. + * We can use Average_* for dpm disabled case. + * But this is available for gfxclk/uclk/socclk. + */ + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_GFXCLK_BIT)) + *value = metrics.CurrClock[PPCLK_GFXCLK]; + else + *value = metrics.AverageGfxclkFrequency; + break; + case PPCLK_UCLK: + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT)) + *value = metrics.CurrClock[PPCLK_UCLK]; + else + *value = metrics.AverageUclkFrequency; + break; + case PPCLK_SOCCLK: + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_SOCCLK_BIT)) + *value = metrics.CurrClock[PPCLK_SOCCLK]; + else + *value = metrics.AverageSocclkFrequency; + break; + default: + *value = metrics.CurrClock[clk_id]; + break; + } + + return ret; +} + +static uint32_t arcturus_find_lowest_dpm_level(struct arcturus_single_dpm_table *table) +{ + uint32_t i; + + for (i = 0; i < table->count; i++) { + if (table->dpm_levels[i].enabled) + break; + } + if (i >= table->count) { + i = 0; + table->dpm_levels[i].enabled = true; + } + + return i; +} + +static uint32_t arcturus_find_highest_dpm_level(struct arcturus_single_dpm_table *table) +{ + int i = 0; + + if (table->count <= 0) { + pr_err("[%s] DPM Table has no entry!", __func__); + return 0; + } + if (table->count > MAX_DPM_NUMBER) { + pr_err("[%s] DPM Table has too many entries!", __func__); + return MAX_DPM_NUMBER - 1; + } + + for (i = table->count - 1; i >= 0; i--) { + if (table->dpm_levels[i].enabled) + break; + } + if (i < 0) { + i = 0; + table->dpm_levels[i].enabled = true; + } + + return i; +} + + + +static int arcturus_force_dpm_limit_value(struct smu_context *smu, bool highest) +{ + struct arcturus_dpm_table *dpm_table = + (struct arcturus_dpm_table *)smu->smu_dpm.dpm_context; + struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(smu->adev, 0); + uint32_t soft_level; + int ret = 0; + + /* gfxclk */ + if (highest) + soft_level = arcturus_find_highest_dpm_level(&(dpm_table->gfx_table)); + else + soft_level = arcturus_find_lowest_dpm_level(&(dpm_table->gfx_table)); + + dpm_table->gfx_table.dpm_state.soft_min_level = + dpm_table->gfx_table.dpm_state.soft_max_level = + dpm_table->gfx_table.dpm_levels[soft_level].value; + + ret = arcturus_upload_dpm_level(smu, false, FEATURE_DPM_GFXCLK_MASK); + if (ret) { + pr_err("Failed to upload boot level to %s!\n", + highest ? "highest" : "lowest"); + return ret; + } + + ret = arcturus_upload_dpm_level(smu, true, FEATURE_DPM_GFXCLK_MASK); + if (ret) { + pr_err("Failed to upload dpm max level to %s!\n!", + highest ? "highest" : "lowest"); + return ret; + } + + if (hive) + /* + * Force XGMI Pstate to highest or lowest + * TODO: revise this when xgmi dpm is functional + */ + ret = smu_v11_0_set_xgmi_pstate(smu, highest ? 1 : 0); + + return ret; +} + +static int arcturus_unforce_dpm_levels(struct smu_context *smu) +{ + struct arcturus_dpm_table *dpm_table = + (struct arcturus_dpm_table *)smu->smu_dpm.dpm_context; + struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(smu->adev, 0); + uint32_t soft_min_level, soft_max_level; + int ret = 0; + + /* gfxclk */ + soft_min_level = arcturus_find_lowest_dpm_level(&(dpm_table->gfx_table)); + soft_max_level = arcturus_find_highest_dpm_level(&(dpm_table->gfx_table)); + dpm_table->gfx_table.dpm_state.soft_min_level = + dpm_table->gfx_table.dpm_levels[soft_min_level].value; + dpm_table->gfx_table.dpm_state.soft_max_level = + dpm_table->gfx_table.dpm_levels[soft_max_level].value; + + ret = arcturus_upload_dpm_level(smu, false, FEATURE_DPM_GFXCLK_MASK); + if (ret) { + pr_err("Failed to upload DPM Bootup Levels!"); + return ret; + } + + ret = arcturus_upload_dpm_level(smu, true, FEATURE_DPM_GFXCLK_MASK); + if (ret) { + pr_err("Failed to upload DPM Max Levels!"); + return ret; + } + + if (hive) + /* + * Reset XGMI Pstate back to default + * TODO: revise this when xgmi dpm is functional + */ + ret = smu_v11_0_set_xgmi_pstate(smu, 0); + + return ret; +} + +static int +arcturus_get_profiling_clk_mask(struct smu_context *smu, + enum amd_dpm_forced_level level, + uint32_t *sclk_mask, + uint32_t *mclk_mask, + uint32_t *soc_mask) +{ + struct arcturus_dpm_table *dpm_table = + (struct arcturus_dpm_table *)smu->smu_dpm.dpm_context; + struct arcturus_single_dpm_table *gfx_dpm_table; + struct arcturus_single_dpm_table *mem_dpm_table; + struct arcturus_single_dpm_table *soc_dpm_table; + + if (!smu->smu_dpm.dpm_context) + return -EINVAL; + + gfx_dpm_table = &dpm_table->gfx_table; + mem_dpm_table = &dpm_table->mem_table; + soc_dpm_table = &dpm_table->soc_table; + + *sclk_mask = 0; + *mclk_mask = 0; + *soc_mask = 0; + + if (gfx_dpm_table->count > ARCTURUS_UMD_PSTATE_GFXCLK_LEVEL && + mem_dpm_table->count > ARCTURUS_UMD_PSTATE_MCLK_LEVEL && + soc_dpm_table->count > ARCTURUS_UMD_PSTATE_SOCCLK_LEVEL) { + *sclk_mask = ARCTURUS_UMD_PSTATE_GFXCLK_LEVEL; + *mclk_mask = ARCTURUS_UMD_PSTATE_MCLK_LEVEL; + *soc_mask = ARCTURUS_UMD_PSTATE_SOCCLK_LEVEL; + } + + if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK) { + *sclk_mask = 0; + } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK) { + *mclk_mask = 0; + } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) { + *sclk_mask = gfx_dpm_table->count - 1; + *mclk_mask = mem_dpm_table->count - 1; + *soc_mask = soc_dpm_table->count - 1; + } + + return 0; +} + +static int arcturus_get_power_limit(struct smu_context *smu, + uint32_t *limit, + bool cap) +{ + PPTable_t *pptable = smu->smu_table.driver_pptable; + uint32_t asic_default_power_limit = 0; + int ret = 0; + int power_src; + + if (!smu->power_limit) { + if (smu_feature_is_enabled(smu, SMU_FEATURE_PPT_BIT)) { + power_src = smu_power_get_index(smu, SMU_POWER_SOURCE_AC); + if (power_src < 0) + return -EINVAL; + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_GetPptLimit, + power_src << 16); + if (ret) { + pr_err("[%s] get PPT limit failed!", __func__); + return ret; + } + smu_read_smc_arg(smu, &asic_default_power_limit); + } else { + /* the last hope to figure out the ppt limit */ + if (!pptable) { + pr_err("Cannot get PPT limit due to pptable missing!"); + return -EINVAL; + } + asic_default_power_limit = + pptable->SocketPowerLimitAc[PPT_THROTTLER_PPT0]; + } + + smu->power_limit = asic_default_power_limit; + } + + if (cap) + *limit = smu_v11_0_get_max_power_limit(smu); + else + *limit = smu->power_limit; + + return 0; +} + +static int arcturus_get_power_profile_mode(struct smu_context *smu, + char *buf) +{ + static const char *profile_name[] = { + "BOOTUP_DEFAULT", + "3D_FULL_SCREEN", + "POWER_SAVING", + "VIDEO", + "VR", + "COMPUTE", + "CUSTOM"}; + uint32_t i, size = 0; + int16_t workload_type = 0; + + if (!smu->pm_enabled || !buf) + return -EINVAL; + + for (i = 0; i <= PP_SMC_POWER_PROFILE_CUSTOM; i++) { + /* + * Conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT + * Not all profile modes are supported on arcturus. + */ + workload_type = smu_workload_get_type(smu, i); + if (workload_type < 0) + continue; + + size += sprintf(buf + size, "%2d %14s%s\n", + i, profile_name[i], (i == smu->power_profile_mode) ? "*" : " "); + } + + return size; +} + +static int arcturus_set_power_profile_mode(struct smu_context *smu, + long *input, + uint32_t size) +{ + int workload_type = 0; + uint32_t profile_mode = input[size]; + int ret = 0; + + if (!smu->pm_enabled) + return -EINVAL; + + if (profile_mode > PP_SMC_POWER_PROFILE_CUSTOM) { + pr_err("Invalid power profile mode %d\n", profile_mode); + return -EINVAL; + } + + /* + * Conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT + * Not all profile modes are supported on arcturus. + */ + workload_type = smu_workload_get_type(smu, profile_mode); + if (workload_type < 0) { + pr_err("Unsupported power profile mode %d on arcturus\n", profile_mode); + return -EINVAL; + } + + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetWorkloadMask, + 1 << workload_type); + if (ret) { + pr_err("Fail to set workload type %d\n", workload_type); + return ret; + } + + smu->power_profile_mode = profile_mode; + + return 0; +} + +static void arcturus_dump_pptable(struct smu_context *smu) +{ + struct smu_table_context *table_context = &smu->smu_table; + PPTable_t *pptable = table_context->driver_pptable; + int i; + + pr_info("Dumped PPTable:\n"); + + pr_info("Version = 0x%08x\n", pptable->Version); + + pr_info("FeaturesToRun[0] = 0x%08x\n", pptable->FeaturesToRun[0]); + pr_info("FeaturesToRun[1] = 0x%08x\n", pptable->FeaturesToRun[1]); + + for (i = 0; i < PPT_THROTTLER_COUNT; i++) { + pr_info("SocketPowerLimitAc[%d] = %d\n", i, pptable->SocketPowerLimitAc[i]); + pr_info("SocketPowerLimitAcTau[%d] = %d\n", i, pptable->SocketPowerLimitAcTau[i]); + } + + pr_info("TdcLimitSoc = %d\n", pptable->TdcLimitSoc); + pr_info("TdcLimitSocTau = %d\n", pptable->TdcLimitSocTau); + pr_info("TdcLimitGfx = %d\n", pptable->TdcLimitGfx); + pr_info("TdcLimitGfxTau = %d\n", pptable->TdcLimitGfxTau); + + pr_info("TedgeLimit = %d\n", pptable->TedgeLimit); + pr_info("ThotspotLimit = %d\n", pptable->ThotspotLimit); + pr_info("TmemLimit = %d\n", pptable->TmemLimit); + pr_info("Tvr_gfxLimit = %d\n", pptable->Tvr_gfxLimit); + pr_info("Tvr_memLimit = %d\n", pptable->Tvr_memLimit); + pr_info("Tvr_socLimit = %d\n", pptable->Tvr_socLimit); + pr_info("FitLimit = %d\n", pptable->FitLimit); + + pr_info("PpmPowerLimit = %d\n", pptable->PpmPowerLimit); + pr_info("PpmTemperatureThreshold = %d\n", pptable->PpmTemperatureThreshold); + + pr_info("ThrottlerControlMask = %d\n", pptable->ThrottlerControlMask); + + pr_info("UlvVoltageOffsetGfx = %d\n", pptable->UlvVoltageOffsetGfx); + pr_info("UlvPadding = 0x%08x\n", pptable->UlvPadding); + + pr_info("UlvGfxclkBypass = %d\n", pptable->UlvGfxclkBypass); + pr_info("Padding234[0] = 0x%02x\n", pptable->Padding234[0]); + pr_info("Padding234[1] = 0x%02x\n", pptable->Padding234[1]); + pr_info("Padding234[2] = 0x%02x\n", pptable->Padding234[2]); + + pr_info("MinVoltageGfx = %d\n", pptable->MinVoltageGfx); + pr_info("MinVoltageSoc = %d\n", pptable->MinVoltageSoc); + pr_info("MaxVoltageGfx = %d\n", pptable->MaxVoltageGfx); + pr_info("MaxVoltageSoc = %d\n", pptable->MaxVoltageSoc); + + pr_info("LoadLineResistanceGfx = %d\n", pptable->LoadLineResistanceGfx); + pr_info("LoadLineResistanceSoc = %d\n", pptable->LoadLineResistanceSoc); + + pr_info("[PPCLK_GFXCLK]\n" + " .VoltageMode = 0x%02x\n" + " .SnapToDiscrete = 0x%02x\n" + " .NumDiscreteLevels = 0x%02x\n" + " .padding = 0x%02x\n" + " .ConversionToAvfsClk{m = 0x%08x b = 0x%08x}\n" + " .SsCurve {a = 0x%08x b = 0x%08x c = 0x%08x}\n" + " .SsFmin = 0x%04x\n" + " .Padding_16 = 0x%04x\n", + pptable->DpmDescriptor[PPCLK_GFXCLK].VoltageMode, + pptable->DpmDescriptor[PPCLK_GFXCLK].SnapToDiscrete, + pptable->DpmDescriptor[PPCLK_GFXCLK].NumDiscreteLevels, + pptable->DpmDescriptor[PPCLK_GFXCLK].padding, + pptable->DpmDescriptor[PPCLK_GFXCLK].ConversionToAvfsClk.m, + pptable->DpmDescriptor[PPCLK_GFXCLK].ConversionToAvfsClk.b, + pptable->DpmDescriptor[PPCLK_GFXCLK].SsCurve.a, + pptable->DpmDescriptor[PPCLK_GFXCLK].SsCurve.b, + pptable->DpmDescriptor[PPCLK_GFXCLK].SsCurve.c, + pptable->DpmDescriptor[PPCLK_GFXCLK].SsFmin, + pptable->DpmDescriptor[PPCLK_GFXCLK].Padding16); + + pr_info("[PPCLK_VCLK]\n" + " .VoltageMode = 0x%02x\n" + " .SnapToDiscrete = 0x%02x\n" + " .NumDiscreteLevels = 0x%02x\n" + " .padding = 0x%02x\n" + " .ConversionToAvfsClk{m = 0x%08x b = 0x%08x}\n" + " .SsCurve {a = 0x%08x b = 0x%08x c = 0x%08x}\n" + " .SsFmin = 0x%04x\n" + " .Padding_16 = 0x%04x\n", + pptable->DpmDescriptor[PPCLK_VCLK].VoltageMode, + pptable->DpmDescriptor[PPCLK_VCLK].SnapToDiscrete, + pptable->DpmDescriptor[PPCLK_VCLK].NumDiscreteLevels, + pptable->DpmDescriptor[PPCLK_VCLK].padding, + pptable->DpmDescriptor[PPCLK_VCLK].ConversionToAvfsClk.m, + pptable->DpmDescriptor[PPCLK_VCLK].ConversionToAvfsClk.b, + pptable->DpmDescriptor[PPCLK_VCLK].SsCurve.a, + pptable->DpmDescriptor[PPCLK_VCLK].SsCurve.b, + pptable->DpmDescriptor[PPCLK_VCLK].SsCurve.c, + pptable->DpmDescriptor[PPCLK_VCLK].SsFmin, + pptable->DpmDescriptor[PPCLK_VCLK].Padding16); + + pr_info("[PPCLK_DCLK]\n" + " .VoltageMode = 0x%02x\n" + " .SnapToDiscrete = 0x%02x\n" + " .NumDiscreteLevels = 0x%02x\n" + " .padding = 0x%02x\n" + " .ConversionToAvfsClk{m = 0x%08x b = 0x%08x}\n" + " .SsCurve {a = 0x%08x b = 0x%08x c = 0x%08x}\n" + " .SsFmin = 0x%04x\n" + " .Padding_16 = 0x%04x\n", + pptable->DpmDescriptor[PPCLK_DCLK].VoltageMode, + pptable->DpmDescriptor[PPCLK_DCLK].SnapToDiscrete, + pptable->DpmDescriptor[PPCLK_DCLK].NumDiscreteLevels, + pptable->DpmDescriptor[PPCLK_DCLK].padding, + pptable->DpmDescriptor[PPCLK_DCLK].ConversionToAvfsClk.m, + pptable->DpmDescriptor[PPCLK_DCLK].ConversionToAvfsClk.b, + pptable->DpmDescriptor[PPCLK_DCLK].SsCurve.a, + pptable->DpmDescriptor[PPCLK_DCLK].SsCurve.b, + pptable->DpmDescriptor[PPCLK_DCLK].SsCurve.c, + pptable->DpmDescriptor[PPCLK_DCLK].SsFmin, + pptable->DpmDescriptor[PPCLK_DCLK].Padding16); + + pr_info("[PPCLK_SOCCLK]\n" + " .VoltageMode = 0x%02x\n" + " .SnapToDiscrete = 0x%02x\n" + " .NumDiscreteLevels = 0x%02x\n" + " .padding = 0x%02x\n" + " .ConversionToAvfsClk{m = 0x%08x b = 0x%08x}\n" + " .SsCurve {a = 0x%08x b = 0x%08x c = 0x%08x}\n" + " .SsFmin = 0x%04x\n" + " .Padding_16 = 0x%04x\n", + pptable->DpmDescriptor[PPCLK_SOCCLK].VoltageMode, + pptable->DpmDescriptor[PPCLK_SOCCLK].SnapToDiscrete, + pptable->DpmDescriptor[PPCLK_SOCCLK].NumDiscreteLevels, + pptable->DpmDescriptor[PPCLK_SOCCLK].padding, + pptable->DpmDescriptor[PPCLK_SOCCLK].ConversionToAvfsClk.m, + pptable->DpmDescriptor[PPCLK_SOCCLK].ConversionToAvfsClk.b, + pptable->DpmDescriptor[PPCLK_SOCCLK].SsCurve.a, + pptable->DpmDescriptor[PPCLK_SOCCLK].SsCurve.b, + pptable->DpmDescriptor[PPCLK_SOCCLK].SsCurve.c, + pptable->DpmDescriptor[PPCLK_SOCCLK].SsFmin, + pptable->DpmDescriptor[PPCLK_SOCCLK].Padding16); + + pr_info("[PPCLK_UCLK]\n" + " .VoltageMode = 0x%02x\n" + " .SnapToDiscrete = 0x%02x\n" + " .NumDiscreteLevels = 0x%02x\n" + " .padding = 0x%02x\n" + " .ConversionToAvfsClk{m = 0x%08x b = 0x%08x}\n" + " .SsCurve {a = 0x%08x b = 0x%08x c = 0x%08x}\n" + " .SsFmin = 0x%04x\n" + " .Padding_16 = 0x%04x\n", + pptable->DpmDescriptor[PPCLK_UCLK].VoltageMode, + pptable->DpmDescriptor[PPCLK_UCLK].SnapToDiscrete, + pptable->DpmDescriptor[PPCLK_UCLK].NumDiscreteLevels, + pptable->DpmDescriptor[PPCLK_UCLK].padding, + pptable->DpmDescriptor[PPCLK_UCLK].ConversionToAvfsClk.m, + pptable->DpmDescriptor[PPCLK_UCLK].ConversionToAvfsClk.b, + pptable->DpmDescriptor[PPCLK_UCLK].SsCurve.a, + pptable->DpmDescriptor[PPCLK_UCLK].SsCurve.b, + pptable->DpmDescriptor[PPCLK_UCLK].SsCurve.c, + pptable->DpmDescriptor[PPCLK_UCLK].SsFmin, + pptable->DpmDescriptor[PPCLK_UCLK].Padding16); + + pr_info("[PPCLK_FCLK]\n" + " .VoltageMode = 0x%02x\n" + " .SnapToDiscrete = 0x%02x\n" + " .NumDiscreteLevels = 0x%02x\n" + " .padding = 0x%02x\n" + " .ConversionToAvfsClk{m = 0x%08x b = 0x%08x}\n" + " .SsCurve {a = 0x%08x b = 0x%08x c = 0x%08x}\n" + " .SsFmin = 0x%04x\n" + " .Padding_16 = 0x%04x\n", + pptable->DpmDescriptor[PPCLK_FCLK].VoltageMode, + pptable->DpmDescriptor[PPCLK_FCLK].SnapToDiscrete, + pptable->DpmDescriptor[PPCLK_FCLK].NumDiscreteLevels, + pptable->DpmDescriptor[PPCLK_FCLK].padding, + pptable->DpmDescriptor[PPCLK_FCLK].ConversionToAvfsClk.m, + pptable->DpmDescriptor[PPCLK_FCLK].ConversionToAvfsClk.b, + pptable->DpmDescriptor[PPCLK_FCLK].SsCurve.a, + pptable->DpmDescriptor[PPCLK_FCLK].SsCurve.b, + pptable->DpmDescriptor[PPCLK_FCLK].SsCurve.c, + pptable->DpmDescriptor[PPCLK_FCLK].SsFmin, + pptable->DpmDescriptor[PPCLK_FCLK].Padding16); + + + pr_info("FreqTableGfx\n"); + for (i = 0; i < NUM_GFXCLK_DPM_LEVELS; i++) + pr_info(" .[%02d] = %d\n", i, pptable->FreqTableGfx[i]); + + pr_info("FreqTableVclk\n"); + for (i = 0; i < NUM_VCLK_DPM_LEVELS; i++) + pr_info(" .[%02d] = %d\n", i, pptable->FreqTableVclk[i]); + + pr_info("FreqTableDclk\n"); + for (i = 0; i < NUM_DCLK_DPM_LEVELS; i++) + pr_info(" .[%02d] = %d\n", i, pptable->FreqTableDclk[i]); + + pr_info("FreqTableSocclk\n"); + for (i = 0; i < NUM_SOCCLK_DPM_LEVELS; i++) + pr_info(" .[%02d] = %d\n", i, pptable->FreqTableSocclk[i]); + + pr_info("FreqTableUclk\n"); + for (i = 0; i < NUM_UCLK_DPM_LEVELS; i++) + pr_info(" .[%02d] = %d\n", i, pptable->FreqTableUclk[i]); + + pr_info("FreqTableFclk\n"); + for (i = 0; i < NUM_FCLK_DPM_LEVELS; i++) + pr_info(" .[%02d] = %d\n", i, pptable->FreqTableFclk[i]); + + pr_info("Mp0clkFreq\n"); + for (i = 0; i < NUM_MP0CLK_DPM_LEVELS; i++) + pr_info(" .[%d] = %d\n", i, pptable->Mp0clkFreq[i]); + + pr_info("Mp0DpmVoltage\n"); + for (i = 0; i < NUM_MP0CLK_DPM_LEVELS; i++) + pr_info(" .[%d] = %d\n", i, pptable->Mp0DpmVoltage[i]); + + pr_info("GfxclkFidle = 0x%x\n", pptable->GfxclkFidle); + pr_info("GfxclkSlewRate = 0x%x\n", pptable->GfxclkSlewRate); + pr_info("Padding567[0] = 0x%x\n", pptable->Padding567[0]); + pr_info("Padding567[1] = 0x%x\n", pptable->Padding567[1]); + pr_info("Padding567[2] = 0x%x\n", pptable->Padding567[2]); + pr_info("Padding567[3] = 0x%x\n", pptable->Padding567[3]); + pr_info("GfxclkDsMaxFreq = %d\n", pptable->GfxclkDsMaxFreq); + pr_info("GfxclkSource = 0x%x\n", pptable->GfxclkSource); + pr_info("Padding456 = 0x%x\n", pptable->Padding456); + + pr_info("EnableTdpm = %d\n", pptable->EnableTdpm); + pr_info("TdpmHighHystTemperature = %d\n", pptable->TdpmHighHystTemperature); + pr_info("TdpmLowHystTemperature = %d\n", pptable->TdpmLowHystTemperature); + pr_info("GfxclkFreqHighTempLimit = %d\n", pptable->GfxclkFreqHighTempLimit); + + pr_info("FanStopTemp = %d\n", pptable->FanStopTemp); + pr_info("FanStartTemp = %d\n", pptable->FanStartTemp); + + pr_info("FanGainEdge = %d\n", pptable->FanGainEdge); + pr_info("FanGainHotspot = %d\n", pptable->FanGainHotspot); + pr_info("FanGainVrGfx = %d\n", pptable->FanGainVrGfx); + pr_info("FanGainVrSoc = %d\n", pptable->FanGainVrSoc); + pr_info("FanGainVrMem = %d\n", pptable->FanGainVrMem); + pr_info("FanGainHbm = %d\n", pptable->FanGainHbm); + + pr_info("FanPwmMin = %d\n", pptable->FanPwmMin); + pr_info("FanAcousticLimitRpm = %d\n", pptable->FanAcousticLimitRpm); + pr_info("FanThrottlingRpm = %d\n", pptable->FanThrottlingRpm); + pr_info("FanMaximumRpm = %d\n", pptable->FanMaximumRpm); + pr_info("FanTargetTemperature = %d\n", pptable->FanTargetTemperature); + pr_info("FanTargetGfxclk = %d\n", pptable->FanTargetGfxclk); + pr_info("FanZeroRpmEnable = %d\n", pptable->FanZeroRpmEnable); + pr_info("FanTachEdgePerRev = %d\n", pptable->FanTachEdgePerRev); + pr_info("FanTempInputSelect = %d\n", pptable->FanTempInputSelect); + + pr_info("FuzzyFan_ErrorSetDelta = %d\n", pptable->FuzzyFan_ErrorSetDelta); + pr_info("FuzzyFan_ErrorRateSetDelta = %d\n", pptable->FuzzyFan_ErrorRateSetDelta); + pr_info("FuzzyFan_PwmSetDelta = %d\n", pptable->FuzzyFan_PwmSetDelta); + pr_info("FuzzyFan_Reserved = %d\n", pptable->FuzzyFan_Reserved); + + pr_info("OverrideAvfsGb[AVFS_VOLTAGE_GFX] = 0x%x\n", pptable->OverrideAvfsGb[AVFS_VOLTAGE_GFX]); + pr_info("OverrideAvfsGb[AVFS_VOLTAGE_SOC] = 0x%x\n", pptable->OverrideAvfsGb[AVFS_VOLTAGE_SOC]); + pr_info("Padding8_Avfs[0] = %d\n", pptable->Padding8_Avfs[0]); + pr_info("Padding8_Avfs[1] = %d\n", pptable->Padding8_Avfs[1]); + + pr_info("dBtcGbGfxPll{a = 0x%x b = 0x%x c = 0x%x}\n", + pptable->dBtcGbGfxPll.a, + pptable->dBtcGbGfxPll.b, + pptable->dBtcGbGfxPll.c); + pr_info("dBtcGbGfxAfll{a = 0x%x b = 0x%x c = 0x%x}\n", + pptable->dBtcGbGfxAfll.a, + pptable->dBtcGbGfxAfll.b, + pptable->dBtcGbGfxAfll.c); + pr_info("dBtcGbSoc{a = 0x%x b = 0x%x c = 0x%x}\n", + pptable->dBtcGbSoc.a, + pptable->dBtcGbSoc.b, + pptable->dBtcGbSoc.c); + + pr_info("qAgingGb[AVFS_VOLTAGE_GFX]{m = 0x%x b = 0x%x}\n", + pptable->qAgingGb[AVFS_VOLTAGE_GFX].m, + pptable->qAgingGb[AVFS_VOLTAGE_GFX].b); + pr_info("qAgingGb[AVFS_VOLTAGE_SOC]{m = 0x%x b = 0x%x}\n", + pptable->qAgingGb[AVFS_VOLTAGE_SOC].m, + pptable->qAgingGb[AVFS_VOLTAGE_SOC].b); + + pr_info("qStaticVoltageOffset[AVFS_VOLTAGE_GFX]{a = 0x%x b = 0x%x c = 0x%x}\n", + pptable->qStaticVoltageOffset[AVFS_VOLTAGE_GFX].a, + pptable->qStaticVoltageOffset[AVFS_VOLTAGE_GFX].b, + pptable->qStaticVoltageOffset[AVFS_VOLTAGE_GFX].c); + pr_info("qStaticVoltageOffset[AVFS_VOLTAGE_SOC]{a = 0x%x b = 0x%x c = 0x%x}\n", + pptable->qStaticVoltageOffset[AVFS_VOLTAGE_SOC].a, + pptable->qStaticVoltageOffset[AVFS_VOLTAGE_SOC].b, + pptable->qStaticVoltageOffset[AVFS_VOLTAGE_SOC].c); + + pr_info("DcTol[AVFS_VOLTAGE_GFX] = 0x%x\n", pptable->DcTol[AVFS_VOLTAGE_GFX]); + pr_info("DcTol[AVFS_VOLTAGE_SOC] = 0x%x\n", pptable->DcTol[AVFS_VOLTAGE_SOC]); + + pr_info("DcBtcEnabled[AVFS_VOLTAGE_GFX] = 0x%x\n", pptable->DcBtcEnabled[AVFS_VOLTAGE_GFX]); + pr_info("DcBtcEnabled[AVFS_VOLTAGE_SOC] = 0x%x\n", pptable->DcBtcEnabled[AVFS_VOLTAGE_SOC]); + pr_info("Padding8_GfxBtc[0] = 0x%x\n", pptable->Padding8_GfxBtc[0]); + pr_info("Padding8_GfxBtc[1] = 0x%x\n", pptable->Padding8_GfxBtc[1]); + + pr_info("DcBtcMin[AVFS_VOLTAGE_GFX] = 0x%x\n", pptable->DcBtcMin[AVFS_VOLTAGE_GFX]); + pr_info("DcBtcMin[AVFS_VOLTAGE_SOC] = 0x%x\n", pptable->DcBtcMin[AVFS_VOLTAGE_SOC]); + pr_info("DcBtcMax[AVFS_VOLTAGE_GFX] = 0x%x\n", pptable->DcBtcMax[AVFS_VOLTAGE_GFX]); + pr_info("DcBtcMax[AVFS_VOLTAGE_SOC] = 0x%x\n", pptable->DcBtcMax[AVFS_VOLTAGE_SOC]); + + pr_info("DcBtcGb[AVFS_VOLTAGE_GFX] = 0x%x\n", pptable->DcBtcGb[AVFS_VOLTAGE_GFX]); + pr_info("DcBtcGb[AVFS_VOLTAGE_SOC] = 0x%x\n", pptable->DcBtcGb[AVFS_VOLTAGE_SOC]); + + pr_info("XgmiDpmPstates\n"); + for (i = 0; i < NUM_XGMI_LEVELS; i++) + pr_info(" .[%d] = %d\n", i, pptable->XgmiDpmPstates[i]); + pr_info("XgmiDpmSpare[0] = 0x%02x\n", pptable->XgmiDpmSpare[0]); + pr_info("XgmiDpmSpare[1] = 0x%02x\n", pptable->XgmiDpmSpare[1]); + + pr_info("VDDGFX_TVmin = %d\n", pptable->VDDGFX_TVmin); + pr_info("VDDSOC_TVmin = %d\n", pptable->VDDSOC_TVmin); + pr_info("VDDGFX_Vmin_HiTemp = %d\n", pptable->VDDGFX_Vmin_HiTemp); + pr_info("VDDGFX_Vmin_LoTemp = %d\n", pptable->VDDGFX_Vmin_LoTemp); + pr_info("VDDSOC_Vmin_HiTemp = %d\n", pptable->VDDSOC_Vmin_HiTemp); + pr_info("VDDSOC_Vmin_LoTemp = %d\n", pptable->VDDSOC_Vmin_LoTemp); + pr_info("VDDGFX_TVminHystersis = %d\n", pptable->VDDGFX_TVminHystersis); + pr_info("VDDSOC_TVminHystersis = %d\n", pptable->VDDSOC_TVminHystersis); + + pr_info("DebugOverrides = 0x%x\n", pptable->DebugOverrides); + pr_info("ReservedEquation0{a = 0x%x b = 0x%x c = 0x%x}\n", + pptable->ReservedEquation0.a, + pptable->ReservedEquation0.b, + pptable->ReservedEquation0.c); + pr_info("ReservedEquation1{a = 0x%x b = 0x%x c = 0x%x}\n", + pptable->ReservedEquation1.a, + pptable->ReservedEquation1.b, + pptable->ReservedEquation1.c); + pr_info("ReservedEquation2{a = 0x%x b = 0x%x c = 0x%x}\n", + pptable->ReservedEquation2.a, + pptable->ReservedEquation2.b, + pptable->ReservedEquation2.c); + pr_info("ReservedEquation3{a = 0x%x b = 0x%x c = 0x%x}\n", + pptable->ReservedEquation3.a, + pptable->ReservedEquation3.b, + pptable->ReservedEquation3.c); + + pr_info("MinVoltageUlvGfx = %d\n", pptable->MinVoltageUlvGfx); + pr_info("PaddingUlv = %d\n", pptable->PaddingUlv); + + pr_info("TotalPowerConfig = %d\n", pptable->TotalPowerConfig); + pr_info("TotalPowerSpare1 = %d\n", pptable->TotalPowerSpare1); + pr_info("TotalPowerSpare2 = %d\n", pptable->TotalPowerSpare2); + + pr_info("PccThresholdLow = %d\n", pptable->PccThresholdLow); + pr_info("PccThresholdHigh = %d\n", pptable->PccThresholdHigh); + + pr_info("Board Parameters:\n"); + pr_info("MaxVoltageStepGfx = 0x%x\n", pptable->MaxVoltageStepGfx); + pr_info("MaxVoltageStepSoc = 0x%x\n", pptable->MaxVoltageStepSoc); + + pr_info("VddGfxVrMapping = 0x%x\n", pptable->VddGfxVrMapping); + pr_info("VddSocVrMapping = 0x%x\n", pptable->VddSocVrMapping); + pr_info("VddMemVrMapping = 0x%x\n", pptable->VddMemVrMapping); + pr_info("BoardVrMapping = 0x%x\n", pptable->BoardVrMapping); + + pr_info("GfxUlvPhaseSheddingMask = 0x%x\n", pptable->GfxUlvPhaseSheddingMask); + pr_info("ExternalSensorPresent = 0x%x\n", pptable->ExternalSensorPresent); + + pr_info("GfxMaxCurrent = 0x%x\n", pptable->GfxMaxCurrent); + pr_info("GfxOffset = 0x%x\n", pptable->GfxOffset); + pr_info("Padding_TelemetryGfx = 0x%x\n", pptable->Padding_TelemetryGfx); + + pr_info("SocMaxCurrent = 0x%x\n", pptable->SocMaxCurrent); + pr_info("SocOffset = 0x%x\n", pptable->SocOffset); + pr_info("Padding_TelemetrySoc = 0x%x\n", pptable->Padding_TelemetrySoc); + + pr_info("MemMaxCurrent = 0x%x\n", pptable->MemMaxCurrent); + pr_info("MemOffset = 0x%x\n", pptable->MemOffset); + pr_info("Padding_TelemetryMem = 0x%x\n", pptable->Padding_TelemetryMem); + + pr_info("BoardMaxCurrent = 0x%x\n", pptable->BoardMaxCurrent); + pr_info("BoardOffset = 0x%x\n", pptable->BoardOffset); + pr_info("Padding_TelemetryBoardInput = 0x%x\n", pptable->Padding_TelemetryBoardInput); + + pr_info("VR0HotGpio = %d\n", pptable->VR0HotGpio); + pr_info("VR0HotPolarity = %d\n", pptable->VR0HotPolarity); + pr_info("VR1HotGpio = %d\n", pptable->VR1HotGpio); + pr_info("VR1HotPolarity = %d\n", pptable->VR1HotPolarity); + + pr_info("PllGfxclkSpreadEnabled = %d\n", pptable->PllGfxclkSpreadEnabled); + pr_info("PllGfxclkSpreadPercent = %d\n", pptable->PllGfxclkSpreadPercent); + pr_info("PllGfxclkSpreadFreq = %d\n", pptable->PllGfxclkSpreadFreq); + + pr_info("UclkSpreadEnabled = %d\n", pptable->UclkSpreadEnabled); + pr_info("UclkSpreadPercent = %d\n", pptable->UclkSpreadPercent); + pr_info("UclkSpreadFreq = %d\n", pptable->UclkSpreadFreq); + + pr_info("FclkSpreadEnabled = %d\n", pptable->FclkSpreadEnabled); + pr_info("FclkSpreadPercent = %d\n", pptable->FclkSpreadPercent); + pr_info("FclkSpreadFreq = %d\n", pptable->FclkSpreadFreq); + + pr_info("FllGfxclkSpreadEnabled = %d\n", pptable->FllGfxclkSpreadEnabled); + pr_info("FllGfxclkSpreadPercent = %d\n", pptable->FllGfxclkSpreadPercent); + pr_info("FllGfxclkSpreadFreq = %d\n", pptable->FllGfxclkSpreadFreq); + + for (i = 0; i < NUM_I2C_CONTROLLERS; i++) { + pr_info("I2cControllers[%d]:\n", i); + pr_info(" .Enabled = %d\n", + pptable->I2cControllers[i].Enabled); + pr_info(" .SlaveAddress = 0x%x\n", + pptable->I2cControllers[i].SlaveAddress); + pr_info(" .ControllerPort = %d\n", + pptable->I2cControllers[i].ControllerPort); + pr_info(" .ControllerName = %d\n", + pptable->I2cControllers[i].ControllerName); + pr_info(" .ThermalThrottler = %d\n", + pptable->I2cControllers[i].ThermalThrotter); + pr_info(" .I2cProtocol = %d\n", + pptable->I2cControllers[i].I2cProtocol); + pr_info(" .Speed = %d\n", + pptable->I2cControllers[i].Speed); + } + + pr_info("MemoryChannelEnabled = %d\n", pptable->MemoryChannelEnabled); + pr_info("DramBitWidth = %d\n", pptable->DramBitWidth); + + pr_info("TotalBoardPower = %d\n", pptable->TotalBoardPower); + + pr_info("XgmiLinkSpeed\n"); + for (i = 0; i < NUM_XGMI_PSTATE_LEVELS; i++) + pr_info(" .[%d] = %d\n", i, pptable->XgmiLinkSpeed[i]); + pr_info("XgmiLinkWidth\n"); + for (i = 0; i < NUM_XGMI_PSTATE_LEVELS; i++) + pr_info(" .[%d] = %d\n", i, pptable->XgmiLinkWidth[i]); + pr_info("XgmiFclkFreq\n"); + for (i = 0; i < NUM_XGMI_PSTATE_LEVELS; i++) + pr_info(" .[%d] = %d\n", i, pptable->XgmiFclkFreq[i]); + pr_info("XgmiSocVoltage\n"); + for (i = 0; i < NUM_XGMI_PSTATE_LEVELS; i++) + pr_info(" .[%d] = %d\n", i, pptable->XgmiSocVoltage[i]); + +} + +static bool arcturus_is_dpm_running(struct smu_context *smu) +{ + int ret = 0; + uint32_t feature_mask[2]; + unsigned long feature_enabled; + ret = smu_feature_get_enabled_mask(smu, feature_mask, 2); + feature_enabled = (unsigned long)((uint64_t)feature_mask[0] | + ((uint64_t)feature_mask[1] << 32)); + return !!(feature_enabled & SMC_DPM_FEATURE); +} + +static int arcturus_dpm_set_uvd_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_VCN_PG_BIT)) { + ret = smu_feature_set_enabled(smu, SMU_FEATURE_VCN_PG_BIT, 1); + if (ret) { + pr_err("[EnableVCNDPM] failed!\n"); + return ret; + } + } + power_gate->vcn_gated = false; + } else { + if (smu_feature_is_enabled(smu, SMU_FEATURE_VCN_PG_BIT)) { + ret = smu_feature_set_enabled(smu, SMU_FEATURE_VCN_PG_BIT, 0); + if (ret) { + pr_err("[DisableVCNDPM] failed!\n"); + return ret; + } + } + power_gate->vcn_gated = true; + } + + return ret; +} + + +static void arcturus_fill_eeprom_i2c_req(SwI2cRequest_t *req, bool write, + uint8_t address, uint32_t numbytes, + uint8_t *data) +{ + int i; + + BUG_ON(numbytes > MAX_SW_I2C_COMMANDS); + + req->I2CcontrollerPort = 0; + req->I2CSpeed = 2; + req->SlaveAddress = address; + req->NumCmds = numbytes; + + for (i = 0; i < numbytes; i++) { + SwI2cCmd_t *cmd = &req->SwI2cCmds[i]; + + /* First 2 bytes are always write for lower 2b EEPROM address */ + if (i < 2) + cmd->Cmd = 1; + else + cmd->Cmd = write; + + + /* Add RESTART for read after address filled */ + cmd->CmdConfig |= (i == 2 && !write) ? CMDCONFIG_RESTART_MASK : 0; + + /* Add STOP in the end */ + cmd->CmdConfig |= (i == (numbytes - 1)) ? CMDCONFIG_STOP_MASK : 0; + + /* Fill with data regardless if read or write to simplify code */ + cmd->RegisterAddr = data[i]; + } +} + +static int arcturus_i2c_eeprom_read_data(struct i2c_adapter *control, + uint8_t address, + uint8_t *data, + uint32_t numbytes) +{ + uint32_t i, ret = 0; + 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]; + + memset(&req, 0, sizeof(req)); + arcturus_fill_eeprom_i2c_req(&req, false, address, numbytes, data); + + mutex_lock(&adev->smu.mutex); + /* Now read data starting with that address */ + ret = smu_update_table(&adev->smu, SMU_TABLE_I2C_COMMANDS, 0, &req, + true); + mutex_unlock(&adev->smu.mutex); + + if (!ret) { + SwI2cRequest_t *res = (SwI2cRequest_t *)table->cpu_addr; + + /* Assume SMU fills res.SwI2cCmds[i].Data with read bytes */ + for (i = 0; i < numbytes; i++) + data[i] = res->SwI2cCmds[i].Data; + + pr_debug("arcturus_i2c_eeprom_read_data, address = %x, bytes = %d, data :", + (uint16_t)address, numbytes); + + print_hex_dump(KERN_DEBUG, "data: ", DUMP_PREFIX_NONE, + 8, 1, data, numbytes, false); + } else + pr_err("arcturus_i2c_eeprom_read_data - error occurred :%x", ret); + + return ret; +} + +static int arcturus_i2c_eeprom_write_data(struct i2c_adapter *control, + uint8_t address, + uint8_t *data, + uint32_t numbytes) +{ + uint32_t ret; + SwI2cRequest_t req; + struct amdgpu_device *adev = to_amdgpu_device(control); + + memset(&req, 0, sizeof(req)); + arcturus_fill_eeprom_i2c_req(&req, true, address, numbytes, data); + + mutex_lock(&adev->smu.mutex); + ret = smu_update_table(&adev->smu, SMU_TABLE_I2C_COMMANDS, 0, &req, true); + mutex_unlock(&adev->smu.mutex); + + if (!ret) { + pr_debug("arcturus_i2c_write(), address = %x, bytes = %d , data: ", + (uint16_t)address, numbytes); + + print_hex_dump(KERN_DEBUG, "data: ", DUMP_PREFIX_NONE, + 8, 1, data, numbytes, false); + /* + * According to EEPROM spec there is a MAX of 10 ms required for + * EEPROM to flush internal RX buffer after STOP was issued at the + * end of write transaction. During this time the EEPROM will not be + * responsive to any more commands - so wait a bit more. + */ + msleep(10); + + } else + pr_err("arcturus_i2c_write- error occurred :%x", ret); + + return ret; +} + +static int arcturus_i2c_eeprom_i2c_xfer(struct i2c_adapter *i2c_adap, + struct i2c_msg *msgs, int num) +{ + uint32_t i, j, ret, data_size, data_chunk_size, next_eeprom_addr = 0; + uint8_t *data_ptr, data_chunk[MAX_SW_I2C_COMMANDS] = { 0 }; + + for (i = 0; i < num; i++) { + /* + * SMU interface allows at most MAX_SW_I2C_COMMANDS bytes of data at + * once and hence the data needs to be spliced into chunks and sent each + * chunk separately + */ + data_size = msgs[i].len - 2; + data_chunk_size = MAX_SW_I2C_COMMANDS - 2; + next_eeprom_addr = (msgs[i].buf[0] << 8 & 0xff00) | (msgs[i].buf[1] & 0xff); + data_ptr = msgs[i].buf + 2; + + for (j = 0; j < data_size / data_chunk_size; j++) { + /* Insert the EEPROM dest addess, bits 0-15 */ + data_chunk[0] = ((next_eeprom_addr >> 8) & 0xff); + data_chunk[1] = (next_eeprom_addr & 0xff); + + if (msgs[i].flags & I2C_M_RD) { + ret = arcturus_i2c_eeprom_read_data(i2c_adap, + (uint8_t)msgs[i].addr, + data_chunk, MAX_SW_I2C_COMMANDS); + + memcpy(data_ptr, data_chunk + 2, data_chunk_size); + } else { + + memcpy(data_chunk + 2, data_ptr, data_chunk_size); + + ret = arcturus_i2c_eeprom_write_data(i2c_adap, + (uint8_t)msgs[i].addr, + data_chunk, MAX_SW_I2C_COMMANDS); + } + + if (ret) { + num = -EIO; + goto fail; + } + + next_eeprom_addr += data_chunk_size; + data_ptr += data_chunk_size; + } + + if (data_size % data_chunk_size) { + data_chunk[0] = ((next_eeprom_addr >> 8) & 0xff); + data_chunk[1] = (next_eeprom_addr & 0xff); + + if (msgs[i].flags & I2C_M_RD) { + ret = arcturus_i2c_eeprom_read_data(i2c_adap, + (uint8_t)msgs[i].addr, + data_chunk, (data_size % data_chunk_size) + 2); + + memcpy(data_ptr, data_chunk + 2, data_size % data_chunk_size); + } else { + memcpy(data_chunk + 2, data_ptr, data_size % data_chunk_size); + + ret = arcturus_i2c_eeprom_write_data(i2c_adap, + (uint8_t)msgs[i].addr, + data_chunk, (data_size % data_chunk_size) + 2); + } + + if (ret) { + num = -EIO; + goto fail; + } + } + } + +fail: + return num; +} + +static u32 arcturus_i2c_eeprom_i2c_func(struct i2c_adapter *adap) +{ + return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL; +} + + +static const struct i2c_algorithm arcturus_i2c_eeprom_i2c_algo = { + .master_xfer = arcturus_i2c_eeprom_i2c_xfer, + .functionality = arcturus_i2c_eeprom_i2c_func, +}; + +static int arcturus_i2c_eeprom_control_init(struct i2c_adapter *control) +{ + struct amdgpu_device *adev = to_amdgpu_device(control); + int res; + + control->owner = THIS_MODULE; + control->class = I2C_CLASS_SPD; + control->dev.parent = &adev->pdev->dev; + control->algo = &arcturus_i2c_eeprom_i2c_algo; + snprintf(control->name, sizeof(control->name), "RAS EEPROM"); + + res = i2c_add_adapter(control); + if (res) + DRM_ERROR("Failed to register hw i2c, err: %d\n", res); + + return res; +} + +static void arcturus_i2c_eeprom_control_fini(struct i2c_adapter *control) +{ + i2c_del_adapter(control); +} + +static uint32_t arcturus_get_pptable_power_limit(struct smu_context *smu) +{ + PPTable_t *pptable = smu->smu_table.driver_pptable; + + return pptable->SocketPowerLimitAc[PPT_THROTTLER_PPT0]; +} + +static const struct pptable_funcs arcturus_ppt_funcs = { + /* translate smu index into arcturus specific index */ + .get_smu_msg_index = arcturus_get_smu_msg_index, + .get_smu_clk_index = arcturus_get_smu_clk_index, + .get_smu_feature_index = arcturus_get_smu_feature_index, + .get_smu_table_index = arcturus_get_smu_table_index, + .get_smu_power_index= arcturus_get_pwr_src_index, + .get_workload_type = arcturus_get_workload_type, + /* internal structurs allocations */ + .tables_init = arcturus_tables_init, + .alloc_dpm_context = arcturus_allocate_dpm_context, + /* pptable related */ + .check_powerplay_table = arcturus_check_powerplay_table, + .store_powerplay_table = arcturus_store_powerplay_table, + .append_powerplay_table = arcturus_append_powerplay_table, + /* init dpm */ + .get_allowed_feature_mask = arcturus_get_allowed_feature_mask, + /* btc */ + .run_btc = arcturus_run_btc, + /* dpm/clk tables */ + .set_default_dpm_table = arcturus_set_default_dpm_table, + .populate_umd_state_clk = arcturus_populate_umd_state_clk, + .get_thermal_temperature_range = arcturus_get_thermal_temperature_range, + .get_current_clk_freq_by_table = arcturus_get_current_clk_freq_by_table, + .print_clk_levels = arcturus_print_clk_levels, + .force_clk_levels = arcturus_force_clk_levels, + .read_sensor = arcturus_read_sensor, + .get_fan_speed_percent = arcturus_get_fan_speed_percent, + .get_fan_speed_rpm = arcturus_get_fan_speed_rpm, + .force_dpm_limit_value = arcturus_force_dpm_limit_value, + .unforce_dpm_levels = arcturus_unforce_dpm_levels, + .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, + /* debug (internal used) */ + .dump_pptable = arcturus_dump_pptable, + .get_power_limit = arcturus_get_power_limit, + .is_dpm_running = arcturus_is_dpm_running, + .dpm_set_uvd_enable = arcturus_dpm_set_uvd_enable, + .i2c_eeprom_init = arcturus_i2c_eeprom_control_init, + .i2c_eeprom_fini = arcturus_i2c_eeprom_control_fini, + .init_microcode = smu_v11_0_init_microcode, + .load_microcode = smu_v11_0_load_microcode, + .init_smc_tables = smu_v11_0_init_smc_tables, + .fini_smc_tables = smu_v11_0_fini_smc_tables, + .init_power = smu_v11_0_init_power, + .fini_power = smu_v11_0_fini_power, + .check_fw_status = smu_v11_0_check_fw_status, + .setup_pptable = smu_v11_0_setup_pptable, + .get_vbios_bootup_values = smu_v11_0_get_vbios_bootup_values, + .get_clk_info_from_vbios = smu_v11_0_get_clk_info_from_vbios, + .check_pptable = smu_v11_0_check_pptable, + .parse_pptable = smu_v11_0_parse_pptable, + .populate_smc_tables = smu_v11_0_populate_smc_pptable, + .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_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, + .send_smc_msg = smu_v11_0_send_msg, + .send_smc_msg_with_param = smu_v11_0_send_msg_with_param, + .read_smc_arg = smu_v11_0_read_arg, + .init_display_count = smu_v11_0_init_display_count, + .set_allowed_mask = smu_v11_0_set_allowed_mask, + .get_enabled_mask = smu_v11_0_get_enabled_mask, + .notify_display_change = smu_v11_0_notify_display_change, + .set_power_limit = smu_v11_0_set_power_limit, + .get_current_clk_freq = smu_v11_0_get_current_clk_freq, + .init_max_sustainable_clocks = smu_v11_0_init_max_sustainable_clocks, + .start_thermal_control = smu_v11_0_start_thermal_control, + .stop_thermal_control = smu_v11_0_stop_thermal_control, + .set_deep_sleep_dcefclk = smu_v11_0_set_deep_sleep_dcefclk, + .display_clock_voltage_request = smu_v11_0_display_clock_voltage_request, + .get_fan_control_mode = smu_v11_0_get_fan_control_mode, + .set_fan_control_mode = smu_v11_0_set_fan_control_mode, + .set_fan_speed_percent = smu_v11_0_set_fan_speed_percent, + .set_fan_speed_rpm = smu_v11_0_set_fan_speed_rpm, + .set_xgmi_pstate = smu_v11_0_set_xgmi_pstate, + .gfx_off_control = smu_v11_0_gfx_off_control, + .register_irq_handler = smu_v11_0_register_irq_handler, + .set_azalia_d3_pme = smu_v11_0_set_azalia_d3_pme, + .get_max_sustainable_clocks_by_dc = smu_v11_0_get_max_sustainable_clocks_by_dc, + .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, + .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, + .get_pptable_power_limit = arcturus_get_pptable_power_limit, +}; + +void arcturus_set_ppt_funcs(struct smu_context *smu) +{ + smu->ppt_funcs = &arcturus_ppt_funcs; +} diff --git a/drivers/gpu/drm/amd/powerplay/arcturus_ppt.h b/drivers/gpu/drm/amd/powerplay/arcturus_ppt.h new file mode 100644 index 000000000000..d756b16924b8 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/arcturus_ppt.h @@ -0,0 +1,72 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __ARCTURUS_PPT_H__ +#define __ARCTURUS_PPT_H__ + +#define ARCTURUS_UMD_PSTATE_GFXCLK_LEVEL 0x3 +#define ARCTURUS_UMD_PSTATE_SOCCLK_LEVEL 0x3 +#define ARCTURUS_UMD_PSTATE_MCLK_LEVEL 0x2 + +#define MAX_DPM_NUMBER 16 +#define MAX_PCIE_CONF 2 + +struct arcturus_dpm_level { + bool enabled; + uint32_t value; + uint32_t param1; +}; + +struct arcturus_dpm_state { + uint32_t soft_min_level; + uint32_t soft_max_level; + uint32_t hard_min_level; + uint32_t hard_max_level; +}; + +struct arcturus_single_dpm_table { + uint32_t count; + struct arcturus_dpm_state dpm_state; + struct arcturus_dpm_level dpm_levels[MAX_DPM_NUMBER]; +}; + +struct arcturus_pcie_table { + uint16_t count; + uint8_t pcie_gen[MAX_PCIE_CONF]; + uint8_t pcie_lane[MAX_PCIE_CONF]; + uint32_t lclk[MAX_PCIE_CONF]; +}; + +struct arcturus_dpm_table { + struct arcturus_single_dpm_table soc_table; + struct arcturus_single_dpm_table gfx_table; + struct arcturus_single_dpm_table mem_table; + struct arcturus_single_dpm_table eclk_table; + struct arcturus_single_dpm_table vclk_table; + struct arcturus_single_dpm_table dclk_table; + struct arcturus_single_dpm_table fclk_table; + struct arcturus_pcie_table pcie_table; +}; + +extern void arcturus_set_ppt_funcs(struct smu_context *smu); + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/Makefile b/drivers/gpu/drm/amd/powerplay/hwmgr/Makefile index 0b3c6d1d52e4..2773966ae434 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/Makefile +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/Makefile @@ -35,7 +35,9 @@ HARDWARE_MGR = hwmgr.o processpptables.o \ vega12_thermal.o \ pp_overdriver.o smu_helper.o \ vega20_processpptables.o vega20_hwmgr.o vega20_powertune.o \ - vega20_thermal.o common_baco.o vega10_baco.o vega20_baco.o + vega20_thermal.o common_baco.o vega10_baco.o vega20_baco.o \ + vega12_baco.o smu9_baco.o tonga_baco.o polaris_baco.o fiji_baco.o \ + ci_baco.o smu7_baco.o AMD_PP_HWMGR = $(addprefix $(AMD_PP_PATH)/hwmgr/,$(HARDWARE_MGR)) diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/ci_baco.c b/drivers/gpu/drm/amd/powerplay/hwmgr/ci_baco.c new file mode 100644 index 000000000000..3be40114e63d --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/ci_baco.c @@ -0,0 +1,195 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "ci_baco.h" + +#include "gmc/gmc_7_1_d.h" +#include "gmc/gmc_7_1_sh_mask.h" + +#include "bif/bif_4_1_d.h" +#include "bif/bif_4_1_sh_mask.h" + +#include "dce/dce_8_0_d.h" +#include "dce/dce_8_0_sh_mask.h" + +#include "smu/smu_7_0_1_d.h" +#include "smu/smu_7_0_1_sh_mask.h" + +#include "gca/gfx_7_2_d.h" +#include "gca/gfx_7_2_sh_mask.h" + +static const struct baco_cmd_entry gpio_tbl[] = +{ + { CMD_WRITE, mmGPIOPAD_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmGPIOPAD_PD_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmGPIOPAD_PU_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmGPIOPAD_MASK, 0, 0, 0, 0xff77ffff }, + { CMD_WRITE, mmDC_GPIO_DVODATA_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmDC_GPIO_DVODATA_MASK, 0, 0, 0, 0xffffffff }, + { CMD_WRITE, mmDC_GPIO_GENERIC_EN, 0, 0, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmDC_GPIO_GENERIC_MASK, 0, 0, 0, 0x03333333 }, + { CMD_WRITE, mmDC_GPIO_SYNCA_EN, 0, 0, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmDC_GPIO_SYNCA_MASK, 0, 0, 0, 0x00001111 } +}; + +static const struct baco_cmd_entry enable_fb_req_rej_tbl[] = +{ + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC0300024 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x1, 0x0, 0, 0x1 }, + { CMD_WRITE, mmBIF_FB_EN, 0, 0, 0, 0x0 } +}; + +static const struct baco_cmd_entry use_bclk_tbl[] = +{ + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL__SPLL_BYPASS_EN_MASK, CG_SPLL_FUNC_CNTL__SPLL_BYPASS_EN__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL_2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL_2__SPLL_BYPASS_CHG_MASK, CG_SPLL_FUNC_CNTL_2__SPLL_BYPASS_CHG__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_STATUS }, + { CMD_WAITFOR, mmGCK_SMC_IND_DATA, CG_SPLL_STATUS__SPLL_CHG_STATUS_MASK, 0, 0xffffffff, 0x2 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL_2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL_2__SPLL_BYPASS_CHG_MASK, CG_SPLL_FUNC_CNTL_2__SPLL_BYPASS_CHG__SHIFT, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL_2__SPLL_CTLREQ_CHG_MASK, CG_SPLL_FUNC_CNTL_2__SPLL_CTLREQ_CHG__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_STATUS }, + { CMD_WAITFOR, mmGCK_SMC_IND_DATA, CG_SPLL_STATUS__SPLL_CHG_STATUS_MASK, 0, 0xffffffff, 0x2 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL_2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL_2__SPLL_CTLREQ_CHG_MASK, CG_SPLL_FUNC_CNTL_2__SPLL_CTLREQ_CHG__SHIFT, 0, 0x0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC0500170 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x4000000, 0x1a, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixMPLL_BYPASSCLK_SEL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL_MASK, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL__SHIFT, 0, 0x2 }, + { CMD_READMODIFYWRITE, mmMPLL_CNTL_MODE, MPLL_CNTL_MODE__MPLL_SW_DIR_CONTROL_MASK, MPLL_CNTL_MODE__MPLL_SW_DIR_CONTROL__SHIFT, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmMPLL_CNTL_MODE, MPLL_CNTL_MODE__MPLL_MCLK_SEL_MASK, MPLL_CNTL_MODE__MPLL_MCLK_SEL__SHIFT, 0, 0x0 } +}; + +static const struct baco_cmd_entry turn_off_plls_tbl[] = +{ + { CMD_READMODIFYWRITE, mmDISPPLL_BG_CNTL, DISPPLL_BG_CNTL__DISPPLL_BG_PDN_MASK, DISPPLL_BG_CNTL__DISPPLL_BG_PDN__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_CLKPIN_CNTL_DC }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_CLKPIN_CNTL_DC__OSC_EN_MASK, CG_CLKPIN_CNTL_DC__OSC_EN__SHIFT, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_CLKPIN_CNTL_DC__XTALIN_SEL_MASK, CG_CLKPIN_CNTL_DC__XTALIN_SEL__SHIFT, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmPLL_CNTL, PLL_CNTL__PLL_RESET_MASK, PLL_CNTL__PLL_RESET__SHIFT, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmPLL_CNTL, PLL_CNTL__PLL_POWER_DOWN_MASK, PLL_CNTL__PLL_POWER_DOWN__SHIFT, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmPLL_CNTL, PLL_CNTL__PLL_BYPASS_CAL_MASK, PLL_CNTL__PLL_BYPASS_CAL__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL__SPLL_RESET_MASK, CG_SPLL_FUNC_CNTL__SPLL_RESET__SHIFT, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL__SPLL_PWRON_MASK, CG_SPLL_FUNC_CNTL__SPLL_PWRON__SHIFT, 0, 0x0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC0500170 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x2000000, 0x19, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x8000000, 0x1b, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmMPLL_CNTL_MODE, MPLL_CNTL_MODE__GLOBAL_MPLL_RESET_MASK, MPLL_CNTL_MODE__GLOBAL_MPLL_RESET__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmMPLL_CONTROL, 0, 0, 0, 0x00000006 }, + { CMD_WRITE, mmMC_IO_RXCNTL_DPHY0_D0, 0, 0, 0, 0x00007740 }, + { CMD_WRITE, mmMC_IO_RXCNTL_DPHY0_D1, 0, 0, 0, 0x00007740 }, + { CMD_WRITE, mmMC_IO_RXCNTL_DPHY1_D0, 0, 0, 0, 0x00007740 }, + { CMD_WRITE, mmMC_IO_RXCNTL_DPHY1_D1, 0, 0, 0, 0x00007740 }, + { CMD_READMODIFYWRITE, mmMCLK_PWRMGT_CNTL, MCLK_PWRMGT_CNTL__MRDCK0_PDNB_MASK, MCLK_PWRMGT_CNTL__MRDCK0_PDNB__SHIFT, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmMCLK_PWRMGT_CNTL, MCLK_PWRMGT_CNTL__MRDCK1_PDNB_MASK, MCLK_PWRMGT_CNTL__MRDCK1_PDNB__SHIFT, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmMC_SEQ_CNTL_2, MC_SEQ_CNTL_2__DRST_PU_MASK, MC_SEQ_CNTL_2__DRST_PU__SHIFT, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmMC_SEQ_CNTL_2, MC_SEQ_CNTL_2__DRST_PD_MASK, MC_SEQ_CNTL_2__DRST_PD__SHIFT, 0, 0x0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_CLKPIN_CNTL_2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_CLKPIN_CNTL_2__FORCE_BIF_REFCLK_EN_MASK, CG_CLKPIN_CNTL_2__FORCE_BIF_REFCLK_EN__SHIFT, 0, 0x0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixMPLL_BYPASSCLK_SEL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL_MASK, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL__SHIFT, 0, 0x4 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixMISC_CLK_CTRL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL_MASK, MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL__SHIFT, 0, 0x2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MISC_CLK_CTRL__ZCLK_SEL_MASK, MISC_CLK_CTRL__ZCLK_SEL__SHIFT, 0, 0x2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MISC_CLK_CTRL__DFT_SMS_PG_CLK_SEL_MASK, MISC_CLK_CTRL__DFT_SMS_PG_CLK_SEL__SHIFT, 0, 0x2 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixTHM_CLK_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, THM_CLK_CNTL__CMON_CLK_SEL_MASK, THM_CLK_CNTL__CMON_CLK_SEL__SHIFT, 0, 0x2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, THM_CLK_CNTL__TMON_CLK_SEL_MASK, THM_CLK_CNTL__TMON_CLK_SEL__SHIFT, 0, 0x2 } +}; + +static const struct baco_cmd_entry enter_baco_tbl[] = +{ + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_EN_MASK, BACO_CNTL__BACO_EN__SHIFT, 0, 0x01 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BCLK_OFF_MASK, BACO_CNTL__BACO_BCLK_OFF__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_BCLK_OFF_MASK, 0, 5, 0x02 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ISO_DIS_MASK, BACO_CNTL__BACO_ISO_DIS__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_ISO_DIS_MASK, 0, 5, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ANA_ISO_DIS_MASK, BACO_CNTL__BACO_ANA_ISO_DIS__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_ANA_ISO_DIS_MASK, 0, 5, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_POWER_OFF_MASK, BACO_CNTL__BACO_POWER_OFF__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_POWER_OFF_MASK, 0, 5, 0x08 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_MODE_MASK, 0, 0xffffffff, 0x40 } +}; + +#define BACO_CNTL__PWRGOOD_MASK BACO_CNTL__PWRGOOD_GPIO_MASK+BACO_CNTL__PWRGOOD_MEM_MASK+BACO_CNTL__PWRGOOD_DVO_MASK + +static const struct baco_cmd_entry exit_baco_tbl[] = +{ + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_RESET_EN_MASK, BACO_CNTL__BACO_RESET_EN__SHIFT, 0, 0x01 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BCLK_OFF_MASK, BACO_CNTL__BACO_BCLK_OFF__SHIFT, 0, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_POWER_OFF_MASK, BACO_CNTL__BACO_POWER_OFF__SHIFT, 0, 0x00 }, + { CMD_DELAY_MS, 0, 0, 0, 20, 0 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__PWRGOOD_BF_MASK, 0, 0xffffffff, 0x20 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ISO_DIS_MASK, BACO_CNTL__BACO_ISO_DIS__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__PWRGOOD_MASK, 0, 5, 0x1c }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ANA_ISO_DIS_MASK, BACO_CNTL__BACO_ANA_ISO_DIS__SHIFT, 0, 0x01 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_RESET_EN_MASK, BACO_CNTL__BACO_RESET_EN__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__RCU_BIF_CONFIG_DONE_MASK, 0, 5, 0x10 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_EN_MASK, BACO_CNTL__BACO_EN__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_MODE_MASK, 0, 0xffffffff, 0x00 } +}; + +static const struct baco_cmd_entry clean_baco_tbl[] = +{ + { CMD_WRITE, mmBIOS_SCRATCH_6, 0, 0, 0, 0 }, + { CMD_WRITE, mmCP_PFP_UCODE_ADDR, 0, 0, 0, 0 } +}; + +int ci_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state) +{ + enum BACO_STATE cur_state; + + smu7_baco_get_state(hwmgr, &cur_state); + + if (cur_state == state) + /* aisc already in the target state */ + return 0; + + if (state == BACO_STATE_IN) { + baco_program_registers(hwmgr, gpio_tbl, ARRAY_SIZE(gpio_tbl)); + baco_program_registers(hwmgr, enable_fb_req_rej_tbl, + ARRAY_SIZE(enable_fb_req_rej_tbl)); + baco_program_registers(hwmgr, use_bclk_tbl, ARRAY_SIZE(use_bclk_tbl)); + baco_program_registers(hwmgr, turn_off_plls_tbl, + ARRAY_SIZE(turn_off_plls_tbl)); + if (baco_program_registers(hwmgr, enter_baco_tbl, + ARRAY_SIZE(enter_baco_tbl))) + return 0; + + } else if (state == BACO_STATE_OUT) { + /* HW requires at least 20ms between regulator off and on */ + msleep(20); + /* Execute Hardware BACO exit sequence */ + if (baco_program_registers(hwmgr, exit_baco_tbl, + ARRAY_SIZE(exit_baco_tbl))) { + if (baco_program_registers(hwmgr, clean_baco_tbl, + ARRAY_SIZE(clean_baco_tbl))) + return 0; + } + } + + return -EINVAL; +} diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/ci_baco.h b/drivers/gpu/drm/amd/powerplay/hwmgr/ci_baco.h new file mode 100644 index 000000000000..17041f187020 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/ci_baco.h @@ -0,0 +1,29 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __CI_BACO_H__ +#define __CI_BACO_H__ +#include "smu7_baco.h" + +extern int ci_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state); + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/common_baco.c b/drivers/gpu/drm/amd/powerplay/hwmgr/common_baco.c index 9c57c1f67749..1c73776bd606 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/common_baco.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/common_baco.c @@ -79,6 +79,25 @@ static bool baco_cmd_handler(struct pp_hwmgr *hwmgr, u32 command, u32 reg, u32 m return ret; } +bool baco_program_registers(struct pp_hwmgr *hwmgr, + const struct baco_cmd_entry *entry, + const u32 array_size) +{ + u32 i, reg = 0; + + for (i = 0; i < array_size; i++) { + if ((entry[i].cmd == CMD_WRITE) || + (entry[i].cmd == CMD_READMODIFYWRITE) || + (entry[i].cmd == CMD_WAITFOR)) + reg = entry[i].reg_offset; + if (!baco_cmd_handler(hwmgr, entry[i].cmd, reg, entry[i].mask, + entry[i].shift, entry[i].val, entry[i].timeout)) + return false; + } + + return true; +} + bool soc15_baco_program_registers(struct pp_hwmgr *hwmgr, const struct soc15_baco_cmd_entry *entry, const u32 array_size) diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/common_baco.h b/drivers/gpu/drm/amd/powerplay/hwmgr/common_baco.h index 95296c916f4e..8393eb62706d 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/common_baco.h +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/common_baco.h @@ -33,6 +33,15 @@ enum baco_cmd_type { CMD_DELAY_US, }; +struct baco_cmd_entry { + enum baco_cmd_type cmd; + uint32_t reg_offset; + uint32_t mask; + uint32_t shift; + uint32_t timeout; + uint32_t val; +}; + struct soc15_baco_cmd_entry { enum baco_cmd_type cmd; uint32_t hwip; @@ -44,6 +53,10 @@ struct soc15_baco_cmd_entry { uint32_t timeout; uint32_t val; }; + +extern bool baco_program_registers(struct pp_hwmgr *hwmgr, + const struct baco_cmd_entry *entry, + const u32 array_size); extern bool soc15_baco_program_registers(struct pp_hwmgr *hwmgr, const struct soc15_baco_cmd_entry *entry, const u32 array_size); diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/fiji_baco.c b/drivers/gpu/drm/amd/powerplay/hwmgr/fiji_baco.c new file mode 100644 index 000000000000..c0368f2dfb21 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/fiji_baco.c @@ -0,0 +1,196 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "fiji_baco.h" + +#include "gmc/gmc_8_1_d.h" +#include "gmc/gmc_8_1_sh_mask.h" + +#include "bif/bif_5_0_d.h" +#include "bif/bif_5_0_sh_mask.h" + +#include "dce/dce_10_0_d.h" +#include "dce/dce_10_0_sh_mask.h" + +#include "smu/smu_7_1_3_d.h" +#include "smu/smu_7_1_3_sh_mask.h" + + +static const struct baco_cmd_entry gpio_tbl[] = +{ + { CMD_WRITE, mmGPIOPAD_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmGPIOPAD_PD_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmGPIOPAD_PU_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmGPIOPAD_MASK, 0, 0, 0, 0xff77ffff }, + { CMD_WRITE, mmDC_GPIO_DVODATA_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmDC_GPIO_DVODATA_MASK, 0, 0, 0, 0xffffffff }, + { CMD_WRITE, mmDC_GPIO_GENERIC_EN, 0, 0, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmDC_GPIO_GENERIC_MASK, 0, 0, 0, 0x03333333 }, + { CMD_WRITE, mmDC_GPIO_SYNCA_EN, 0, 0, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmDC_GPIO_SYNCA_MASK, 0, 0, 0, 0x00001111 } +}; + +static const struct baco_cmd_entry enable_fb_req_rej_tbl[] = +{ + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC0300024 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x1, 0x0, 0, 0x1 }, + { CMD_WRITE, mmBIF_FB_EN, 0, 0, 0, 0x0 } +}; + +static const struct baco_cmd_entry use_bclk_tbl[] = +{ + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL__SPLL_BYPASS_EN_MASK, CG_SPLL_FUNC_CNTL__SPLL_BYPASS_EN__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL_2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL_2__SPLL_BYPASS_CHG_MASK, CG_SPLL_FUNC_CNTL_2__SPLL_BYPASS_CHG__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_STATUS }, + { CMD_WAITFOR, mmGCK_SMC_IND_DATA, CG_SPLL_STATUS__SPLL_CHG_STATUS_MASK, 0, 0xffffffff, 0x2 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL_2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL_2__SPLL_BYPASS_CHG_MASK, CG_SPLL_FUNC_CNTL_2__SPLL_BYPASS_CHG__SHIFT, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL_2__SPLL_CTLREQ_CHG_MASK, CG_SPLL_FUNC_CNTL_2__SPLL_CTLREQ_CHG__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_STATUS }, + { CMD_WAITFOR, mmGCK_SMC_IND_DATA, CG_SPLL_STATUS__SPLL_CHG_STATUS_MASK, 0, 0xffffffff, 0x2 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL_2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL_2__SPLL_CTLREQ_CHG_MASK, CG_SPLL_FUNC_CNTL_2__SPLL_CTLREQ_CHG__SHIFT, 0, 0x0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC0500170 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x4000000, 0x1a, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixMPLL_BYPASSCLK_SEL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL_MASK, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL__SHIFT, 0, 0x2 } +}; + +static const struct baco_cmd_entry turn_off_plls_tbl[] = +{ + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL__SPLL_RESET_MASK, CG_SPLL_FUNC_CNTL__SPLL_RESET__SHIFT, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL__SPLL_PWRON_MASK, CG_SPLL_FUNC_CNTL__SPLL_PWRON__SHIFT, 0, 0x0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC0500170 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x2000000, 0x19, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x8000000, 0x1b, 0, 0x0 } +}; + +static const struct baco_cmd_entry clk_req_b_tbl[] = +{ + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_CLKPIN_CNTL_2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_CLKPIN_CNTL_2__FORCE_BIF_REFCLK_EN_MASK, CG_CLKPIN_CNTL_2__FORCE_BIF_REFCLK_EN__SHIFT, 0, 0x0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixMPLL_BYPASSCLK_SEL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL_MASK, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL__SHIFT, 0, 0x4 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixMISC_CLK_CTRL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL_MASK, MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL__SHIFT, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MISC_CLK_CTRL__ZCLK_SEL_MASK, MISC_CLK_CTRL__ZCLK_SEL__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_CLKPIN_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_CLKPIN_CNTL__BCLK_AS_XCLK_MASK, CG_CLKPIN_CNTL__BCLK_AS_XCLK__SHIFT, 0, 0x0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixTHM_CLK_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, THM_CLK_CNTL__CMON_CLK_SEL_MASK, THM_CLK_CNTL__CMON_CLK_SEL__SHIFT, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, THM_CLK_CNTL__TMON_CLK_SEL_MASK, THM_CLK_CNTL__TMON_CLK_SEL__SHIFT, 0, 0x1 } +}; + +static const struct baco_cmd_entry enter_baco_tbl[] = +{ + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_EN_MASK, BACO_CNTL__BACO_EN__SHIFT, 0, 0x01 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BIF_SCLK_SWITCH_MASK, BACO_CNTL__BACO_BIF_SCLK_SWITCH__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_BIF_SCLK_SWITCH_MASK, 0, 5, 0x40000 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BCLK_OFF_MASK, BACO_CNTL__BACO_BCLK_OFF__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_BCLK_OFF_MASK, 0, 5, 0x02 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ISO_DIS_MASK, BACO_CNTL__BACO_ISO_DIS__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_ISO_DIS_MASK, 0, 5, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ANA_ISO_DIS_MASK, BACO_CNTL__BACO_ANA_ISO_DIS__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_ANA_ISO_DIS_MASK, 0, 5, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_POWER_OFF_MASK, BACO_CNTL__BACO_POWER_OFF__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_POWER_OFF_MASK, 0, 5, 0x08 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_MODE_MASK, 0, 0xffffffff, 0x40 } +}; + +#define BACO_CNTL__PWRGOOD_MASK BACO_CNTL__PWRGOOD_GPIO_MASK+BACO_CNTL__PWRGOOD_MEM_MASK+BACO_CNTL__PWRGOOD_DVO_MASK + +static const struct baco_cmd_entry exit_baco_tbl[] = +{ + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_RESET_EN_MASK, BACO_CNTL__BACO_RESET_EN__SHIFT, 0, 0x01 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BCLK_OFF_MASK, BACO_CNTL__BACO_BCLK_OFF__SHIFT, 0, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_POWER_OFF_MASK, BACO_CNTL__BACO_POWER_OFF__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__PWRGOOD_BF_MASK, 0, 0xffffffff, 0x200 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ISO_DIS_MASK, BACO_CNTL__BACO_ISO_DIS__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__PWRGOOD_MASK, 0, 5, 0x1c00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ANA_ISO_DIS_MASK, BACO_CNTL__BACO_ANA_ISO_DIS__SHIFT, 0, 0x01 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BIF_SCLK_SWITCH_MASK, BACO_CNTL__BACO_BIF_SCLK_SWITCH__SHIFT, 0, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_RESET_EN_MASK, BACO_CNTL__BACO_RESET_EN__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__RCU_BIF_CONFIG_DONE_MASK, 0, 5, 0x100 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_EN_MASK, BACO_CNTL__BACO_EN__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_MODE_MASK, 0, 0xffffffff, 0x00 } +}; + +static const struct baco_cmd_entry clean_baco_tbl[] = +{ + { CMD_WRITE, mmBIOS_SCRATCH_0, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_1, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_2, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_3, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_4, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_5, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_6, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_7, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_8, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_9, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_10, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_11, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_12, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_13, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_14, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_15, 0, 0, 0, 0 } +}; + +int fiji_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state) +{ + enum BACO_STATE cur_state; + + smu7_baco_get_state(hwmgr, &cur_state); + + if (cur_state == state) + /* aisc already in the target state */ + return 0; + + if (state == BACO_STATE_IN) { + baco_program_registers(hwmgr, gpio_tbl, ARRAY_SIZE(gpio_tbl)); + baco_program_registers(hwmgr, enable_fb_req_rej_tbl, + ARRAY_SIZE(enable_fb_req_rej_tbl)); + baco_program_registers(hwmgr, use_bclk_tbl, ARRAY_SIZE(use_bclk_tbl)); + baco_program_registers(hwmgr, turn_off_plls_tbl, + ARRAY_SIZE(turn_off_plls_tbl)); + baco_program_registers(hwmgr, clk_req_b_tbl, ARRAY_SIZE(clk_req_b_tbl)); + if (baco_program_registers(hwmgr, enter_baco_tbl, + ARRAY_SIZE(enter_baco_tbl))) + return 0; + + } else if (state == BACO_STATE_OUT) { + /* HW requires at least 20ms between regulator off and on */ + msleep(20); + /* Execute Hardware BACO exit sequence */ + if (baco_program_registers(hwmgr, exit_baco_tbl, + ARRAY_SIZE(exit_baco_tbl))) { + if (baco_program_registers(hwmgr, clean_baco_tbl, + ARRAY_SIZE(clean_baco_tbl))) + return 0; + } + } + + return -EINVAL; +} diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/fiji_baco.h b/drivers/gpu/drm/amd/powerplay/hwmgr/fiji_baco.h new file mode 100644 index 000000000000..47f402900bdb --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/fiji_baco.h @@ -0,0 +1,29 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __FIJI_BACO_H__ +#define __FIJI_BACO_H__ +#include "smu7_baco.h" + +extern int fiji_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state); + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c b/drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c index c1c51c115e57..cc57fb953e62 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c @@ -76,7 +76,7 @@ int phm_set_power_state(struct pp_hwmgr *hwmgr, int phm_enable_dynamic_state_management(struct pp_hwmgr *hwmgr) { struct amdgpu_device *adev = NULL; - int ret = -EINVAL;; + int ret = -EINVAL; PHM_FUNC_CHECK(hwmgr); adev = hwmgr->adev; @@ -225,7 +225,16 @@ int phm_register_irq_handlers(struct pp_hwmgr *hwmgr) int phm_start_thermal_controller(struct pp_hwmgr *hwmgr) { int ret = 0; - struct PP_TemperatureRange range = {TEMP_RANGE_MIN, TEMP_RANGE_MAX}; + struct PP_TemperatureRange range = { + TEMP_RANGE_MIN, + TEMP_RANGE_MAX, + TEMP_RANGE_MAX, + TEMP_RANGE_MIN, + TEMP_RANGE_MAX, + TEMP_RANGE_MAX, + TEMP_RANGE_MIN, + TEMP_RANGE_MAX, + TEMP_RANGE_MAX}; struct amdgpu_device *adev = hwmgr->adev; if (hwmgr->hwmgr_func->get_thermal_temperature_range) @@ -239,6 +248,13 @@ int phm_start_thermal_controller(struct pp_hwmgr *hwmgr) adev->pm.dpm.thermal.min_temp = range.min; adev->pm.dpm.thermal.max_temp = range.max; + adev->pm.dpm.thermal.max_edge_emergency_temp = range.edge_emergency_max; + adev->pm.dpm.thermal.min_hotspot_temp = range.hotspot_min; + adev->pm.dpm.thermal.max_hotspot_crit_temp = range.hotspot_crit_max; + adev->pm.dpm.thermal.max_hotspot_emergency_temp = range.hotspot_emergency_max; + adev->pm.dpm.thermal.min_mem_temp = range.mem_min; + adev->pm.dpm.thermal.max_mem_crit_temp = range.mem_crit_max; + adev->pm.dpm.thermal.max_mem_emergency_temp = range.mem_emergency_max; return ret; } diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c index 6cd6497c6fc2..d2909c91d65b 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c @@ -81,6 +81,8 @@ static void hwmgr_init_workload_prority(struct pp_hwmgr *hwmgr) int hwmgr_early_init(struct pp_hwmgr *hwmgr) { + struct amdgpu_device *adev; + if (!hwmgr) return -EINVAL; @@ -92,9 +94,13 @@ int hwmgr_early_init(struct pp_hwmgr *hwmgr) hwmgr_set_user_specify_caps(hwmgr); hwmgr->fan_ctrl_is_in_default_mode = true; hwmgr_init_workload_prority(hwmgr); + hwmgr->gfxoff_state_changed_by_workload = false; + + adev = hwmgr->adev; switch (hwmgr->chip_family) { case AMDGPU_FAMILY_CI: + adev->pm.pp_feature &= ~PP_GFXOFF_MASK; hwmgr->smumgr_funcs = &ci_smu_funcs; ci_set_asic_special_caps(hwmgr); hwmgr->feature_mask &= ~(PP_VBI_TIME_SUPPORT_MASK | @@ -105,12 +111,14 @@ int hwmgr_early_init(struct pp_hwmgr *hwmgr) smu7_init_function_pointers(hwmgr); break; case AMDGPU_FAMILY_CZ: + adev->pm.pp_feature &= ~PP_GFXOFF_MASK; hwmgr->od_enabled = false; hwmgr->smumgr_funcs = &smu8_smu_funcs; hwmgr->feature_mask &= ~PP_GFXOFF_MASK; smu8_init_function_pointers(hwmgr); break; case AMDGPU_FAMILY_VI: + adev->pm.pp_feature &= ~PP_GFXOFF_MASK; hwmgr->feature_mask &= ~PP_GFXOFF_MASK; switch (hwmgr->chip_id) { case CHIP_TOPAZ: @@ -152,6 +160,7 @@ int hwmgr_early_init(struct pp_hwmgr *hwmgr) case AMDGPU_FAMILY_AI: switch (hwmgr->chip_id) { case CHIP_VEGA10: + adev->pm.pp_feature &= ~PP_GFXOFF_MASK; hwmgr->feature_mask &= ~PP_GFXOFF_MASK; hwmgr->smumgr_funcs = &vega10_smu_funcs; vega10_hwmgr_init(hwmgr); @@ -161,6 +170,7 @@ int hwmgr_early_init(struct pp_hwmgr *hwmgr) vega12_hwmgr_init(hwmgr); break; case CHIP_VEGA20: + adev->pm.pp_feature &= ~PP_GFXOFF_MASK; hwmgr->feature_mask &= ~PP_GFXOFF_MASK; hwmgr->smumgr_funcs = &vega20_smu_funcs; vega20_hwmgr_init(hwmgr); @@ -193,6 +203,7 @@ int hwmgr_sw_init(struct pp_hwmgr *hwmgr) return -EINVAL; phm_register_irq_handlers(hwmgr); + pr_info("hwmgr_sw_init smu backed is %s\n", hwmgr->smumgr_funcs->name); return hwmgr->smumgr_funcs->smu_init(hwmgr); } @@ -325,7 +336,7 @@ int hwmgr_resume(struct pp_hwmgr *hwmgr) if (ret) return ret; - ret = psm_adjust_power_state_dynamic(hwmgr, true, NULL); + ret = psm_adjust_power_state_dynamic(hwmgr, false, NULL); return ret; } diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/polaris_baco.c b/drivers/gpu/drm/amd/powerplay/hwmgr/polaris_baco.c new file mode 100644 index 000000000000..8f8e296f2fe9 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/polaris_baco.c @@ -0,0 +1,222 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "polaris_baco.h" + +#include "gmc/gmc_8_1_d.h" +#include "gmc/gmc_8_1_sh_mask.h" + +#include "bif/bif_5_0_d.h" +#include "bif/bif_5_0_sh_mask.h" + +#include "dce/dce_11_0_d.h" +#include "dce/dce_11_0_sh_mask.h" + +#include "smu/smu_7_1_3_d.h" +#include "smu/smu_7_1_3_sh_mask.h" + +static const struct baco_cmd_entry gpio_tbl[] = +{ + { CMD_WRITE, mmGPIOPAD_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmGPIOPAD_PD_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmGPIOPAD_PU_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmGPIOPAD_MASK, 0, 0, 0, 0xff77ffff }, + { CMD_WRITE, mmDC_GPIO_DVODATA_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmDC_GPIO_DVODATA_MASK, 0, 0, 0, 0xffffffff }, + { CMD_WRITE, mmDC_GPIO_GENERIC_EN, 0, 0, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmDC_GPIO_GENERIC_MASK, 0, 0, 0, 0x03333333 }, + { CMD_WRITE, mmDC_GPIO_SYNCA_EN, 0, 0, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmDC_GPIO_SYNCA_MASK, 0, 0, 0, 0x00001111 } +}; + +static const struct baco_cmd_entry enable_fb_req_rej_tbl[] = +{ + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC0300024 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x1, 0x0, 0, 0x1 }, + { CMD_WRITE, mmBIF_FB_EN, 0, 0, 0, 0x0 } +}; + +static const struct baco_cmd_entry use_bclk_tbl[] = +{ + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL__SPLL_BYPASS_EN_MASK, CG_SPLL_FUNC_CNTL__SPLL_BYPASS_EN__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC0500170 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x4000000, 0x1a, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixGCK_DFS_BYPASS_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, GCK_DFS_BYPASS_CNTL__BYPASSACLK_MASK, GCK_DFS_BYPASS_CNTL__BYPASSACLK__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixMPLL_BYPASSCLK_SEL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL_MASK, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL__SHIFT, 0, 0x2 }, + { CMD_READMODIFYWRITE, mmMPLL_CNTL_MODE, MPLL_CNTL_MODE__MPLL_SW_DIR_CONTROL_MASK, MPLL_CNTL_MODE__MPLL_SW_DIR_CONTROL__SHIFT, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmMPLL_CNTL_MODE, MPLL_CNTL_MODE__MPLL_MCLK_SEL_MASK, MPLL_CNTL_MODE__MPLL_MCLK_SEL__SHIFT, 0, 0x0 } +}; + +static const struct baco_cmd_entry turn_off_plls_tbl[] = +{ + { CMD_READMODIFYWRITE, mmDC_GPIO_PAD_STRENGTH_1, DC_GPIO_PAD_STRENGTH_1__GENLK_STRENGTH_SP_MASK, DC_GPIO_PAD_STRENGTH_1__GENLK_STRENGTH_SP__SHIFT, 0, 0x1 }, + { CMD_DELAY_US, 0, 0, 0, 1, 0x0 }, + { CMD_READMODIFYWRITE, mmMC_SEQ_DRAM, MC_SEQ_DRAM__RST_CTL_MASK, MC_SEQ_DRAM__RST_CTL__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC05002B0 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x10, 0x4, 0, 0x1 }, + { CMD_WAITFOR, mmGCK_SMC_IND_DATA, 0x10, 0, 1, 0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC050032C }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x10, 0x4, 0, 0x1 }, + { CMD_WAITFOR, mmGCK_SMC_IND_DATA, 0x10, 0, 1, 0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC0500080 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x1, 0x0, 0, 0x1 }, + { CMD_READMODIFYWRITE, 0xda2, 0x40, 0x6, 0, 0x0 }, + { CMD_DELAY_US, 0, 0, 0, 3, 0x0 }, + { CMD_READMODIFYWRITE, 0xda2, 0x8, 0x3, 0, 0x0 }, + { CMD_READMODIFYWRITE, 0xda2, 0x3fff00, 0x8, 0, 0x32 }, + { CMD_DELAY_US, 0, 0, 0, 3, 0x0 }, + { CMD_READMODIFYWRITE, mmMPLL_FUNC_CNTL_2, MPLL_FUNC_CNTL_2__ISO_DIS_P_MASK, MPLL_FUNC_CNTL_2__ISO_DIS_P__SHIFT, 0, 0x0 }, + { CMD_DELAY_US, 0, 0, 0, 5, 0x0 } +}; + +static const struct baco_cmd_entry clk_req_b_tbl[] = +{ + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixTHM_CLK_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, THM_CLK_CNTL__CMON_CLK_SEL_MASK, THM_CLK_CNTL__CMON_CLK_SEL__SHIFT, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, THM_CLK_CNTL__TMON_CLK_SEL_MASK, THM_CLK_CNTL__TMON_CLK_SEL__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixMISC_CLK_CTRL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL_MASK, MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL__SHIFT, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MISC_CLK_CTRL__ZCLK_SEL_MASK, MISC_CLK_CTRL__ZCLK_SEL__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_CLKPIN_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_CLKPIN_CNTL__BCLK_AS_XCLK_MASK, CG_CLKPIN_CNTL__BCLK_AS_XCLK__SHIFT, 0, 0x0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_CLKPIN_CNTL_2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_CLKPIN_CNTL_2__FORCE_BIF_REFCLK_EN_MASK, CG_CLKPIN_CNTL_2__FORCE_BIF_REFCLK_EN__SHIFT, 0, 0x0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixMPLL_BYPASSCLK_SEL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL_MASK, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL__SHIFT, 0, 0x4 } +}; + +static const struct baco_cmd_entry enter_baco_tbl[] = +{ + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_EN_MASK, BACO_CNTL__BACO_EN__SHIFT, 0, 0x01 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BIF_SCLK_SWITCH_MASK, BACO_CNTL__BACO_BIF_SCLK_SWITCH__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_BIF_SCLK_SWITCH_MASK, 0, 5, 0x40000 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BCLK_OFF_MASK, BACO_CNTL__BACO_BCLK_OFF__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_BCLK_OFF_MASK, 0, 5, 0x02 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ISO_DIS_MASK, BACO_CNTL__BACO_ISO_DIS__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_ISO_DIS_MASK, 0, 5, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ANA_ISO_DIS_MASK, BACO_CNTL__BACO_ANA_ISO_DIS__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_ANA_ISO_DIS_MASK, 0, 5, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_POWER_OFF_MASK, BACO_CNTL__BACO_POWER_OFF__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_POWER_OFF_MASK, 0, 5, 0x08 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_MODE_MASK, 0, 0xffffffff, 0x40 } +}; + +#define BACO_CNTL__PWRGOOD_MASK BACO_CNTL__PWRGOOD_GPIO_MASK+BACO_CNTL__PWRGOOD_MEM_MASK+BACO_CNTL__PWRGOOD_DVO_MASK + +static const struct baco_cmd_entry exit_baco_tbl[] = +{ + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_RESET_EN_MASK, BACO_CNTL__BACO_RESET_EN__SHIFT, 0, 0x01 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BCLK_OFF_MASK, BACO_CNTL__BACO_BCLK_OFF__SHIFT, 0, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_POWER_OFF_MASK, BACO_CNTL__BACO_POWER_OFF__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__PWRGOOD_BF_MASK, 0, 0xffffffff, 0x200 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ISO_DIS_MASK, BACO_CNTL__BACO_ISO_DIS__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__PWRGOOD_MASK, 0, 5, 0x1c00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ANA_ISO_DIS_MASK, BACO_CNTL__BACO_ANA_ISO_DIS__SHIFT, 0, 0x01 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BIF_SCLK_SWITCH_MASK, BACO_CNTL__BACO_BIF_SCLK_SWITCH__SHIFT, 0, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_RESET_EN_MASK, BACO_CNTL__BACO_RESET_EN__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__RCU_BIF_CONFIG_DONE_MASK, 0, 5, 0x100 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_EN_MASK, BACO_CNTL__BACO_EN__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_MODE_MASK, 0, 0xffffffff, 0x00 } +}; + +static const struct baco_cmd_entry clean_baco_tbl[] = +{ + { CMD_WRITE, mmBIOS_SCRATCH_6, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_7, 0, 0, 0, 0 } +}; + +static const struct baco_cmd_entry use_bclk_tbl_vg[] = +{ + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL__SPLL_BYPASS_EN_MASK, CG_SPLL_FUNC_CNTL__SPLL_BYPASS_EN__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC0500170 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x4000000, 0x1a, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixGCK_DFS_BYPASS_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, GCK_DFS_BYPASS_CNTL__BYPASSACLK_MASK, GCK_DFS_BYPASS_CNTL__BYPASSACLK__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixMPLL_BYPASSCLK_SEL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL_MASK, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL__SHIFT, 0, 0x2 } +}; + +static const struct baco_cmd_entry turn_off_plls_tbl_vg[] = +{ + { CMD_READMODIFYWRITE, mmDC_GPIO_PAD_STRENGTH_1, DC_GPIO_PAD_STRENGTH_1__GENLK_STRENGTH_SP_MASK, DC_GPIO_PAD_STRENGTH_1__GENLK_STRENGTH_SP__SHIFT, 0, 0x1 }, + { CMD_DELAY_US, 0, 0, 0, 1, 0x0 }, + { CMD_READMODIFYWRITE, mmMC_SEQ_DRAM, MC_SEQ_DRAM__RST_CTL_MASK, MC_SEQ_DRAM__RST_CTL__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC05002B0 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x10, 0x4, 0, 0x1 }, + { CMD_WAITFOR, mmGCK_SMC_IND_DATA, 0x10, 0, 1, 0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC050032C }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x10, 0x4, 0, 0x1 }, + { CMD_WAITFOR, mmGCK_SMC_IND_DATA, 0x10, 0, 1, 0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC0500080 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x1, 0x0, 0, 0x1 }, + { CMD_DELAY_US, 0, 0, 0, 3, 0x0 }, + { CMD_DELAY_US, 0, 0, 0, 3, 0x0 }, + { CMD_DELAY_US, 0, 0, 0, 5, 0x0 } +}; + +int polaris_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state) +{ + enum BACO_STATE cur_state; + + smu7_baco_get_state(hwmgr, &cur_state); + + if (cur_state == state) + /* aisc already in the target state */ + return 0; + + if (state == BACO_STATE_IN) { + baco_program_registers(hwmgr, gpio_tbl, ARRAY_SIZE(gpio_tbl)); + baco_program_registers(hwmgr, enable_fb_req_rej_tbl, + ARRAY_SIZE(enable_fb_req_rej_tbl)); + if (hwmgr->chip_id == CHIP_VEGAM) { + baco_program_registers(hwmgr, use_bclk_tbl_vg, ARRAY_SIZE(use_bclk_tbl_vg)); + baco_program_registers(hwmgr, turn_off_plls_tbl_vg, + ARRAY_SIZE(turn_off_plls_tbl_vg)); + } else { + baco_program_registers(hwmgr, use_bclk_tbl, ARRAY_SIZE(use_bclk_tbl)); + baco_program_registers(hwmgr, turn_off_plls_tbl, + ARRAY_SIZE(turn_off_plls_tbl)); + } + baco_program_registers(hwmgr, clk_req_b_tbl, ARRAY_SIZE(clk_req_b_tbl)); + if (baco_program_registers(hwmgr, enter_baco_tbl, + ARRAY_SIZE(enter_baco_tbl))) + return 0; + + } else if (state == BACO_STATE_OUT) { + /* HW requires at least 20ms between regulator off and on */ + msleep(20); + /* Execute Hardware BACO exit sequence */ + if (baco_program_registers(hwmgr, exit_baco_tbl, + ARRAY_SIZE(exit_baco_tbl))) { + if (baco_program_registers(hwmgr, clean_baco_tbl, + ARRAY_SIZE(clean_baco_tbl))) + return 0; + } + } + + return -EINVAL; +} diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/polaris_baco.h b/drivers/gpu/drm/amd/powerplay/hwmgr/polaris_baco.h new file mode 100644 index 000000000000..87a5fa0a157a --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/polaris_baco.h @@ -0,0 +1,29 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __POLARIS_BACO_H__ +#define __POLARIS_BACO_H__ +#include "smu7_baco.h" + +extern int polaris_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state); + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/process_pptables_v1_0.c b/drivers/gpu/drm/amd/powerplay/hwmgr/process_pptables_v1_0.c index ae64ff7153d6..b760f95e7fa7 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/process_pptables_v1_0.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/process_pptables_v1_0.c @@ -916,8 +916,10 @@ static int init_thermal_controller( PHM_PlatformCaps_ThermalController ); - if (0 == powerplay_table->usFanTableOffset) + if (0 == powerplay_table->usFanTableOffset) { + hwmgr->thermal_controller.use_hw_fan_control = 1; return 0; + } fan_table = (const PPTable_Generic_SubTable_Header *) (((unsigned long)powerplay_table) + @@ -1065,8 +1067,6 @@ static int pp_tables_v1_0_initialize(struct pp_hwmgr *hwmgr) PP_ASSERT_WITH_CODE((NULL != hwmgr->pptable), "Failed to allocate hwmgr->pptable!", return -ENOMEM); - memset(hwmgr->pptable, 0x00, sizeof(struct phm_ppt_v1_information)); - powerplay_table = get_powerplay_table(hwmgr); PP_ASSERT_WITH_CODE((NULL != powerplay_table), diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c index 0ad8fe4a6277..1115761982a7 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c @@ -35,6 +35,7 @@ #include "smu10_hwmgr.h" #include "power_state.h" #include "soc15_common.h" +#include "smu10.h" #define SMU10_MAX_DEEPSLEEP_DIVIDER_ID 5 #define SMU10_MINIMUM_ENGINE_CLOCK 800 /* 8Mhz, the low boundary of engine clock allowed on this chip */ @@ -114,11 +115,6 @@ static int smu10_initialize_dpm_defaults(struct pp_hwmgr *hwmgr) smu10_data->num_active_display = 0; smu10_data->deep_sleep_dcefclk = 0; - if (hwmgr->feature_mask & PP_GFXOFF_MASK) - smu10_data->gfx_off_controled_by_driver = true; - else - smu10_data->gfx_off_controled_by_driver = false; - phm_cap_unset(hwmgr->platform_descriptor.platformCaps, PHM_PlatformCaps_SclkDeepSleep); @@ -209,18 +205,13 @@ static int smu10_set_clock_limit(struct pp_hwmgr *hwmgr, const void *input) return 0; } -static inline uint32_t convert_10k_to_mhz(uint32_t clock) -{ - return (clock + 99) / 100; -} - static int smu10_set_min_deep_sleep_dcefclk(struct pp_hwmgr *hwmgr, uint32_t clock) { struct smu10_hwmgr *smu10_data = (struct smu10_hwmgr *)(hwmgr->backend); if (smu10_data->need_min_deep_sleep_dcefclk && - smu10_data->deep_sleep_dcefclk != convert_10k_to_mhz(clock)) { - smu10_data->deep_sleep_dcefclk = convert_10k_to_mhz(clock); + smu10_data->deep_sleep_dcefclk != clock) { + smu10_data->deep_sleep_dcefclk = clock; smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetMinDeepSleepDcefclk, smu10_data->deep_sleep_dcefclk); @@ -233,8 +224,8 @@ static int smu10_set_hard_min_dcefclk_by_freq(struct pp_hwmgr *hwmgr, uint32_t c struct smu10_hwmgr *smu10_data = (struct smu10_hwmgr *)(hwmgr->backend); if (smu10_data->dcf_actual_hard_min_freq && - smu10_data->dcf_actual_hard_min_freq != convert_10k_to_mhz(clock)) { - smu10_data->dcf_actual_hard_min_freq = convert_10k_to_mhz(clock); + smu10_data->dcf_actual_hard_min_freq != clock) { + smu10_data->dcf_actual_hard_min_freq = clock; smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetHardMinDcefclkByFreq, smu10_data->dcf_actual_hard_min_freq); @@ -247,8 +238,8 @@ static int smu10_set_hard_min_fclk_by_freq(struct pp_hwmgr *hwmgr, uint32_t cloc struct smu10_hwmgr *smu10_data = (struct smu10_hwmgr *)(hwmgr->backend); if (smu10_data->f_actual_hard_min_freq && - smu10_data->f_actual_hard_min_freq != convert_10k_to_mhz(clock)) { - smu10_data->f_actual_hard_min_freq = convert_10k_to_mhz(clock); + smu10_data->f_actual_hard_min_freq != clock) { + smu10_data->f_actual_hard_min_freq = clock; smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetHardMinFclkByFreq, smu10_data->f_actual_hard_min_freq); @@ -330,9 +321,9 @@ static bool smu10_is_gfx_on(struct pp_hwmgr *hwmgr) static int smu10_disable_gfx_off(struct pp_hwmgr *hwmgr) { - struct smu10_hwmgr *smu10_data = (struct smu10_hwmgr *)(hwmgr->backend); + struct amdgpu_device *adev = hwmgr->adev; - if (smu10_data->gfx_off_controled_by_driver) { + if (adev->pm.pp_feature & PP_GFXOFF_MASK) { smum_send_msg_to_smc(hwmgr, PPSMC_MSG_DisableGfxOff); /* confirm gfx is back to "on" state */ @@ -350,9 +341,9 @@ static int smu10_disable_dpm_tasks(struct pp_hwmgr *hwmgr) static int smu10_enable_gfx_off(struct pp_hwmgr *hwmgr) { - struct smu10_hwmgr *smu10_data = (struct smu10_hwmgr *)(hwmgr->backend); + struct amdgpu_device *adev = hwmgr->adev; - if (smu10_data->gfx_off_controled_by_driver) + if (adev->pm.pp_feature & PP_GFXOFF_MASK) smum_send_msg_to_smc(hwmgr, PPSMC_MSG_EnableGfxOff); return 0; @@ -577,7 +568,6 @@ static int smu10_dpm_force_dpm_level(struct pp_hwmgr *hwmgr, enum amd_dpm_forced_level level) { struct smu10_hwmgr *data = hwmgr->backend; - struct amdgpu_device *adev = hwmgr->adev; uint32_t min_sclk = hwmgr->display_config->min_core_set_clock; uint32_t min_mclk = hwmgr->display_config->min_mem_set_clock/100; @@ -586,11 +576,6 @@ static int smu10_dpm_force_dpm_level(struct pp_hwmgr *hwmgr, return 0; } - /* Disable UMDPSTATE support on rv2 temporarily */ - if ((adev->asic_type == CHIP_RAVEN) && - (adev->rev_id >= 8)) - return 0; - if (min_sclk < data->gfx_min_freq_limit) min_sclk = data->gfx_min_freq_limit; @@ -1126,6 +1111,7 @@ static int smu10_thermal_get_temperature(struct pp_hwmgr *hwmgr) static int smu10_read_sensor(struct pp_hwmgr *hwmgr, int idx, void *value, int *size) { + struct smu10_hwmgr *smu10_data = (struct smu10_hwmgr *)(hwmgr->backend); uint32_t sclk, mclk; int ret = 0; @@ -1147,6 +1133,10 @@ static int smu10_read_sensor(struct pp_hwmgr *hwmgr, int idx, case AMDGPU_PP_SENSOR_GPU_TEMP: *((uint32_t *)value) = smu10_thermal_get_temperature(hwmgr); break; + case AMDGPU_PP_SENSOR_VCN_POWER_STATE: + *(uint32_t *)value = smu10_data->vcn_power_gated ? 0 : 1; + *size = 4; + break; default: ret = -EINVAL; break; @@ -1190,21 +1180,144 @@ static int smu10_powergate_sdma(struct pp_hwmgr *hwmgr, bool gate) static void smu10_powergate_vcn(struct pp_hwmgr *hwmgr, bool bgate) { + struct smu10_hwmgr *smu10_data = (struct smu10_hwmgr *)(hwmgr->backend); + if (bgate) { amdgpu_device_ip_set_powergating_state(hwmgr->adev, AMD_IP_BLOCK_TYPE_VCN, AMD_PG_STATE_GATE); smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_PowerDownVcn, 0); + smu10_data->vcn_power_gated = true; } else { smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_PowerUpVcn, 0); amdgpu_device_ip_set_powergating_state(hwmgr->adev, AMD_IP_BLOCK_TYPE_VCN, AMD_PG_STATE_UNGATE); + smu10_data->vcn_power_gated = false; } } +static int conv_power_profile_to_pplib_workload(int power_profile) +{ + int pplib_workload = 0; + + switch (power_profile) { + case PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT: + pplib_workload = WORKLOAD_DEFAULT_BIT; + break; + case PP_SMC_POWER_PROFILE_FULLSCREEN3D: + pplib_workload = WORKLOAD_PPLIB_FULL_SCREEN_3D_BIT; + break; + case PP_SMC_POWER_PROFILE_POWERSAVING: + pplib_workload = WORKLOAD_PPLIB_POWER_SAVING_BIT; + break; + case PP_SMC_POWER_PROFILE_VIDEO: + pplib_workload = WORKLOAD_PPLIB_VIDEO_BIT; + break; + case PP_SMC_POWER_PROFILE_VR: + pplib_workload = WORKLOAD_PPLIB_VR_BIT; + break; + case PP_SMC_POWER_PROFILE_COMPUTE: + pplib_workload = WORKLOAD_PPLIB_COMPUTE_BIT; + break; + } + + return pplib_workload; +} + +static int smu10_get_power_profile_mode(struct pp_hwmgr *hwmgr, char *buf) +{ + uint32_t i, size = 0; + static const uint8_t + profile_mode_setting[6][4] = {{70, 60, 0, 0,}, + {70, 60, 1, 3,}, + {90, 60, 0, 0,}, + {70, 60, 0, 0,}, + {70, 90, 0, 0,}, + {30, 60, 0, 6,}, + }; + static const char *profile_name[6] = { + "BOOTUP_DEFAULT", + "3D_FULL_SCREEN", + "POWER_SAVING", + "VIDEO", + "VR", + "COMPUTE"}; + static const char *title[6] = {"NUM", + "MODE_NAME", + "BUSY_SET_POINT", + "FPS", + "USE_RLC_BUSY", + "MIN_ACTIVE_LEVEL"}; + + if (!buf) + return -EINVAL; + + size += sprintf(buf + size, "%s %16s %s %s %s %s\n",title[0], + title[1], title[2], title[3], title[4], title[5]); + + for (i = 0; i <= PP_SMC_POWER_PROFILE_COMPUTE; i++) + size += sprintf(buf + size, "%3d %14s%s: %14d %3d %10d %14d\n", + i, profile_name[i], (i == hwmgr->power_profile_mode) ? "*" : " ", + profile_mode_setting[i][0], profile_mode_setting[i][1], + profile_mode_setting[i][2], profile_mode_setting[i][3]); + + return size; +} + +static bool smu10_is_raven1_refresh(struct pp_hwmgr *hwmgr) +{ + struct amdgpu_device *adev = hwmgr->adev; + if ((adev->asic_type == CHIP_RAVEN) && + (adev->rev_id != 0x15d8) && + (hwmgr->smu_version >= 0x41e2b)) + return true; + else + return false; +} + +static int smu10_set_power_profile_mode(struct pp_hwmgr *hwmgr, long *input, uint32_t size) +{ + int workload_type = 0; + int result = 0; + + if (input[size] > PP_SMC_POWER_PROFILE_COMPUTE) { + pr_err("Invalid power profile mode %ld\n", input[size]); + return -EINVAL; + } + if (hwmgr->power_profile_mode == input[size]) + return 0; + + /* conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT */ + workload_type = + conv_power_profile_to_pplib_workload(input[size]); + if (workload_type && + smu10_is_raven1_refresh(hwmgr) && + !hwmgr->gfxoff_state_changed_by_workload) { + smu10_gfx_off_control(hwmgr, false); + hwmgr->gfxoff_state_changed_by_workload = true; + } + result = smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_ActiveProcessNotify, + 1 << workload_type); + if (!result) + hwmgr->power_profile_mode = input[size]; + if (workload_type && hwmgr->gfxoff_state_changed_by_workload) { + smu10_gfx_off_control(hwmgr, true); + hwmgr->gfxoff_state_changed_by_workload = false; + } + + return 0; +} + +static int smu10_asic_reset(struct pp_hwmgr *hwmgr, enum SMU_ASIC_RESET_MODE mode) +{ + return smum_send_msg_to_smc_with_parameter(hwmgr, + PPSMC_MSG_DeviceDriverReset, + mode); +} + static const struct pp_hwmgr_func smu10_hwmgr_funcs = { .backend_init = smu10_hwmgr_backend_init, .backend_fini = smu10_hwmgr_backend_fini, @@ -1246,6 +1359,9 @@ static const struct pp_hwmgr_func smu10_hwmgr_funcs = { .powergate_sdma = smu10_powergate_sdma, .set_hard_min_dcefclk_by_freq = smu10_set_hard_min_dcefclk_by_freq, .set_hard_min_fclk_by_freq = smu10_set_hard_min_fclk_by_freq, + .get_power_profile_mode = smu10_get_power_profile_mode, + .set_power_profile_mode = smu10_set_power_profile_mode, + .asic_reset = smu10_asic_reset, }; int smu10_init_function_pointers(struct pp_hwmgr *hwmgr) diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_baco.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_baco.c new file mode 100644 index 000000000000..044cda005aed --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_baco.c @@ -0,0 +1,91 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "smu7_baco.h" +#include "tonga_baco.h" +#include "fiji_baco.h" +#include "polaris_baco.h" +#include "ci_baco.h" + +#include "bif/bif_5_0_d.h" +#include "bif/bif_5_0_sh_mask.h" + +#include "smu/smu_7_1_2_d.h" +#include "smu/smu_7_1_2_sh_mask.h" + +int smu7_baco_get_capability(struct pp_hwmgr *hwmgr, bool *cap) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)(hwmgr->adev); + uint32_t reg; + + *cap = false; + if (!phm_cap_enabled(hwmgr->platform_descriptor.platformCaps, PHM_PlatformCaps_BACO)) + return 0; + + reg = RREG32(mmCC_BIF_BX_FUSESTRAP0); + + if (reg & CC_BIF_BX_FUSESTRAP0__STRAP_BIF_PX_CAPABLE_MASK) + *cap = true; + + return 0; +} + +int smu7_baco_get_state(struct pp_hwmgr *hwmgr, enum BACO_STATE *state) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)(hwmgr->adev); + uint32_t reg; + + reg = RREG32(mmBACO_CNTL); + + if (reg & BACO_CNTL__BACO_MODE_MASK) + /* gfx has already entered BACO state */ + *state = BACO_STATE_IN; + else + *state = BACO_STATE_OUT; + return 0; +} + +int smu7_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)(hwmgr->adev); + + switch (adev->asic_type) { + case CHIP_TOPAZ: + case CHIP_TONGA: + return tonga_baco_set_state(hwmgr, state); + case CHIP_FIJI: + return fiji_baco_set_state(hwmgr, state); + case CHIP_POLARIS10: + case CHIP_POLARIS11: + case CHIP_POLARIS12: + case CHIP_VEGAM: + return polaris_baco_set_state(hwmgr, state); +#ifdef CONFIG_DRM_AMDGPU_CIK + case CHIP_BONAIRE: + case CHIP_HAWAII: + return ci_baco_set_state(hwmgr, state); +#endif + default: + return -EINVAL; + } +} diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_baco.h b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_baco.h new file mode 100644 index 000000000000..be0d98abb536 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_baco.h @@ -0,0 +1,32 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __SMU7_BACO_H__ +#define __SMU7_BACO_H__ +#include "hwmgr.h" +#include "common_baco.h" + +extern int smu7_baco_get_capability(struct pp_hwmgr *hwmgr, bool *cap); +extern int smu7_baco_get_state(struct pp_hwmgr *hwmgr, enum BACO_STATE *state); +extern int smu7_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state); + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c index 83d3d935f3ac..f73dff68e799 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c @@ -24,6 +24,7 @@ #include <linux/delay.h> #include <linux/fb.h> #include <linux/module.h> +#include <linux/pci.h> #include <linux/slab.h> #include <asm/div64.h> #include <drm/amdgpu_drm.h> @@ -47,6 +48,7 @@ #include "smu7_clockpowergating.h" #include "processpptables.h" #include "pp_thermal.h" +#include "smu7_baco.h" #include "ivsrcid/ivsrcid_vislands30.h" @@ -77,7 +79,7 @@ #define PCIE_BUS_CLK 10000 #define TCLK (PCIE_BUS_CLK / 10) -static const struct profile_mode_setting smu7_profiling[7] = +static struct profile_mode_setting smu7_profiling[7] = {{0, 0, 0, 0, 0, 0, 0, 0}, {1, 0, 100, 30, 1, 0, 100, 10}, {1, 10, 0, 30, 0, 0, 0, 0}, @@ -1993,7 +1995,6 @@ static int smu7_sort_lookup_table(struct pp_hwmgr *hwmgr, struct phm_ppt_v1_voltage_lookup_table *lookup_table) { uint32_t table_size, i, j; - struct phm_ppt_v1_voltage_lookup_record tmp_voltage_lookup_record; table_size = lookup_table->count; PP_ASSERT_WITH_CODE(0 != lookup_table->count, @@ -2004,9 +2005,8 @@ static int smu7_sort_lookup_table(struct pp_hwmgr *hwmgr, for (j = i + 1; j > 0; j--) { if (lookup_table->entries[j].us_vdd < lookup_table->entries[j - 1].us_vdd) { - tmp_voltage_lookup_record = lookup_table->entries[j - 1]; - lookup_table->entries[j - 1] = lookup_table->entries[j]; - lookup_table->entries[j] = tmp_voltage_lookup_record; + swap(lookup_table->entries[j - 1], + lookup_table->entries[j]); } } } @@ -2955,9 +2955,10 @@ static int smu7_apply_state_adjust_rules(struct pp_hwmgr *hwmgr, if (hwmgr->display_config->num_display == 0) disable_mclk_switching = false; else - disable_mclk_switching = ((1 < hwmgr->display_config->num_display) || - disable_mclk_switching_for_frame_lock || - smu7_vblank_too_short(hwmgr, hwmgr->display_config->min_vblank_time)); + disable_mclk_switching = ((1 < hwmgr->display_config->num_display) && + !hwmgr->display_config->multi_monitor_in_sync) || + disable_mclk_switching_for_frame_lock || + smu7_vblank_too_short(hwmgr, hwmgr->display_config->min_vblank_time); sclk = smu7_ps->performance_levels[0].engine_clock; mclk = smu7_ps->performance_levels[0].memory_clock; @@ -3476,25 +3477,38 @@ static int smu7_get_pp_table_entry(struct pp_hwmgr *hwmgr, static int smu7_get_gpu_power(struct pp_hwmgr *hwmgr, u32 *query) { + struct amdgpu_device *adev = hwmgr->adev; int i; u32 tmp = 0; if (!query) return -EINVAL; - smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_GetCurrPkgPwr, 0); - tmp = cgs_read_register(hwmgr->device, mmSMC_MSG_ARG_0); - *query = tmp; + /* + * PPSMC_MSG_GetCurrPkgPwr is not supported on: + * - Hawaii + * - Bonaire + * - Fiji + * - Tonga + */ + if ((adev->asic_type != CHIP_HAWAII) && + (adev->asic_type != CHIP_BONAIRE) && + (adev->asic_type != CHIP_FIJI) && + (adev->asic_type != CHIP_TONGA)) { + smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_GetCurrPkgPwr, 0); + tmp = cgs_read_register(hwmgr->device, mmSMC_MSG_ARG_0); + *query = tmp; - if (tmp != 0) - return 0; + if (tmp != 0) + return 0; + } smum_send_msg_to_smc(hwmgr, PPSMC_MSG_PmStatusLogStart); cgs_write_ind_register(hwmgr->device, CGS_IND_REG__SMC, ixSMU_PM_STATUS_95, 0); for (i = 0; i < 10; i++) { - mdelay(500); + msleep(500); smum_send_msg_to_smc(hwmgr, PPSMC_MSG_PmStatusLogSample); tmp = cgs_read_ind_register(hwmgr->device, CGS_IND_REG__SMC, @@ -3532,9 +3546,12 @@ static int smu7_read_sensor(struct pp_hwmgr *hwmgr, int idx, *size = 4; return 0; case AMDGPU_PP_SENSOR_GPU_LOAD: + case AMDGPU_PP_SENSOR_MEM_LOAD: offset = data->soft_regs_start + smum_get_offsetof(hwmgr, SMU_SoftRegisters, - AverageGraphicsActivity); + (idx == AMDGPU_PP_SENSOR_GPU_LOAD) ? + AverageGraphicsActivity: + AverageMemoryActivity); activity_percent = cgs_read_ind_register(hwmgr->device, CGS_IND_REG__SMC, offset); activity_percent += 0x80; @@ -3965,6 +3982,13 @@ static int smu7_set_power_state_tasks(struct pp_hwmgr *hwmgr, const void *input) "Failed to populate and upload SCLK MCLK DPM levels!", result = tmp_result); + /* + * If a custom pp table is loaded, set DPMTABLE_OD_UPDATE_VDDC flag. + * That effectively disables AVFS feature. + */ + if (hwmgr->hardcode_pp_table != NULL) + data->need_update_smu7_dpm_table |= DPMTABLE_OD_UPDATE_VDDC; + tmp_result = smu7_update_avfs(hwmgr); PP_ASSERT_WITH_CODE((0 == tmp_result), "Failed to update avfs voltages!", @@ -4064,6 +4088,11 @@ static int smu7_program_display_gap(struct pp_hwmgr *hwmgr) data->frame_time_x2 = frame_time_in_us * 2 / 100; + if (data->frame_time_x2 < 280) { + pr_debug("%s: enforce minimal VBITimeout: %d -> 280\n", __func__, data->frame_time_x2); + data->frame_time_x2 = 280; + } + display_gap2 = pre_vbi_time_in_us * (ref_clock / 100); cgs_write_ind_register(hwmgr->device, CGS_IND_REG__SMC, ixCG_DISPLAY_GAP_CNTL2, display_gap2); @@ -4984,17 +5013,27 @@ static int smu7_set_power_profile_mode(struct pp_hwmgr *hwmgr, long *input, uint mode = input[size]; switch (mode) { case PP_SMC_POWER_PROFILE_CUSTOM: - if (size < 8) + if (size < 8 && size != 0) return -EINVAL; - - tmp.bupdate_sclk = input[0]; - tmp.sclk_up_hyst = input[1]; - tmp.sclk_down_hyst = input[2]; - tmp.sclk_activity = input[3]; - tmp.bupdate_mclk = input[4]; - tmp.mclk_up_hyst = input[5]; - tmp.mclk_down_hyst = input[6]; - tmp.mclk_activity = input[7]; + /* If only CUSTOM is passed in, use the saved values. Check + * that we actually have a CUSTOM profile by ensuring that + * the "use sclk" or the "use mclk" bits are set + */ + tmp = smu7_profiling[PP_SMC_POWER_PROFILE_CUSTOM]; + if (size == 0) { + if (tmp.bupdate_sclk == 0 && tmp.bupdate_mclk == 0) + return -EINVAL; + } else { + tmp.bupdate_sclk = input[0]; + tmp.sclk_up_hyst = input[1]; + tmp.sclk_down_hyst = input[2]; + tmp.sclk_activity = input[3]; + tmp.bupdate_mclk = input[4]; + tmp.mclk_up_hyst = input[5]; + tmp.mclk_down_hyst = input[6]; + tmp.mclk_activity = input[7]; + smu7_profiling[PP_SMC_POWER_PROFILE_CUSTOM] = tmp; + } if (!smum_update_dpm_settings(hwmgr, &tmp)) { memcpy(&data->current_profile_setting, &tmp, sizeof(struct profile_mode_setting)); hwmgr->power_profile_mode = mode; @@ -5125,6 +5164,9 @@ static const struct pp_hwmgr_func smu7_hwmgr_funcs = { .get_power_profile_mode = smu7_get_power_profile_mode, .set_power_profile_mode = smu7_set_power_profile_mode, .get_performance_level = smu7_get_performance_level, + .get_asic_baco_capability = smu7_baco_get_capability, + .get_asic_baco_state = smu7_baco_get_state, + .set_asic_baco_state = smu7_baco_set_state, .power_off_asic = smu7_power_off_asic, }; diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu9_baco.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu9_baco.c new file mode 100644 index 000000000000..de0a37f7c632 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu9_baco.c @@ -0,0 +1,66 @@ +/* + * Copyright 2018 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "soc15.h" +#include "soc15_hw_ip.h" +#include "vega10_ip_offset.h" +#include "soc15_common.h" +#include "vega10_inc.h" +#include "smu9_baco.h" + +int smu9_baco_get_capability(struct pp_hwmgr *hwmgr, bool *cap) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)(hwmgr->adev); + uint32_t reg, data; + + *cap = false; + if (!phm_cap_enabled(hwmgr->platform_descriptor.platformCaps, PHM_PlatformCaps_BACO)) + return 0; + + WREG32(0x12074, 0xFFF0003B); + data = RREG32(0x12075); + + if (data == 0x1) { + reg = RREG32_SOC15(NBIF, 0, mmRCC_BIF_STRAP0); + + if (reg & RCC_BIF_STRAP0__STRAP_PX_CAPABLE_MASK) + *cap = true; + } + + return 0; +} + +int smu9_baco_get_state(struct pp_hwmgr *hwmgr, enum BACO_STATE *state) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)(hwmgr->adev); + uint32_t reg; + + reg = RREG32_SOC15(NBIF, 0, mmBACO_CNTL); + + if (reg & BACO_CNTL__BACO_MODE_MASK) + /* gfx has already entered BACO state */ + *state = BACO_STATE_IN; + else + *state = BACO_STATE_OUT; + return 0; +} diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu9_baco.h b/drivers/gpu/drm/amd/powerplay/hwmgr/smu9_baco.h new file mode 100644 index 000000000000..84e90f801ac3 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu9_baco.h @@ -0,0 +1,31 @@ +/* + * Copyright 2018 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __SMU9_BACO_H__ +#define __SMU9_BACO_H__ +#include "hwmgr.h" +#include "common_baco.h" + +extern int smu9_baco_get_capability(struct pp_hwmgr *hwmgr, bool *cap); +extern int smu9_baco_get_state(struct pp_hwmgr *hwmgr, enum BACO_STATE *state); + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu_helper.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu_helper.c index 101c09b212ad..d09690fca452 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu_helper.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu_helper.c @@ -20,6 +20,9 @@ * OTHER DEALINGS IN THE SOFTWARE. * */ + +#include <linux/pci.h> + #include "hwmgr.h" #include "pp_debug.h" #include "ppatomctrl.h" diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/tonga_baco.c b/drivers/gpu/drm/amd/powerplay/hwmgr/tonga_baco.c new file mode 100644 index 000000000000..ea743bea8e29 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/tonga_baco.c @@ -0,0 +1,231 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "tonga_baco.h" + +#include "gmc/gmc_8_1_d.h" +#include "gmc/gmc_8_1_sh_mask.h" + +#include "bif/bif_5_0_d.h" +#include "bif/bif_5_0_sh_mask.h" + +#include "dce/dce_10_0_d.h" +#include "dce/dce_10_0_sh_mask.h" + +#include "smu/smu_7_1_2_d.h" +#include "smu/smu_7_1_2_sh_mask.h" + + +static const struct baco_cmd_entry gpio_tbl[] = +{ + { CMD_WRITE, mmGPIOPAD_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmGPIOPAD_PD_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmGPIOPAD_PU_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmGPIOPAD_MASK, 0, 0, 0, 0xff77ffff }, + { CMD_WRITE, mmDC_GPIO_DVODATA_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmDC_GPIO_DVODATA_MASK, 0, 0, 0, 0xffffffff }, + { CMD_WRITE, mmDC_GPIO_GENERIC_EN, 0, 0, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmDC_GPIO_GENERIC_MASK, 0, 0, 0, 0x03333333 }, + { CMD_WRITE, mmDC_GPIO_SYNCA_EN, 0, 0, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmDC_GPIO_SYNCA_MASK, 0, 0, 0, 0x00001111 } +}; + +static const struct baco_cmd_entry enable_fb_req_rej_tbl[] = +{ + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC0300024 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x1, 0x0, 0, 0x1 }, + { CMD_WRITE, mmBIF_FB_EN, 0, 0, 0, 0x0 } +}; + +static const struct baco_cmd_entry use_bclk_tbl[] = +{ + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL__SPLL_BYPASS_EN_MASK, CG_SPLL_FUNC_CNTL__SPLL_BYPASS_EN__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL_2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL_2__SPLL_BYPASS_CHG_MASK, CG_SPLL_FUNC_CNTL_2__SPLL_BYPASS_CHG__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_STATUS }, + { CMD_WAITFOR, mmGCK_SMC_IND_DATA, CG_SPLL_STATUS__SPLL_CHG_STATUS_MASK, 0, 0xffffffff, 0x2 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL_2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL_2__SPLL_BYPASS_CHG_MASK, CG_SPLL_FUNC_CNTL_2__SPLL_BYPASS_CHG__SHIFT, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL_2__SPLL_CTLREQ_CHG_MASK, CG_SPLL_FUNC_CNTL_2__SPLL_CTLREQ_CHG__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_STATUS }, + { CMD_WAITFOR, mmGCK_SMC_IND_DATA, CG_SPLL_STATUS__SPLL_CHG_STATUS_MASK, 0, 0xffffffff, 0x2 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL_2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL_2__SPLL_CTLREQ_CHG_MASK, CG_SPLL_FUNC_CNTL_2__SPLL_CTLREQ_CHG__SHIFT, 0, 0x0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC0500170 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x4000000, 0x1a, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixMPLL_BYPASSCLK_SEL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL_MASK, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL__SHIFT, 0, 0x2 }, + { CMD_READMODIFYWRITE, mmMPLL_CNTL_MODE, MPLL_CNTL_MODE__MPLL_SW_DIR_CONTROL_MASK, MPLL_CNTL_MODE__MPLL_SW_DIR_CONTROL__SHIFT, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmMPLL_CNTL_MODE, MPLL_CNTL_MODE__MPLL_MCLK_SEL_MASK, MPLL_CNTL_MODE__MPLL_MCLK_SEL__SHIFT, 0, 0x0 } +}; + +static const struct baco_cmd_entry turn_off_plls_tbl[] = +{ + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_SPLL_FUNC_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL__SPLL_RESET_MASK, CG_SPLL_FUNC_CNTL__SPLL_RESET__SHIFT, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_SPLL_FUNC_CNTL__SPLL_PWRON_MASK, CG_SPLL_FUNC_CNTL__SPLL_PWRON__SHIFT, 0, 0x0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, 0xC0500170 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x2000000, 0x19, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, 0x8000000, 0x1b, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmMPLL_CNTL_MODE, MPLL_CNTL_MODE__GLOBAL_MPLL_RESET_MASK, MPLL_CNTL_MODE__GLOBAL_MPLL_RESET__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmMPLL_CONTROL, 0, 0, 0, 0x00000006 }, + { CMD_WRITE, mmMC_IO_RXCNTL_DPHY0_D0, 0, 0, 0, 0x00007740 }, + { CMD_WRITE, mmMC_IO_RXCNTL_DPHY0_D1, 0, 0, 0, 0x00007740 }, + { CMD_WRITE, mmMC_IO_RXCNTL_DPHY1_D0, 0, 0, 0, 0x00007740 }, + { CMD_WRITE, mmMC_IO_RXCNTL_DPHY1_D1, 0, 0, 0, 0x00007740 }, + { CMD_READMODIFYWRITE, mmMCLK_PWRMGT_CNTL, MCLK_PWRMGT_CNTL__MRDCK0_PDNB_MASK, MCLK_PWRMGT_CNTL__MRDCK0_PDNB__SHIFT, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmMCLK_PWRMGT_CNTL, MCLK_PWRMGT_CNTL__MRDCK1_PDNB_MASK, MCLK_PWRMGT_CNTL__MRDCK1_PDNB__SHIFT, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmMC_SEQ_CNTL_2, MC_SEQ_CNTL_2__DRST_PU_MASK, MC_SEQ_CNTL_2__DRST_PU__SHIFT, 0, 0x0 }, + { CMD_READMODIFYWRITE, mmMC_SEQ_CNTL_2, MC_SEQ_CNTL_2__DRST_PD_MASK, MC_SEQ_CNTL_2__DRST_PD__SHIFT, 0, 0x0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_CLKPIN_CNTL_2 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_CLKPIN_CNTL_2__FORCE_BIF_REFCLK_EN_MASK, CG_CLKPIN_CNTL_2__FORCE_BIF_REFCLK_EN__SHIFT, 0, 0x0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixMPLL_BYPASSCLK_SEL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL_MASK, MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL__SHIFT, 0, 0x4 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixMISC_CLK_CTRL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL_MASK, MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL__SHIFT, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, MISC_CLK_CTRL__ZCLK_SEL_MASK, MISC_CLK_CTRL__ZCLK_SEL__SHIFT, 0, 0x1 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixCG_CLKPIN_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, CG_CLKPIN_CNTL__BCLK_AS_XCLK_MASK, CG_CLKPIN_CNTL__BCLK_AS_XCLK__SHIFT, 0, 0x0 }, + { CMD_WRITE, mmGCK_SMC_IND_INDEX, 0, 0, 0, ixTHM_CLK_CNTL }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, THM_CLK_CNTL__CMON_CLK_SEL_MASK, THM_CLK_CNTL__CMON_CLK_SEL__SHIFT, 0, 0x1 }, + { CMD_READMODIFYWRITE, mmGCK_SMC_IND_DATA, THM_CLK_CNTL__TMON_CLK_SEL_MASK, THM_CLK_CNTL__TMON_CLK_SEL__SHIFT, 0, 0x1 } +}; + +static const struct baco_cmd_entry enter_baco_tbl[] = +{ + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_EN_MASK, BACO_CNTL__BACO_EN__SHIFT, 0, 0x01 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BIF_SCLK_SWITCH_MASK, BACO_CNTL__BACO_BIF_SCLK_SWITCH__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_BIF_SCLK_SWITCH_MASK, 0, 5, 0x40000 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BCLK_OFF_MASK, BACO_CNTL__BACO_BCLK_OFF__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_BCLK_OFF_MASK, 0, 5, 0x02 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ISO_DIS_MASK, BACO_CNTL__BACO_ISO_DIS__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_ISO_DIS_MASK, 0, 5, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ANA_ISO_DIS_MASK, BACO_CNTL__BACO_ANA_ISO_DIS__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_ANA_ISO_DIS_MASK, 0, 5, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_POWER_OFF_MASK, BACO_CNTL__BACO_POWER_OFF__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_POWER_OFF_MASK, 0, 5, 0x08 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_MODE_MASK, 0, 0xffffffff, 0x40 } +}; + +#define BACO_CNTL__PWRGOOD_MASK BACO_CNTL__PWRGOOD_GPIO_MASK+BACO_CNTL__PWRGOOD_MEM_MASK+BACO_CNTL__PWRGOOD_DVO_MASK + +static const struct baco_cmd_entry exit_baco_tbl[] = +{ + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_RESET_EN_MASK, BACO_CNTL__BACO_RESET_EN__SHIFT, 0, 0x01 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BCLK_OFF_MASK, BACO_CNTL__BACO_BCLK_OFF__SHIFT, 0, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_POWER_OFF_MASK, BACO_CNTL__BACO_POWER_OFF__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__PWRGOOD_BF_MASK, 0, 0xffffffff, 0x200 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ISO_DIS_MASK, BACO_CNTL__BACO_ISO_DIS__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__PWRGOOD_MASK, 0, 5, 0x1c00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ANA_ISO_DIS_MASK, BACO_CNTL__BACO_ANA_ISO_DIS__SHIFT, 0, 0x01 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BIF_SCLK_SWITCH_MASK, BACO_CNTL__BACO_BIF_SCLK_SWITCH__SHIFT, 0, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_RESET_EN_MASK, BACO_CNTL__BACO_RESET_EN__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__RCU_BIF_CONFIG_DONE_MASK, 0, 5, 0x100 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_EN_MASK, BACO_CNTL__BACO_EN__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_MODE_MASK, 0, 0xffffffff, 0x00 } +}; + +static const struct baco_cmd_entry clean_baco_tbl[] = +{ + { CMD_WRITE, mmBIOS_SCRATCH_6, 0, 0, 0, 0 }, + { CMD_WRITE, mmBIOS_SCRATCH_7, 0, 0, 0, 0 } +}; + +static const struct baco_cmd_entry gpio_tbl_iceland[] = +{ + { CMD_WRITE, mmGPIOPAD_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmGPIOPAD_PD_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmGPIOPAD_PU_EN, 0, 0, 0, 0x0 }, + { CMD_WRITE, mmGPIOPAD_MASK, 0, 0, 0, 0xff77ffff } +}; + +static const struct baco_cmd_entry exit_baco_tbl_iceland[] = +{ + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_RESET_EN_MASK, BACO_CNTL__BACO_RESET_EN__SHIFT, 0, 0x01 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BCLK_OFF_MASK, BACO_CNTL__BACO_BCLK_OFF__SHIFT, 0, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_POWER_OFF_MASK, BACO_CNTL__BACO_POWER_OFF__SHIFT, 0, 0x00 }, + { CMD_DELAY_MS, 0, 0, 0, 20, 0 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__PWRGOOD_BF_MASK, 0, 0xffffffff, 0x200 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ISO_DIS_MASK, BACO_CNTL__BACO_ISO_DIS__SHIFT, 0, 0x01 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__PWRGOOD_MASK, 0, 5, 0x1c00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_ANA_ISO_DIS_MASK, BACO_CNTL__BACO_ANA_ISO_DIS__SHIFT, 0, 0x01 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_BIF_SCLK_SWITCH_MASK, BACO_CNTL__BACO_BIF_SCLK_SWITCH__SHIFT, 0, 0x00 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_RESET_EN_MASK, BACO_CNTL__BACO_RESET_EN__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__RCU_BIF_CONFIG_DONE_MASK, 0, 5, 0x100 }, + { CMD_READMODIFYWRITE, mmBACO_CNTL, BACO_CNTL__BACO_EN_MASK, BACO_CNTL__BACO_EN__SHIFT, 0, 0x00 }, + { CMD_WAITFOR, mmBACO_CNTL, BACO_CNTL__BACO_MODE_MASK, 0, 0xffffffff, 0x00 } +}; + +static const struct baco_cmd_entry clean_baco_tbl_iceland[] = +{ + { CMD_WRITE, mmBIOS_SCRATCH_7, 0, 0, 0, 0 } +}; + +int tonga_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state) +{ + enum BACO_STATE cur_state; + + smu7_baco_get_state(hwmgr, &cur_state); + + if (cur_state == state) + /* aisc already in the target state */ + return 0; + + if (state == BACO_STATE_IN) { + if (hwmgr->chip_id == CHIP_TOPAZ) + baco_program_registers(hwmgr, gpio_tbl_iceland, ARRAY_SIZE(gpio_tbl_iceland)); + else + baco_program_registers(hwmgr, gpio_tbl, ARRAY_SIZE(gpio_tbl)); + baco_program_registers(hwmgr, enable_fb_req_rej_tbl, + ARRAY_SIZE(enable_fb_req_rej_tbl)); + baco_program_registers(hwmgr, use_bclk_tbl, ARRAY_SIZE(use_bclk_tbl)); + baco_program_registers(hwmgr, turn_off_plls_tbl, + ARRAY_SIZE(turn_off_plls_tbl)); + if (baco_program_registers(hwmgr, enter_baco_tbl, + ARRAY_SIZE(enter_baco_tbl))) + return 0; + + } else if (state == BACO_STATE_OUT) { + /* HW requires at least 20ms between regulator off and on */ + msleep(20); + /* Execute Hardware BACO exit sequence */ + if (hwmgr->chip_id == CHIP_TOPAZ) { + if (baco_program_registers(hwmgr, exit_baco_tbl_iceland, + ARRAY_SIZE(exit_baco_tbl_iceland))) { + if (baco_program_registers(hwmgr, clean_baco_tbl_iceland, + ARRAY_SIZE(clean_baco_tbl_iceland))) + return 0; + } + } else { + if (baco_program_registers(hwmgr, exit_baco_tbl, + ARRAY_SIZE(exit_baco_tbl))) { + if (baco_program_registers(hwmgr, clean_baco_tbl, + ARRAY_SIZE(clean_baco_tbl))) + return 0; + } + } + } + + return -EINVAL; +} diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/tonga_baco.h b/drivers/gpu/drm/amd/powerplay/hwmgr/tonga_baco.h new file mode 100644 index 000000000000..5dc16cc8a295 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/tonga_baco.h @@ -0,0 +1,29 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __TONGA_BACO_H__ +#define __TONGA_BACO_H__ +#include "smu7_baco.h" + +extern int tonga_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state); + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_baco.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_baco.c index 7337be5602e4..d168af4a4d78 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_baco.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_baco.c @@ -85,48 +85,11 @@ static const struct soc15_baco_cmd_entry clean_baco_tbl[] = {CMD_WRITE, SOC15_REG_ENTRY(NBIF, 0, mmBIOS_SCRATCH_7), 0, 0, 0, 0}, }; -int vega10_baco_get_capability(struct pp_hwmgr *hwmgr, bool *cap) -{ - struct amdgpu_device *adev = (struct amdgpu_device *)(hwmgr->adev); - uint32_t reg, data; - - *cap = false; - if (!phm_cap_enabled(hwmgr->platform_descriptor.platformCaps, PHM_PlatformCaps_BACO)) - return 0; - - WREG32(0x12074, 0xFFF0003B); - data = RREG32(0x12075); - - if (data == 0x1) { - reg = RREG32_SOC15(NBIF, 0, mmRCC_BIF_STRAP0); - - if (reg & RCC_BIF_STRAP0__STRAP_PX_CAPABLE_MASK) - *cap = true; - } - - return 0; -} - -int vega10_baco_get_state(struct pp_hwmgr *hwmgr, enum BACO_STATE *state) -{ - struct amdgpu_device *adev = (struct amdgpu_device *)(hwmgr->adev); - uint32_t reg; - - reg = RREG32_SOC15(NBIF, 0, mmBACO_CNTL); - - if (reg & BACO_CNTL__BACO_MODE_MASK) - /* gfx has already entered BACO state */ - *state = BACO_STATE_IN; - else - *state = BACO_STATE_OUT; - return 0; -} - int vega10_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state) { enum BACO_STATE cur_state; - vega10_baco_get_state(hwmgr, &cur_state); + smu9_baco_get_state(hwmgr, &cur_state); if (cur_state == state) /* aisc already in the target state */ diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_baco.h b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_baco.h index f7a3ffa744b3..96d793f026a5 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_baco.h +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_baco.h @@ -22,11 +22,8 @@ */ #ifndef __VEGA10_BACO_H__ #define __VEGA10_BACO_H__ -#include "hwmgr.h" -#include "common_baco.h" +#include "smu9_baco.h" -extern int vega10_baco_get_capability(struct pp_hwmgr *hwmgr, bool *cap); -extern int vega10_baco_get_state(struct pp_hwmgr *hwmgr, enum BACO_STATE *state); extern int vega10_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state); #endif diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c index 5c4f701939ea..d71a492c87a3 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c @@ -24,6 +24,7 @@ #include <linux/delay.h> #include <linux/fb.h> #include <linux/module.h> +#include <linux/pci.h> #include <linux/slab.h> #include "hwmgr.h" @@ -356,6 +357,7 @@ static void vega10_init_dpm_defaults(struct pp_hwmgr *hwmgr) struct vega10_hwmgr *data = hwmgr->backend; int i; uint32_t sub_vendor_id, hw_revision; + uint32_t top32, bottom32; struct amdgpu_device *adev = hwmgr->adev; vega10_initialize_power_tune_defaults(hwmgr); @@ -499,6 +501,14 @@ static void vega10_init_dpm_defaults(struct pp_hwmgr *hwmgr) (hw_revision == 0) && (sub_vendor_id != 0x1002)) data->smu_features[GNLD_PCC_LIMIT].supported = true; + + /* Get the SN to turn into a Unique ID */ + smum_send_msg_to_smc(hwmgr, PPSMC_MSG_ReadSerialNumTop32); + top32 = smum_get_argument(hwmgr); + smum_send_msg_to_smc(hwmgr, PPSMC_MSG_ReadSerialNumBottom32); + bottom32 = smum_get_argument(hwmgr); + + adev->unique_id = ((uint64_t)bottom32 << 32) | top32; } #ifdef PPLIB_VEGA10_EVV_SUPPORT @@ -702,7 +712,6 @@ static int vega10_sort_lookup_table(struct pp_hwmgr *hwmgr, struct phm_ppt_v1_voltage_lookup_table *lookup_table) { uint32_t table_size, i, j; - struct phm_ppt_v1_voltage_lookup_record tmp_voltage_lookup_record; PP_ASSERT_WITH_CODE(lookup_table && lookup_table->count, "Lookup table is empty", return -EINVAL); @@ -714,9 +723,8 @@ static int vega10_sort_lookup_table(struct pp_hwmgr *hwmgr, for (j = i + 1; j > 0; j--) { if (lookup_table->entries[j].us_vdd < lookup_table->entries[j - 1].us_vdd) { - tmp_voltage_lookup_record = lookup_table->entries[j - 1]; - lookup_table->entries[j - 1] = lookup_table->entries[j]; - lookup_table->entries[j] = tmp_voltage_lookup_record; + swap(lookup_table->entries[j - 1], + lookup_table->entries[j]); } } } @@ -1427,6 +1435,15 @@ static int vega10_setup_default_dpm_tables(struct pp_hwmgr *hwmgr) vega10_setup_default_pcie_table(hwmgr); + /* Zero out the saved copy of the CUSTOM profile + * This will be checked when trying to set the profile + * and will require that new values be passed in + */ + data->custom_profile_mode[0] = 0; + data->custom_profile_mode[1] = 0; + data->custom_profile_mode[2] = 0; + data->custom_profile_mode[3] = 0; + /* save a copy of the default DPM table */ memcpy(&(data->golden_dpm_table), &(data->dpm_table), sizeof(struct vega10_dpm_table)); @@ -2258,8 +2275,8 @@ static int vega10_populate_avfs_parameters(struct pp_hwmgr *hwmgr) pp_table->AcgAvfsGb.m1 = avfs_params.ulAcgGbFuseTableM1; pp_table->AcgAvfsGb.m2 = avfs_params.ulAcgGbFuseTableM2; pp_table->AcgAvfsGb.b = avfs_params.ulAcgGbFuseTableB; - pp_table->AcgAvfsGb.m1_shift = 0; - pp_table->AcgAvfsGb.m2_shift = 0; + pp_table->AcgAvfsGb.m1_shift = 24; + pp_table->AcgAvfsGb.m2_shift = 12; pp_table->AcgAvfsGb.b_shift = 0; } else { @@ -2355,6 +2372,10 @@ static int vega10_avfs_enable(struct pp_hwmgr *hwmgr, bool enable) struct vega10_hwmgr *data = hwmgr->backend; if (data->smu_features[GNLD_AVFS].supported) { + /* Already enabled or disabled */ + if (!(enable ^ data->smu_features[GNLD_AVFS].enabled)) + return 0; + if (enable) { PP_ASSERT_WITH_CODE(!vega10_enable_smc_features(hwmgr, true, @@ -2457,11 +2478,6 @@ static void vega10_check_dpm_table_updated(struct pp_hwmgr *hwmgr) return; } } - - if (data->need_update_dpm_table & DPMTABLE_OD_UPDATE_VDDC) { - data->need_update_dpm_table &= ~DPMTABLE_OD_UPDATE_VDDC; - data->need_update_dpm_table |= DPMTABLE_OD_UPDATE_SCLK | DPMTABLE_OD_UPDATE_MCLK; - } } /** @@ -3202,7 +3218,8 @@ static int vega10_apply_state_adjust_rules(struct pp_hwmgr *hwmgr, if (hwmgr->display_config->num_display == 0) disable_mclk_switching = false; else - disable_mclk_switching = (hwmgr->display_config->num_display > 1) || + disable_mclk_switching = ((1 < hwmgr->display_config->num_display) && + !hwmgr->display_config->multi_monitor_in_sync) || disable_mclk_switching_for_frame_lock || disable_mclk_switching_for_vr || force_mclk_high; @@ -3672,8 +3689,19 @@ static int vega10_set_power_state_tasks(struct pp_hwmgr *hwmgr, PP_ASSERT_WITH_CODE(!result, "Failed to upload PPtable!", return result); + /* + * If a custom pp table is loaded, set DPMTABLE_OD_UPDATE_VDDC flag. + * That effectively disables AVFS feature. + */ + if(hwmgr->hardcode_pp_table != NULL) + data->need_update_dpm_table |= DPMTABLE_OD_UPDATE_VDDC; + vega10_update_avfs(hwmgr); + /* + * Clear all OD flags except DPMTABLE_OD_UPDATE_VDDC. + * That will help to keep AVFS disabled. + */ data->need_update_dpm_table &= DPMTABLE_OD_UPDATE_VDDC; return 0; @@ -3776,6 +3804,18 @@ static int vega10_read_sensor(struct pp_hwmgr *hwmgr, int idx, *((uint32_t *)value) = vega10_thermal_get_temperature(hwmgr); *size = 4; break; + case AMDGPU_PP_SENSOR_HOTSPOT_TEMP: + smum_send_msg_to_smc(hwmgr, PPSMC_MSG_GetTemperatureHotspot); + *((uint32_t *)value) = smum_get_argument(hwmgr) * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + *size = 4; + break; + case AMDGPU_PP_SENSOR_MEM_TEMP: + smum_send_msg_to_smc(hwmgr, PPSMC_MSG_GetTemperatureHBM); + *((uint32_t *)value) = smum_get_argument(hwmgr) * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + *size = 4; + break; case AMDGPU_PP_SENSOR_UVD_POWER: *((uint32_t *)value) = data->uvd_power_gated ? 0 : 1; *size = 4; @@ -4843,12 +4883,22 @@ static int vega10_notify_cac_buffer_info(struct pp_hwmgr *hwmgr, static int vega10_get_thermal_temperature_range(struct pp_hwmgr *hwmgr, struct PP_TemperatureRange *thermal_data) { - struct phm_ppt_v2_information *table_info = - (struct phm_ppt_v2_information *)hwmgr->pptable; + struct vega10_hwmgr *data = hwmgr->backend; + PPTable_t *pp_table = &(data->smc_state_table.pp_table); memcpy(thermal_data, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange)); - thermal_data->max = table_info->tdp_table->usSoftwareShutdownTemp * + thermal_data->max = pp_table->TedgeLimit * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->edge_emergency_max = (pp_table->TedgeLimit + CTF_OFFSET_EDGE) * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->hotspot_crit_max = pp_table->ThotspotLimit * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->hotspot_emergency_max = (pp_table->ThotspotLimit + CTF_OFFSET_HOTSPOT) * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->mem_crit_max = pp_table->ThbmLimit * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->mem_emergency_max = (pp_table->ThbmLimit + CTF_OFFSET_HBM)* PP_TEMPERATURE_UNITS_PER_CENTIGRADES; return 0; @@ -4904,16 +4954,23 @@ static int vega10_set_power_profile_mode(struct pp_hwmgr *hwmgr, long *input, ui uint8_t FPS; uint8_t use_rlc_busy; uint8_t min_active_level; + uint32_t power_profile_mode = input[size]; - hwmgr->power_profile_mode = input[size]; - - smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetWorkloadMask, - 1<<hwmgr->power_profile_mode); - - if (hwmgr->power_profile_mode == PP_SMC_POWER_PROFILE_CUSTOM) { - if (size == 0 || size > 4) + if (power_profile_mode == PP_SMC_POWER_PROFILE_CUSTOM) { + if (size != 0 && size != 4) return -EINVAL; + /* If size = 0 and the CUSTOM profile has been set already + * then just apply the profile. The copy stored in the hwmgr + * is zeroed out on init + */ + if (size == 0) { + if (data->custom_profile_mode[0] != 0) + goto out; + else + return -EINVAL; + } + data->custom_profile_mode[0] = busy_set_point = input[0]; data->custom_profile_mode[1] = FPS = input[1]; data->custom_profile_mode[2] = use_rlc_busy = input[2]; @@ -4924,6 +4981,11 @@ static int vega10_set_power_profile_mode(struct pp_hwmgr *hwmgr, long *input, ui use_rlc_busy << 16 | min_active_level<<24); } +out: + smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetWorkloadMask, + 1 << power_profile_mode); + hwmgr->power_profile_mode = power_profile_mode; + return 0; } @@ -4967,13 +5029,70 @@ static bool vega10_check_clk_voltage_valid(struct pp_hwmgr *hwmgr, return true; } +static void vega10_odn_update_power_state(struct pp_hwmgr *hwmgr) +{ + struct vega10_hwmgr *data = hwmgr->backend; + struct pp_power_state *ps = hwmgr->request_ps; + struct vega10_power_state *vega10_ps; + struct vega10_single_dpm_table *gfx_dpm_table = + &data->dpm_table.gfx_table; + struct vega10_single_dpm_table *soc_dpm_table = + &data->dpm_table.soc_table; + struct vega10_single_dpm_table *mem_dpm_table = + &data->dpm_table.mem_table; + int max_level; + + if (!ps) + return; + + vega10_ps = cast_phw_vega10_power_state(&ps->hardware); + max_level = vega10_ps->performance_level_count - 1; + + if (vega10_ps->performance_levels[max_level].gfx_clock != + gfx_dpm_table->dpm_levels[gfx_dpm_table->count - 1].value) + vega10_ps->performance_levels[max_level].gfx_clock = + gfx_dpm_table->dpm_levels[gfx_dpm_table->count - 1].value; + + if (vega10_ps->performance_levels[max_level].soc_clock != + soc_dpm_table->dpm_levels[soc_dpm_table->count - 1].value) + vega10_ps->performance_levels[max_level].soc_clock = + soc_dpm_table->dpm_levels[soc_dpm_table->count - 1].value; + + if (vega10_ps->performance_levels[max_level].mem_clock != + mem_dpm_table->dpm_levels[mem_dpm_table->count - 1].value) + vega10_ps->performance_levels[max_level].mem_clock = + mem_dpm_table->dpm_levels[mem_dpm_table->count - 1].value; + + if (!hwmgr->ps) + return; + + ps = (struct pp_power_state *)((unsigned long)(hwmgr->ps) + hwmgr->ps_size * (hwmgr->num_ps - 1)); + vega10_ps = cast_phw_vega10_power_state(&ps->hardware); + max_level = vega10_ps->performance_level_count - 1; + + if (vega10_ps->performance_levels[max_level].gfx_clock != + gfx_dpm_table->dpm_levels[gfx_dpm_table->count - 1].value) + vega10_ps->performance_levels[max_level].gfx_clock = + gfx_dpm_table->dpm_levels[gfx_dpm_table->count - 1].value; + + if (vega10_ps->performance_levels[max_level].soc_clock != + soc_dpm_table->dpm_levels[soc_dpm_table->count - 1].value) + vega10_ps->performance_levels[max_level].soc_clock = + soc_dpm_table->dpm_levels[soc_dpm_table->count - 1].value; + + if (vega10_ps->performance_levels[max_level].mem_clock != + mem_dpm_table->dpm_levels[mem_dpm_table->count - 1].value) + vega10_ps->performance_levels[max_level].mem_clock = + mem_dpm_table->dpm_levels[mem_dpm_table->count - 1].value; +} + static void vega10_odn_update_soc_table(struct pp_hwmgr *hwmgr, enum PP_OD_DPM_TABLE_COMMAND type) { struct vega10_hwmgr *data = hwmgr->backend; struct phm_ppt_v2_information *table_info = hwmgr->pptable; struct phm_ppt_v1_clock_voltage_dependency_table *dep_table = table_info->vdd_dep_on_socclk; - struct vega10_single_dpm_table *dpm_table = &data->golden_dpm_table.soc_table; + struct vega10_single_dpm_table *dpm_table = &data->golden_dpm_table.mem_table; struct vega10_odn_clock_voltage_dependency_table *podn_vdd_dep_on_socclk = &data->odn_dpm_table.vdd_dep_on_socclk; @@ -4984,9 +5103,7 @@ static void vega10_odn_update_soc_table(struct pp_hwmgr *hwmgr, if (type == PP_OD_EDIT_SCLK_VDDC_TABLE) { podn_vdd_dep = &data->odn_dpm_table.vdd_dep_on_sclk; - for (i = 0; i < podn_vdd_dep->count - 1; i++) - od_vddc_lookup_table->entries[i].us_vdd = podn_vdd_dep->entries[i].vddc; - if (od_vddc_lookup_table->entries[i].us_vdd < podn_vdd_dep->entries[i].vddc) + for (i = 0; i < podn_vdd_dep->count; i++) od_vddc_lookup_table->entries[i].us_vdd = podn_vdd_dep->entries[i].vddc; } else if (type == PP_OD_EDIT_MCLK_VDDC_TABLE) { podn_vdd_dep = &data->odn_dpm_table.vdd_dep_on_mclk; @@ -4997,7 +5114,8 @@ static void vega10_odn_update_soc_table(struct pp_hwmgr *hwmgr, break; } if (j == od_vddc_lookup_table->count) { - od_vddc_lookup_table->entries[j-1].us_vdd = + j = od_vddc_lookup_table->count - 1; + od_vddc_lookup_table->entries[j].us_vdd = podn_vdd_dep->entries[i].vddc; data->need_update_dpm_table |= DPMTABLE_OD_UPDATE_VDDC; } @@ -5005,25 +5123,38 @@ static void vega10_odn_update_soc_table(struct pp_hwmgr *hwmgr, } dpm_table = &data->dpm_table.soc_table; for (i = 0; i < dep_table->count; i++) { - if (dep_table->entries[i].vddInd == podn_vdd_dep->entries[dep_table->count-1].vddInd && - dep_table->entries[i].clk < podn_vdd_dep->entries[dep_table->count-1].clk) { + if (dep_table->entries[i].vddInd == podn_vdd_dep->entries[podn_vdd_dep->count-1].vddInd && + dep_table->entries[i].clk < podn_vdd_dep->entries[podn_vdd_dep->count-1].clk) { data->need_update_dpm_table |= DPMTABLE_UPDATE_SOCCLK; - podn_vdd_dep_on_socclk->entries[i].clk = podn_vdd_dep->entries[dep_table->count-1].clk; - dpm_table->dpm_levels[i].value = podn_vdd_dep_on_socclk->entries[i].clk; + for (; (i < dep_table->count) && + (dep_table->entries[i].clk < podn_vdd_dep->entries[podn_vdd_dep->count - 1].clk); i++) { + podn_vdd_dep_on_socclk->entries[i].clk = podn_vdd_dep->entries[podn_vdd_dep->count-1].clk; + dpm_table->dpm_levels[i].value = podn_vdd_dep_on_socclk->entries[i].clk; + } + break; + } else { + dpm_table->dpm_levels[i].value = dep_table->entries[i].clk; + podn_vdd_dep_on_socclk->entries[i].vddc = dep_table->entries[i].vddc; + podn_vdd_dep_on_socclk->entries[i].vddInd = dep_table->entries[i].vddInd; + podn_vdd_dep_on_socclk->entries[i].clk = dep_table->entries[i].clk; } } if (podn_vdd_dep_on_socclk->entries[podn_vdd_dep_on_socclk->count - 1].clk < - podn_vdd_dep->entries[dep_table->count-1].clk) { + podn_vdd_dep->entries[podn_vdd_dep->count - 1].clk) { data->need_update_dpm_table |= DPMTABLE_UPDATE_SOCCLK; - podn_vdd_dep_on_socclk->entries[podn_vdd_dep_on_socclk->count - 1].clk = podn_vdd_dep->entries[dep_table->count-1].clk; - dpm_table->dpm_levels[podn_vdd_dep_on_socclk->count - 1].value = podn_vdd_dep->entries[dep_table->count-1].clk; + podn_vdd_dep_on_socclk->entries[podn_vdd_dep_on_socclk->count - 1].clk = + podn_vdd_dep->entries[podn_vdd_dep->count - 1].clk; + dpm_table->dpm_levels[podn_vdd_dep_on_socclk->count - 1].value = + podn_vdd_dep->entries[podn_vdd_dep->count - 1].clk; } if (podn_vdd_dep_on_socclk->entries[podn_vdd_dep_on_socclk->count - 1].vddInd < - podn_vdd_dep->entries[dep_table->count-1].vddInd) { + podn_vdd_dep->entries[podn_vdd_dep->count - 1].vddInd) { data->need_update_dpm_table |= DPMTABLE_UPDATE_SOCCLK; - podn_vdd_dep_on_socclk->entries[podn_vdd_dep_on_socclk->count - 1].vddInd = podn_vdd_dep->entries[dep_table->count-1].vddInd; + podn_vdd_dep_on_socclk->entries[podn_vdd_dep_on_socclk->count - 1].vddInd = + podn_vdd_dep->entries[podn_vdd_dep->count - 1].vddInd; } } + vega10_odn_update_power_state(hwmgr); } static int vega10_odn_edit_dpm_table(struct pp_hwmgr *hwmgr, @@ -5058,6 +5189,11 @@ static int vega10_odn_edit_dpm_table(struct pp_hwmgr *hwmgr, } else if (PP_OD_RESTORE_DEFAULT_TABLE == type) { memcpy(&(data->dpm_table), &(data->golden_dpm_table), sizeof(struct vega10_dpm_table)); vega10_odn_initial_default_setting(hwmgr); + vega10_odn_update_power_state(hwmgr); + /* force to update all clock tables */ + data->need_update_dpm_table = DPMTABLE_UPDATE_SCLK | + DPMTABLE_UPDATE_MCLK | + DPMTABLE_UPDATE_SOCCLK; return 0; } else if (PP_OD_COMMIT_DPM_TABLE == type) { vega10_check_dpm_table_updated(hwmgr); @@ -5087,6 +5223,30 @@ static int vega10_odn_edit_dpm_table(struct pp_hwmgr *hwmgr, return 0; } +static int vega10_set_mp1_state(struct pp_hwmgr *hwmgr, + enum pp_mp1_state mp1_state) +{ + uint16_t msg; + int ret; + + switch (mp1_state) { + case PP_MP1_STATE_UNLOAD: + msg = PPSMC_MSG_PrepareMp1ForUnload; + break; + case PP_MP1_STATE_SHUTDOWN: + case PP_MP1_STATE_RESET: + case PP_MP1_STATE_NONE: + default: + return 0; + } + + PP_ASSERT_WITH_CODE((ret = smum_send_msg_to_smc(hwmgr, msg)) == 0, + "[PrepareMp1] Failed!", + return ret); + + return 0; +} + static int vega10_get_performance_level(struct pp_hwmgr *hwmgr, const struct pp_hw_power_state *state, PHM_PerformanceLevelDesignation designation, uint32_t index, PHM_PerformanceLevel *level) @@ -5110,6 +5270,59 @@ static int vega10_get_performance_level(struct pp_hwmgr *hwmgr, const struct pp_ return 0; } +static int vega10_disable_power_features_for_compute_performance(struct pp_hwmgr *hwmgr, bool disable) +{ + struct vega10_hwmgr *data = hwmgr->backend; + uint32_t feature_mask = 0; + + if (disable) { + feature_mask |= data->smu_features[GNLD_ULV].enabled ? + data->smu_features[GNLD_ULV].smu_feature_bitmap : 0; + feature_mask |= data->smu_features[GNLD_DS_GFXCLK].enabled ? + data->smu_features[GNLD_DS_GFXCLK].smu_feature_bitmap : 0; + feature_mask |= data->smu_features[GNLD_DS_SOCCLK].enabled ? + data->smu_features[GNLD_DS_SOCCLK].smu_feature_bitmap : 0; + feature_mask |= data->smu_features[GNLD_DS_LCLK].enabled ? + data->smu_features[GNLD_DS_LCLK].smu_feature_bitmap : 0; + feature_mask |= data->smu_features[GNLD_DS_DCEFCLK].enabled ? + data->smu_features[GNLD_DS_DCEFCLK].smu_feature_bitmap : 0; + } else { + feature_mask |= (!data->smu_features[GNLD_ULV].enabled) ? + data->smu_features[GNLD_ULV].smu_feature_bitmap : 0; + feature_mask |= (!data->smu_features[GNLD_DS_GFXCLK].enabled) ? + data->smu_features[GNLD_DS_GFXCLK].smu_feature_bitmap : 0; + feature_mask |= (!data->smu_features[GNLD_DS_SOCCLK].enabled) ? + data->smu_features[GNLD_DS_SOCCLK].smu_feature_bitmap : 0; + feature_mask |= (!data->smu_features[GNLD_DS_LCLK].enabled) ? + data->smu_features[GNLD_DS_LCLK].smu_feature_bitmap : 0; + feature_mask |= (!data->smu_features[GNLD_DS_DCEFCLK].enabled) ? + data->smu_features[GNLD_DS_DCEFCLK].smu_feature_bitmap : 0; + } + + if (feature_mask) + PP_ASSERT_WITH_CODE(!vega10_enable_smc_features(hwmgr, + !disable, feature_mask), + "enable/disable power features for compute performance Failed!", + return -EINVAL); + + if (disable) { + data->smu_features[GNLD_ULV].enabled = false; + data->smu_features[GNLD_DS_GFXCLK].enabled = false; + data->smu_features[GNLD_DS_SOCCLK].enabled = false; + data->smu_features[GNLD_DS_LCLK].enabled = false; + data->smu_features[GNLD_DS_DCEFCLK].enabled = false; + } else { + data->smu_features[GNLD_ULV].enabled = true; + data->smu_features[GNLD_DS_GFXCLK].enabled = true; + data->smu_features[GNLD_DS_SOCCLK].enabled = true; + data->smu_features[GNLD_DS_LCLK].enabled = true; + data->smu_features[GNLD_DS_DCEFCLK].enabled = true; + } + + return 0; + +} + static const struct pp_hwmgr_func vega10_hwmgr_funcs = { .backend_init = vega10_hwmgr_backend_init, .backend_fini = vega10_hwmgr_backend_fini, @@ -5170,18 +5383,25 @@ static const struct pp_hwmgr_func vega10_hwmgr_funcs = { .set_power_limit = vega10_set_power_limit, .odn_edit_dpm_table = vega10_odn_edit_dpm_table, .get_performance_level = vega10_get_performance_level, - .get_asic_baco_capability = vega10_baco_get_capability, - .get_asic_baco_state = vega10_baco_get_state, + .get_asic_baco_capability = smu9_baco_get_capability, + .get_asic_baco_state = smu9_baco_get_state, .set_asic_baco_state = vega10_baco_set_state, .enable_mgpu_fan_boost = vega10_enable_mgpu_fan_boost, .get_ppfeature_status = vega10_get_ppfeature_status, .set_ppfeature_status = vega10_set_ppfeature_status, + .set_mp1_state = vega10_set_mp1_state, + .disable_power_features_for_compute_performance = + vega10_disable_power_features_for_compute_performance, }; int vega10_hwmgr_init(struct pp_hwmgr *hwmgr) { + struct amdgpu_device *adev = hwmgr->adev; + hwmgr->hwmgr_func = &vega10_hwmgr_funcs; hwmgr->pptable_func = &vega10_pptable_funcs; + if (amdgpu_passthrough(adev)) + return vega10_baco_set_cap(hwmgr); return 0; } diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_processpptables.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_processpptables.c index b6767d74dc85..f29af5ca0aa0 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_processpptables.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_processpptables.c @@ -21,6 +21,7 @@ * */ #include <linux/module.h> +#include <linux/pci.h> #include <linux/slab.h> #include <linux/fb.h> @@ -1371,3 +1372,27 @@ int vega10_get_powerplay_table_entry(struct pp_hwmgr *hwmgr, return result; } + +int vega10_baco_set_cap(struct pp_hwmgr *hwmgr) +{ + int result = 0; + + const ATOM_Vega10_POWERPLAYTABLE *powerplay_table; + + powerplay_table = get_powerplay_table(hwmgr); + + PP_ASSERT_WITH_CODE((powerplay_table != NULL), + "Missing PowerPlay Table!", return -1); + + result = check_powerplay_tables(hwmgr, powerplay_table); + + PP_ASSERT_WITH_CODE((result == 0), + "check_powerplay_tables failed", return result); + + set_hw_cap( + hwmgr, + 0 != (le32_to_cpu(powerplay_table->ulPlatformCaps) & ATOM_VEGA10_PP_PLATFORM_CAP_BACO), + PHM_PlatformCaps_BACO); + return result; +} + diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_processpptables.h b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_processpptables.h index d83ed2af7aa3..da5fbec9b0cd 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_processpptables.h +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_processpptables.h @@ -59,4 +59,5 @@ extern int vega10_get_number_of_powerplay_table_entries(struct pp_hwmgr *hwmgr); extern int vega10_get_powerplay_table_entry(struct pp_hwmgr *hwmgr, uint32_t entry_index, struct pp_power_state *power_state, int (*call_back_func)(struct pp_hwmgr *, void *, struct pp_power_state *, void *, uint32_t)); +extern int vega10_baco_set_cap(struct pp_hwmgr *hwmgr); #endif diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_baco.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_baco.c new file mode 100644 index 000000000000..9d8ca94a8f0c --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_baco.c @@ -0,0 +1,119 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "soc15.h" +#include "soc15_hw_ip.h" +#include "vega10_ip_offset.h" +#include "soc15_common.h" +#include "vega12_inc.h" +#include "vega12_ppsmc.h" +#include "vega12_baco.h" + +static const struct soc15_baco_cmd_entry pre_baco_tbl[] = +{ + { CMD_READMODIFYWRITE, NBIF_HWID, 0, mmBIF_DOORBELL_CNTL_BASE_IDX, mmBIF_DOORBELL_CNTL, BIF_DOORBELL_CNTL__DOORBELL_MONITOR_EN_MASK, BIF_DOORBELL_CNTL__DOORBELL_MONITOR_EN__SHIFT, 0, 0 }, + { CMD_WRITE, NBIF_HWID, 0, mmBIF_FB_EN_BASE_IDX, mmBIF_FB_EN, 0, 0, 0, 0 }, + { CMD_READMODIFYWRITE, NBIF_HWID, 0, mmRCC_BACO_CNTL_MISC_BASE_IDX, mmBACO_CNTL, BACO_CNTL__BACO_DSTATE_BYPASS_MASK, BACO_CNTL__BACO_DSTATE_BYPASS__SHIFT, 0, 1 }, + { CMD_READMODIFYWRITE, NBIF_HWID, 0, mmRCC_BACO_CNTL_MISC_BASE_IDX, mmBACO_CNTL, BACO_CNTL__BACO_RST_INTR_MASK_MASK, BACO_CNTL__BACO_RST_INTR_MASK__SHIFT, 0, 1 } +}; + +static const struct soc15_baco_cmd_entry enter_baco_tbl[] = +{ + { CMD_WAITFOR, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__SOC_DOMAIN_IDLE_MASK, THM_BACO_CNTL__SOC_DOMAIN_IDLE__SHIFT, 0xffffffff, 0x80000000 }, + { CMD_READMODIFYWRITE, NBIF_HWID, 0, mmRCC_BACO_CNTL_MISC_BASE_IDX, mmBACO_CNTL, BACO_CNTL__BACO_EN_MASK, BACO_CNTL__BACO_EN__SHIFT, 0, 1 }, + { CMD_READMODIFYWRITE, NBIF_HWID, 0, mmRCC_BACO_CNTL_MISC_BASE_IDX, mmBACO_CNTL, BACO_CNTL__BACO_BIF_LCLK_SWITCH_MASK, BACO_CNTL__BACO_BIF_LCLK_SWITCH__SHIFT, 0, 1 }, + { CMD_READMODIFYWRITE, NBIF_HWID, 0, mmRCC_BACO_CNTL_MISC_BASE_IDX, mmBACO_CNTL, BACO_CNTL__BACO_DUMMY_EN_MASK, BACO_CNTL__BACO_DUMMY_EN__SHIFT, 0, 1 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_SOC_VDCI_RESET_MASK, THM_BACO_CNTL__BACO_SOC_VDCI_RESET__SHIFT, 0, 1 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_SMNCLK_MUX_MASK, THM_BACO_CNTL__BACO_SMNCLK_MUX__SHIFT, 0, 1 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_ISO_EN_MASK, THM_BACO_CNTL__BACO_ISO_EN__SHIFT, 0, 1 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_AEB_ISO_EN_MASK, THM_BACO_CNTL__BACO_AEB_ISO_EN__SHIFT, 0, 1 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_ANA_ISO_EN_MASK, THM_BACO_CNTL__BACO_ANA_ISO_EN__SHIFT, 0, 1 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_SOC_REFCLK_OFF_MASK, THM_BACO_CNTL__BACO_SOC_REFCLK_OFF__SHIFT, 0, 1 }, + { CMD_READMODIFYWRITE, NBIF_HWID, 0, mmRCC_BACO_CNTL_MISC_BASE_IDX, mmBACO_CNTL, BACO_CNTL__BACO_POWER_OFF_MASK, BACO_CNTL__BACO_POWER_OFF__SHIFT, 0, 1 }, + { CMD_DELAY_MS, 0, 0, 0, 5, 0 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_RESET_EN_MASK, THM_BACO_CNTL__BACO_RESET_EN__SHIFT, 0, 1 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_PWROKRAW_CNTL_MASK, THM_BACO_CNTL__BACO_PWROKRAW_CNTL__SHIFT, 0, 0 }, + { CMD_WAITFOR, NBIF_HWID, 0, mmRCC_BACO_CNTL_MISC_BASE_IDX, mmBACO_CNTL, BACO_CNTL__BACO_MODE_MASK, BACO_CNTL__BACO_MODE__SHIFT, 0xffffffff, 0x100 } +}; + +static const struct soc15_baco_cmd_entry exit_baco_tbl[] = +{ + { CMD_READMODIFYWRITE, NBIF_HWID, 0, mmRCC_BACO_CNTL_MISC_BASE_IDX, mmBACO_CNTL, BACO_CNTL__BACO_POWER_OFF_MASK, BACO_CNTL__BACO_POWER_OFF__SHIFT, 0, 0 }, + { CMD_DELAY_MS, 0, 0, 0, 0, 0, 0, 10, 0 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_SOC_REFCLK_OFF_MASK, THM_BACO_CNTL__BACO_SOC_REFCLK_OFF__SHIFT, 0, 0 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_ANA_ISO_EN_MASK, THM_BACO_CNTL__BACO_ANA_ISO_EN__SHIFT, 0, 0 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_AEB_ISO_EN_MASK, THM_BACO_CNTL__BACO_AEB_ISO_EN__SHIFT, 0, 0 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_ISO_EN_MASK, THM_BACO_CNTL__BACO_ISO_EN__SHIFT, 0, 0 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_PWROKRAW_CNTL_MASK, THM_BACO_CNTL__BACO_PWROKRAW_CNTL__SHIFT, 0, 1 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_SMNCLK_MUX_MASK, THM_BACO_CNTL__BACO_SMNCLK_MUX__SHIFT, 0, 0 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_SOC_VDCI_RESET_MASK, THM_BACO_CNTL__BACO_SOC_VDCI_RESET__SHIFT, 0, 0 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_EXIT_MASK, THM_BACO_CNTL__BACO_EXIT__SHIFT, 0, 1 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_RESET_EN_MASK, THM_BACO_CNTL__BACO_RESET_EN__SHIFT, 0, 0 }, + { CMD_WAITFOR, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_EXIT_MASK, 0, 0xffffffff, 0 }, + { CMD_READMODIFYWRITE, THM_HWID, 0, mmTHM_BACO_CNTL_BASE_IDX, mmTHM_BACO_CNTL, THM_BACO_CNTL__BACO_SB_AXI_FENCE_MASK, THM_BACO_CNTL__BACO_SB_AXI_FENCE__SHIFT, 0, 0 }, + { CMD_READMODIFYWRITE, NBIF_HWID, 0, mmRCC_BACO_CNTL_MISC_BASE_IDX, mmBACO_CNTL, BACO_CNTL__BACO_DUMMY_EN_MASK, BACO_CNTL__BACO_DUMMY_EN__SHIFT, 0, 0 }, + { CMD_READMODIFYWRITE, NBIF_HWID, 0, mmRCC_BACO_CNTL_MISC_BASE_IDX, mmBACO_CNTL, BACO_CNTL__BACO_BIF_LCLK_SWITCH_MASK, BACO_CNTL__BACO_BIF_LCLK_SWITCH__SHIFT, 0, 0 }, + { CMD_READMODIFYWRITE, NBIF_HWID, 0, mmRCC_BACO_CNTL_MISC_BASE_IDX, mmBACO_CNTL, BACO_CNTL__BACO_EN_MASK, BACO_CNTL__BACO_EN__SHIFT, 0, 0 }, + { CMD_WAITFOR, NBIF_HWID, 0, mmRCC_BACO_CNTL_MISC_BASE_IDX, mmBACO_CNTL, BACO_CNTL__BACO_MODE_MASK, 0, 0xffffffff, 0 } +}; + +static const struct soc15_baco_cmd_entry clean_baco_tbl[] = +{ + { CMD_WRITE, NBIF_HWID, 0, mmBIOS_SCRATCH_6_BASE_IDX, mmBIOS_SCRATCH_6, 0, 0, 0, 0 }, + { CMD_WRITE, NBIF_HWID, 0, mmBIOS_SCRATCH_7_BASE_IDX, mmBIOS_SCRATCH_7, 0, 0, 0, 0 } +}; + +int vega12_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state) +{ + enum BACO_STATE cur_state; + + smu9_baco_get_state(hwmgr, &cur_state); + + if (cur_state == state) + /* aisc already in the target state */ + return 0; + + if (state == BACO_STATE_IN) { + if (soc15_baco_program_registers(hwmgr, pre_baco_tbl, + ARRAY_SIZE(pre_baco_tbl))) { + if (smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_EnterBaco, 0)) + return -EINVAL; + + if (soc15_baco_program_registers(hwmgr, enter_baco_tbl, + ARRAY_SIZE(enter_baco_tbl))) + return 0; + } + } else if (state == BACO_STATE_OUT) { + /* HW requires at least 20ms between regulator off and on */ + msleep(20); + /* Execute Hardware BACO exit sequence */ + if (soc15_baco_program_registers(hwmgr, exit_baco_tbl, + ARRAY_SIZE(exit_baco_tbl))) { + if (soc15_baco_program_registers(hwmgr, clean_baco_tbl, + ARRAY_SIZE(clean_baco_tbl))) + return 0; + } + } + + return -EINVAL; +} diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_baco.h b/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_baco.h new file mode 100644 index 000000000000..57b72e5a95ae --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_baco.h @@ -0,0 +1,29 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __VEGA12_BACO_H__ +#define __VEGA12_BACO_H__ +#include "smu9_baco.h" + +extern int vega12_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state); + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c index bdb48e94eff6..7af9ad450ac4 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c @@ -45,6 +45,7 @@ #include "ppinterrupt.h" #include "pp_overdriver.h" #include "pp_thermal.h" +#include "vega12_baco.h" static int vega12_force_clock_level(struct pp_hwmgr *hwmgr, @@ -288,6 +289,8 @@ static int vega12_set_features_platform_caps(struct pp_hwmgr *hwmgr) static void vega12_init_dpm_defaults(struct pp_hwmgr *hwmgr) { struct vega12_hwmgr *data = (struct vega12_hwmgr *)(hwmgr->backend); + struct amdgpu_device *adev = hwmgr->adev; + uint32_t top32, bottom32; int i; data->smu_features[GNLD_DPM_PREFETCHER].smu_feature_id = @@ -352,6 +355,14 @@ static void vega12_init_dpm_defaults(struct pp_hwmgr *hwmgr) ((data->registry_data.disallowed_features >> i) & 1) ? false : true; } + + /* Get the SN to turn into a Unique ID */ + smum_send_msg_to_smc(hwmgr, PPSMC_MSG_ReadSerialNumTop32); + top32 = smum_get_argument(hwmgr); + smum_send_msg_to_smc(hwmgr, PPSMC_MSG_ReadSerialNumBottom32); + bottom32 = smum_get_argument(hwmgr); + + adev->unique_id = ((uint64_t)bottom32 << 32) | top32; } static int vega12_set_private_data_based_on_pptable(struct pp_hwmgr *hwmgr) @@ -1236,21 +1247,39 @@ static uint32_t vega12_dpm_get_mclk(struct pp_hwmgr *hwmgr, bool low) return (mem_clk * 100); } +static int vega12_get_metrics_table(struct pp_hwmgr *hwmgr, SmuMetrics_t *metrics_table) +{ + struct vega12_hwmgr *data = + (struct vega12_hwmgr *)(hwmgr->backend); + int ret = 0; + + if (!data->metrics_time || time_after(jiffies, data->metrics_time + HZ / 2)) { + ret = smum_smc_table_manager(hwmgr, (uint8_t *)metrics_table, + TABLE_SMU_METRICS, true); + if (ret) { + pr_info("Failed to export SMU metrics table!\n"); + return ret; + } + memcpy(&data->metrics_table, metrics_table, sizeof(SmuMetrics_t)); + data->metrics_time = jiffies; + } else + memcpy(metrics_table, &data->metrics_table, sizeof(SmuMetrics_t)); + + return ret; +} + static int vega12_get_gpu_power(struct pp_hwmgr *hwmgr, uint32_t *query) { -#if 0 - uint32_t value; + SmuMetrics_t metrics_table; + int ret = 0; - PP_ASSERT_WITH_CODE(!smum_send_msg_to_smc(hwmgr, - PPSMC_MSG_GetCurrPkgPwr), - "Failed to get current package power!", - return -EINVAL); + ret = vega12_get_metrics_table(hwmgr, &metrics_table); + if (ret) + return ret; - value = smum_get_argument(hwmgr); - /* power value is an integer */ - *query = value << 8; -#endif - return 0; + *query = metrics_table.CurrSocketPower << 8; + + return ret; } static int vega12_get_current_gfx_clk_freq(struct pp_hwmgr *hwmgr, uint32_t *gfx_freq) @@ -1289,25 +1318,27 @@ static int vega12_get_current_mclk_freq(struct pp_hwmgr *hwmgr, uint32_t *mclk_f static int vega12_get_current_activity_percent( struct pp_hwmgr *hwmgr, + int idx, uint32_t *activity_percent) { + SmuMetrics_t metrics_table; int ret = 0; - uint32_t current_activity = 50; -#if 0 - ret = smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_GetAverageGfxActivity, 0); - if (!ret) { - current_activity = smum_get_argument(hwmgr); - if (current_activity > 100) { - PP_ASSERT(false, - "[GetCurrentActivityPercent] Activity Percentage Exceeds 100!"); - current_activity = 100; - } - } else - PP_ASSERT(false, - "[GetCurrentActivityPercent] Attempt To Send Get Average Graphics Activity to SMU Failed!"); -#endif - *activity_percent = current_activity; + ret = vega12_get_metrics_table(hwmgr, &metrics_table); + if (ret) + return ret; + + switch (idx) { + case AMDGPU_PP_SENSOR_GPU_LOAD: + *activity_percent = metrics_table.AverageGfxActivity; + break; + case AMDGPU_PP_SENSOR_MEM_LOAD: + *activity_percent = metrics_table.AverageUclkActivity; + break; + default: + pr_err("Invalid index for retrieving clock activity\n"); + return -EINVAL; + } return ret; } @@ -1316,6 +1347,7 @@ static int vega12_read_sensor(struct pp_hwmgr *hwmgr, int idx, void *value, int *size) { struct vega12_hwmgr *data = (struct vega12_hwmgr *)(hwmgr->backend); + SmuMetrics_t metrics_table; int ret = 0; switch (idx) { @@ -1330,7 +1362,8 @@ static int vega12_read_sensor(struct pp_hwmgr *hwmgr, int idx, *size = 4; break; case AMDGPU_PP_SENSOR_GPU_LOAD: - ret = vega12_get_current_activity_percent(hwmgr, (uint32_t *)value); + case AMDGPU_PP_SENSOR_MEM_LOAD: + ret = vega12_get_current_activity_percent(hwmgr, idx, (uint32_t *)value); if (!ret) *size = 4; break; @@ -1338,6 +1371,24 @@ static int vega12_read_sensor(struct pp_hwmgr *hwmgr, int idx, *((uint32_t *)value) = vega12_thermal_get_temperature(hwmgr); *size = 4; break; + case AMDGPU_PP_SENSOR_HOTSPOT_TEMP: + ret = vega12_get_metrics_table(hwmgr, &metrics_table); + if (ret) + return ret; + + *((uint32_t *)value) = metrics_table.TemperatureHotspot * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + *size = 4; + break; + case AMDGPU_PP_SENSOR_MEM_TEMP: + ret = vega12_get_metrics_table(hwmgr, &metrics_table); + if (ret) + return ret; + + *((uint32_t *)value) = metrics_table.TemperatureHBM * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + *size = 4; + break; case AMDGPU_PP_SENSOR_UVD_POWER: *((uint32_t *)value) = data->uvd_power_gated ? 0 : 1; *size = 4; @@ -1348,6 +1399,8 @@ static int vega12_read_sensor(struct pp_hwmgr *hwmgr, int idx, break; case AMDGPU_PP_SENSOR_GPU_POWER: ret = vega12_get_gpu_power(hwmgr, (uint32_t *)value); + if (!ret) + *size = 4; break; case AMDGPU_PP_SENSOR_ENABLED_SMC_FEATURES_MASK: ret = vega12_get_enabled_smc_features(hwmgr, (uint64_t *)value); @@ -2525,12 +2578,23 @@ static int vega12_notify_cac_buffer_info(struct pp_hwmgr *hwmgr, static int vega12_get_thermal_temperature_range(struct pp_hwmgr *hwmgr, struct PP_TemperatureRange *thermal_data) { - struct phm_ppt_v3_information *pptable_information = - (struct phm_ppt_v3_information *)hwmgr->pptable; + struct vega12_hwmgr *data = + (struct vega12_hwmgr *)(hwmgr->backend); + PPTable_t *pp_table = &(data->smc_state_table.pp_table); memcpy(thermal_data, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange)); - thermal_data->max = pptable_information->us_software_shutdown_temp * + thermal_data->max = pp_table->TedgeLimit * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->edge_emergency_max = (pp_table->TedgeLimit + CTF_OFFSET_EDGE) * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->hotspot_crit_max = pp_table->ThotspotLimit * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->hotspot_emergency_max = (pp_table->ThotspotLimit + CTF_OFFSET_HOTSPOT) * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->mem_crit_max = pp_table->ThbmLimit * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->mem_emergency_max = (pp_table->ThbmLimit + CTF_OFFSET_HBM)* PP_TEMPERATURE_UNITS_PER_CENTIGRADES; return 0; @@ -2575,6 +2639,30 @@ static int vega12_get_performance_level(struct pp_hwmgr *hwmgr, const struct pp_ return 0; } +static int vega12_set_mp1_state(struct pp_hwmgr *hwmgr, + enum pp_mp1_state mp1_state) +{ + uint16_t msg; + int ret; + + switch (mp1_state) { + case PP_MP1_STATE_UNLOAD: + msg = PPSMC_MSG_PrepareMp1ForUnload; + break; + case PP_MP1_STATE_SHUTDOWN: + case PP_MP1_STATE_RESET: + case PP_MP1_STATE_NONE: + default: + return 0; + } + + PP_ASSERT_WITH_CODE((ret = smum_send_msg_to_smc(hwmgr, msg)) == 0, + "[PrepareMp1] Failed!", + return ret); + + return 0; +} + static const struct pp_hwmgr_func vega12_hwmgr_funcs = { .backend_init = vega12_hwmgr_backend_init, .backend_fini = vega12_hwmgr_backend_fini, @@ -2626,8 +2714,12 @@ static const struct pp_hwmgr_func vega12_hwmgr_funcs = { .start_thermal_controller = vega12_start_thermal_controller, .powergate_gfx = vega12_gfx_off_control, .get_performance_level = vega12_get_performance_level, + .get_asic_baco_capability = smu9_baco_get_capability, + .get_asic_baco_state = smu9_baco_get_state, + .set_asic_baco_state = vega12_baco_set_state, .get_ppfeature_status = vega12_get_ppfeature_status, .set_ppfeature_status = vega12_set_ppfeature_status, + .set_mp1_state = vega12_set_mp1_state, }; int vega12_hwmgr_init(struct pp_hwmgr *hwmgr) diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.h b/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.h index b3e424d28994..73875399666a 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.h +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.h @@ -396,6 +396,9 @@ struct vega12_hwmgr { /* ---- Gfxoff ---- */ bool gfxoff_controlled_by_driver; + + unsigned long metrics_time; + SmuMetrics_t metrics_table; }; #define VEGA12_DPM2_NEAR_TDP_DEC 10 diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_inc.h b/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_inc.h index 30b278c50222..e6d9e84059e1 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_inc.h +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_inc.h @@ -35,5 +35,7 @@ #include "asic_reg/gc/gc_9_2_1_sh_mask.h" #include "asic_reg/nbio/nbio_6_1_offset.h" +#include "asic_reg/nbio/nbio_6_1_offset.h" +#include "asic_reg/nbio/nbio_6_1_sh_mask.h" #endif diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_baco.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_baco.c index 5e8602a79b1c..9b5e72bdceca 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_baco.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_baco.c @@ -27,8 +27,9 @@ #include "vega20_inc.h" #include "vega20_ppsmc.h" #include "vega20_baco.h" +#include "vega20_smumgr.h" - +#include "amdgpu_ras.h" static const struct soc15_baco_cmd_entry clean_baco_tbl[] = { @@ -73,6 +74,7 @@ int vega20_baco_get_state(struct pp_hwmgr *hwmgr, enum BACO_STATE *state) int vega20_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state) { struct amdgpu_device *adev = (struct amdgpu_device *)(hwmgr->adev); + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); enum BACO_STATE cur_state; uint32_t data; @@ -83,13 +85,19 @@ int vega20_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state) return 0; if (state == BACO_STATE_IN) { - data = RREG32_SOC15(THM, 0, mmTHM_BACO_CNTL); - data |= 0x80000000; - WREG32_SOC15(THM, 0, mmTHM_BACO_CNTL, data); - - - if(smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_EnterBaco, 0)) - return -EINVAL; + if (!ras || !ras->supported) { + data = RREG32_SOC15(THM, 0, mmTHM_BACO_CNTL); + data |= 0x80000000; + WREG32_SOC15(THM, 0, mmTHM_BACO_CNTL, data); + + if(smum_send_msg_to_smc_with_parameter(hwmgr, + PPSMC_MSG_EnterBaco, 0)) + return -EINVAL; + } else { + if(smum_send_msg_to_smc_with_parameter(hwmgr, + PPSMC_MSG_EnterBaco, 1)) + return -EINVAL; + } } else if (state == BACO_STATE_OUT) { if (smum_send_msg_to_smc(hwmgr, PPSMC_MSG_ExitBaco)) @@ -101,3 +109,14 @@ int vega20_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state) return 0; } + +int vega20_baco_apply_vdci_flush_workaround(struct pp_hwmgr *hwmgr) +{ + int ret = 0; + + ret = vega20_set_pptable_driver_address(hwmgr); + if (ret) + return ret; + + return smum_send_msg_to_smc(hwmgr, PPSMC_MSG_BacoWorkAroundFlushVDCI); +} diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_baco.h b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_baco.h index 51c7f8392925..f06471e712dc 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_baco.h +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_baco.h @@ -28,5 +28,6 @@ extern int vega20_baco_get_capability(struct pp_hwmgr *hwmgr, bool *cap); extern int vega20_baco_get_state(struct pp_hwmgr *hwmgr, enum BACO_STATE *state); extern int vega20_baco_set_state(struct pp_hwmgr *hwmgr, enum BACO_STATE state); +extern int vega20_baco_apply_vdci_flush_workaround(struct pp_hwmgr *hwmgr); #endif diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c index 23b5b94a4939..5bcf0d684151 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c @@ -97,6 +97,27 @@ static void vega20_set_default_registry_data(struct pp_hwmgr *hwmgr) if (hwmgr->smu_version < 0x282100) data->registry_data.disallowed_features |= FEATURE_ECC_MASK; + if (!(hwmgr->feature_mask & PP_PCIE_DPM_MASK)) + data->registry_data.disallowed_features |= FEATURE_DPM_LINK_MASK; + + if (!(hwmgr->feature_mask & PP_SCLK_DPM_MASK)) + data->registry_data.disallowed_features |= FEATURE_DPM_GFXCLK_MASK; + + if (!(hwmgr->feature_mask & PP_SOCCLK_DPM_MASK)) + data->registry_data.disallowed_features |= FEATURE_DPM_SOCCLK_MASK; + + if (!(hwmgr->feature_mask & PP_MCLK_DPM_MASK)) + data->registry_data.disallowed_features |= FEATURE_DPM_UCLK_MASK; + + if (!(hwmgr->feature_mask & PP_DCEFCLK_DPM_MASK)) + data->registry_data.disallowed_features |= FEATURE_DPM_DCEFCLK_MASK; + + if (!(hwmgr->feature_mask & PP_ULV_MASK)) + data->registry_data.disallowed_features |= FEATURE_ULV_MASK; + + if (!(hwmgr->feature_mask & PP_SCLK_DEEP_SLEEP_MASK)) + data->registry_data.disallowed_features |= FEATURE_DS_GFXCLK_MASK; + data->registry_data.od_state_in_dc_support = 0; data->registry_data.thermal_support = 1; data->registry_data.skip_baco_hardware = 0; @@ -162,6 +183,9 @@ static int vega20_set_features_platform_caps(struct pp_hwmgr *hwmgr) PHM_PlatformCaps_TablelessHardwareInterface); phm_cap_set(hwmgr->platform_descriptor.platformCaps, + PHM_PlatformCaps_BACO); + + phm_cap_set(hwmgr->platform_descriptor.platformCaps, PHM_PlatformCaps_EnableSMU7ThermalManagement); if (adev->pg_flags & AMD_PG_SUPPORT_UVD) @@ -303,6 +327,8 @@ static int vega20_set_features_platform_caps(struct pp_hwmgr *hwmgr) static void vega20_init_dpm_defaults(struct pp_hwmgr *hwmgr) { struct vega20_hwmgr *data = (struct vega20_hwmgr *)(hwmgr->backend); + struct amdgpu_device *adev = hwmgr->adev; + uint32_t top32, bottom32; int i; data->smu_features[GNLD_DPM_PREFETCHER].smu_feature_id = @@ -372,6 +398,14 @@ static void vega20_init_dpm_defaults(struct pp_hwmgr *hwmgr) ((data->registry_data.disallowed_features >> i) & 1) ? false : true; } + + /* Get the SN to turn into a Unique ID */ + smum_send_msg_to_smc(hwmgr, PPSMC_MSG_ReadSerialNumTop32); + top32 = smum_get_argument(hwmgr); + smum_send_msg_to_smc(hwmgr, PPSMC_MSG_ReadSerialNumBottom32); + bottom32 = smum_get_argument(hwmgr); + + adev->unique_id = ((uint64_t)bottom32 << 32) | top32; } static int vega20_set_private_data_based_on_pptable(struct pp_hwmgr *hwmgr) @@ -434,6 +468,7 @@ static int vega20_hwmgr_backend_init(struct pp_hwmgr *hwmgr) hwmgr->platform_descriptor.clockStep.memoryClock = 500; data->total_active_cus = adev->gfx.cu_info.number; + data->is_custom_profile_set = false; return 0; } @@ -450,6 +485,7 @@ static int vega20_init_sclk_threshold(struct pp_hwmgr *hwmgr) static int vega20_setup_asic_task(struct pp_hwmgr *hwmgr) { + struct amdgpu_device *adev = (struct amdgpu_device *)(hwmgr->adev); int ret = 0; ret = vega20_init_sclk_threshold(hwmgr); @@ -457,7 +493,15 @@ static int vega20_setup_asic_task(struct pp_hwmgr *hwmgr) "Failed to init sclk threshold!", return ret); - return 0; + if (adev->in_gpu_reset && + (amdgpu_asic_reset_method(adev) == AMD_RESET_METHOD_BACO)) { + + ret = vega20_baco_apply_vdci_flush_workaround(hwmgr); + if (ret) + pr_err("Failed to apply vega20 baco workaround!\n"); + } + + return ret; } /* @@ -2060,7 +2104,11 @@ static int vega20_get_gpu_power(struct pp_hwmgr *hwmgr, if (ret) return ret; - *query = metrics_table.CurrSocketPower << 8; + /* For the 40.46 release, they changed the value name */ + if (hwmgr->smu_version == 0x282e00) + *query = metrics_table.AverageSocketPower << 8; + else + *query = metrics_table.CurrSocketPower << 8; return ret; } @@ -2084,6 +2132,7 @@ static int vega20_get_current_clk_freq(struct pp_hwmgr *hwmgr, } static int vega20_get_current_activity_percent(struct pp_hwmgr *hwmgr, + int idx, uint32_t *activity_percent) { int ret = 0; @@ -2093,7 +2142,17 @@ static int vega20_get_current_activity_percent(struct pp_hwmgr *hwmgr, if (ret) return ret; - *activity_percent = metrics_table.AverageGfxActivity; + switch (idx) { + case AMDGPU_PP_SENSOR_GPU_LOAD: + *activity_percent = metrics_table.AverageGfxActivity; + break; + case AMDGPU_PP_SENSOR_MEM_LOAD: + *activity_percent = metrics_table.AverageUclkActivity; + break; + default: + pr_err("Invalid index for retrieving clock activity\n"); + return -EINVAL; + } return ret; } @@ -2124,14 +2183,33 @@ static int vega20_read_sensor(struct pp_hwmgr *hwmgr, int idx, *size = 4; break; case AMDGPU_PP_SENSOR_GPU_LOAD: - ret = vega20_get_current_activity_percent(hwmgr, (uint32_t *)value); + case AMDGPU_PP_SENSOR_MEM_LOAD: + ret = vega20_get_current_activity_percent(hwmgr, idx, (uint32_t *)value); if (!ret) *size = 4; break; - case AMDGPU_PP_SENSOR_GPU_TEMP: + case AMDGPU_PP_SENSOR_HOTSPOT_TEMP: *((uint32_t *)value) = vega20_thermal_get_temperature(hwmgr); *size = 4; break; + case AMDGPU_PP_SENSOR_EDGE_TEMP: + ret = vega20_get_metrics_table(hwmgr, &metrics_table); + if (ret) + return ret; + + *((uint32_t *)value) = metrics_table.TemperatureEdge * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + *size = 4; + break; + case AMDGPU_PP_SENSOR_MEM_TEMP: + ret = vega20_get_metrics_table(hwmgr, &metrics_table); + if (ret) + return ret; + + *((uint32_t *)value) = metrics_table.TemperatureHBM * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + *size = 4; + break; case AMDGPU_PP_SENSOR_UVD_POWER: *((uint32_t *)value) = data->uvd_power_gated ? 0 : 1; *size = 4; @@ -2278,12 +2356,16 @@ static int vega20_force_dpm_highest(struct pp_hwmgr *hwmgr) data->dpm_table.soc_table.dpm_state.soft_max_level = data->dpm_table.soc_table.dpm_levels[soft_level].value; - ret = vega20_upload_dpm_min_level(hwmgr, 0xFFFFFFFF); + ret = vega20_upload_dpm_min_level(hwmgr, FEATURE_DPM_GFXCLK_MASK | + FEATURE_DPM_UCLK_MASK | + FEATURE_DPM_SOCCLK_MASK); PP_ASSERT_WITH_CODE(!ret, "Failed to upload boot level to highest!", return ret); - ret = vega20_upload_dpm_max_level(hwmgr, 0xFFFFFFFF); + ret = vega20_upload_dpm_max_level(hwmgr, FEATURE_DPM_GFXCLK_MASK | + FEATURE_DPM_UCLK_MASK | + FEATURE_DPM_SOCCLK_MASK); PP_ASSERT_WITH_CODE(!ret, "Failed to upload dpm max level to highest!", return ret); @@ -2316,12 +2398,16 @@ static int vega20_force_dpm_lowest(struct pp_hwmgr *hwmgr) data->dpm_table.soc_table.dpm_state.soft_max_level = data->dpm_table.soc_table.dpm_levels[soft_level].value; - ret = vega20_upload_dpm_min_level(hwmgr, 0xFFFFFFFF); + ret = vega20_upload_dpm_min_level(hwmgr, FEATURE_DPM_GFXCLK_MASK | + FEATURE_DPM_UCLK_MASK | + FEATURE_DPM_SOCCLK_MASK); PP_ASSERT_WITH_CODE(!ret, "Failed to upload boot level to highest!", return ret); - ret = vega20_upload_dpm_max_level(hwmgr, 0xFFFFFFFF); + ret = vega20_upload_dpm_max_level(hwmgr, FEATURE_DPM_GFXCLK_MASK | + FEATURE_DPM_UCLK_MASK | + FEATURE_DPM_SOCCLK_MASK); PP_ASSERT_WITH_CODE(!ret, "Failed to upload dpm max level to highest!", return ret); @@ -2332,14 +2418,54 @@ static int vega20_force_dpm_lowest(struct pp_hwmgr *hwmgr) static int vega20_unforce_dpm_levels(struct pp_hwmgr *hwmgr) { + struct vega20_hwmgr *data = + (struct vega20_hwmgr *)(hwmgr->backend); + uint32_t soft_min_level, soft_max_level; int ret = 0; - ret = vega20_upload_dpm_min_level(hwmgr, 0xFFFFFFFF); + /* gfxclk soft min/max settings */ + soft_min_level = + vega20_find_lowest_dpm_level(&(data->dpm_table.gfx_table)); + soft_max_level = + vega20_find_highest_dpm_level(&(data->dpm_table.gfx_table)); + + data->dpm_table.gfx_table.dpm_state.soft_min_level = + data->dpm_table.gfx_table.dpm_levels[soft_min_level].value; + data->dpm_table.gfx_table.dpm_state.soft_max_level = + data->dpm_table.gfx_table.dpm_levels[soft_max_level].value; + + /* uclk soft min/max settings */ + soft_min_level = + vega20_find_lowest_dpm_level(&(data->dpm_table.mem_table)); + soft_max_level = + vega20_find_highest_dpm_level(&(data->dpm_table.mem_table)); + + data->dpm_table.mem_table.dpm_state.soft_min_level = + data->dpm_table.mem_table.dpm_levels[soft_min_level].value; + data->dpm_table.mem_table.dpm_state.soft_max_level = + data->dpm_table.mem_table.dpm_levels[soft_max_level].value; + + /* socclk soft min/max settings */ + soft_min_level = + vega20_find_lowest_dpm_level(&(data->dpm_table.soc_table)); + soft_max_level = + vega20_find_highest_dpm_level(&(data->dpm_table.soc_table)); + + data->dpm_table.soc_table.dpm_state.soft_min_level = + data->dpm_table.soc_table.dpm_levels[soft_min_level].value; + data->dpm_table.soc_table.dpm_state.soft_max_level = + data->dpm_table.soc_table.dpm_levels[soft_max_level].value; + + ret = vega20_upload_dpm_min_level(hwmgr, FEATURE_DPM_GFXCLK_MASK | + FEATURE_DPM_UCLK_MASK | + FEATURE_DPM_SOCCLK_MASK); PP_ASSERT_WITH_CODE(!ret, "Failed to upload DPM Bootup Levels!", return ret); - ret = vega20_upload_dpm_max_level(hwmgr, 0xFFFFFFFF); + ret = vega20_upload_dpm_max_level(hwmgr, FEATURE_DPM_GFXCLK_MASK | + FEATURE_DPM_UCLK_MASK | + FEATURE_DPM_SOCCLK_MASK); PP_ASSERT_WITH_CODE(!ret, "Failed to upload DPM Max Levels!", return ret); @@ -2992,6 +3118,34 @@ static int vega20_odn_edit_dpm_table(struct pp_hwmgr *hwmgr, return 0; } +static int vega20_set_mp1_state(struct pp_hwmgr *hwmgr, + enum pp_mp1_state mp1_state) +{ + uint16_t msg; + int ret; + + switch (mp1_state) { + case PP_MP1_STATE_SHUTDOWN: + msg = PPSMC_MSG_PrepareMp1ForShutdown; + break; + case PP_MP1_STATE_UNLOAD: + msg = PPSMC_MSG_PrepareMp1ForUnload; + break; + case PP_MP1_STATE_RESET: + msg = PPSMC_MSG_PrepareMp1ForReset; + break; + case PP_MP1_STATE_NONE: + default: + return 0; + } + + PP_ASSERT_WITH_CODE((ret = smum_send_msg_to_smc(hwmgr, msg)) == 0, + "[PrepareMp1] Failed!", + return ret); + + return 0; +} + static int vega20_get_ppfeature_status(struct pp_hwmgr *hwmgr, char *buf) { static const char *ppfeature_name[] = { @@ -3450,7 +3604,18 @@ static void vega20_power_gate_vce(struct pp_hwmgr *hwmgr, bool bgate) return ; data->vce_power_gated = bgate; - vega20_enable_disable_vce_dpm(hwmgr, !bgate); + if (bgate) { + vega20_enable_disable_vce_dpm(hwmgr, !bgate); + amdgpu_device_ip_set_powergating_state(hwmgr->adev, + AMD_IP_BLOCK_TYPE_VCE, + AMD_PG_STATE_GATE); + } else { + amdgpu_device_ip_set_powergating_state(hwmgr->adev, + AMD_IP_BLOCK_TYPE_VCE, + AMD_PG_STATE_UNGATE); + vega20_enable_disable_vce_dpm(hwmgr, !bgate); + } + } static void vega20_power_gate_uvd(struct pp_hwmgr *hwmgr, bool bgate) @@ -3826,16 +3991,19 @@ static int vega20_set_power_profile_mode(struct pp_hwmgr *hwmgr, long *input, ui { DpmActivityMonitorCoeffInt_t activity_monitor; int workload_type, result = 0; + uint32_t power_profile_mode = input[size]; - hwmgr->power_profile_mode = input[size]; - - if (hwmgr->power_profile_mode > PP_SMC_POWER_PROFILE_CUSTOM) { - pr_err("Invalid power profile mode %d\n", hwmgr->power_profile_mode); + if (power_profile_mode > PP_SMC_POWER_PROFILE_CUSTOM) { + pr_err("Invalid power profile mode %d\n", power_profile_mode); return -EINVAL; } - if (hwmgr->power_profile_mode == PP_SMC_POWER_PROFILE_CUSTOM) { - if (size < 10) + if (power_profile_mode == PP_SMC_POWER_PROFILE_CUSTOM) { + struct vega20_hwmgr *data = + (struct vega20_hwmgr *)(hwmgr->backend); + if (size == 0 && !data->is_custom_profile_set) + return -EINVAL; + if (size < 10 && size != 0) return -EINVAL; result = vega20_get_activity_monitor_coeff(hwmgr, @@ -3845,6 +4013,13 @@ static int vega20_set_power_profile_mode(struct pp_hwmgr *hwmgr, long *input, ui "[SetPowerProfile] Failed to get activity monitor!", return result); + /* If size==0, then we want to apply the already-configured + * CUSTOM profile again. Just apply it, since we checked its + * validity above + */ + if (size == 0) + goto out; + switch (input[0]) { case 0: /* Gfxclk */ activity_monitor.Gfx_FPS = input[1]; @@ -3895,17 +4070,21 @@ static int vega20_set_power_profile_mode(struct pp_hwmgr *hwmgr, long *input, ui result = vega20_set_activity_monitor_coeff(hwmgr, (uint8_t *)(&activity_monitor), WORKLOAD_PPLIB_CUSTOM_BIT); + data->is_custom_profile_set = true; PP_ASSERT_WITH_CODE(!result, "[SetPowerProfile] Failed to set activity monitor!", return result); } +out: /* conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT */ workload_type = - conv_power_profile_to_pplib_workload(hwmgr->power_profile_mode); + conv_power_profile_to_pplib_workload(power_profile_mode); smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetWorkloadMask, 1 << workload_type); + hwmgr->power_profile_mode = power_profile_mode; + return 0; } @@ -3939,17 +4118,78 @@ static int vega20_notify_cac_buffer_info(struct pp_hwmgr *hwmgr, static int vega20_get_thermal_temperature_range(struct pp_hwmgr *hwmgr, struct PP_TemperatureRange *thermal_data) { - struct phm_ppt_v3_information *pptable_information = - (struct phm_ppt_v3_information *)hwmgr->pptable; + struct vega20_hwmgr *data = + (struct vega20_hwmgr *)(hwmgr->backend); + PPTable_t *pp_table = &(data->smc_state_table.pp_table); memcpy(thermal_data, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange)); - thermal_data->max = pptable_information->us_software_shutdown_temp * + thermal_data->max = pp_table->TedgeLimit * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->edge_emergency_max = (pp_table->TedgeLimit + CTF_OFFSET_EDGE) * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->hotspot_crit_max = pp_table->ThotspotLimit * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->hotspot_emergency_max = (pp_table->ThotspotLimit + CTF_OFFSET_HOTSPOT) * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->mem_crit_max = pp_table->ThbmLimit * + PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->mem_emergency_max = (pp_table->ThbmLimit + CTF_OFFSET_HBM)* PP_TEMPERATURE_UNITS_PER_CENTIGRADES; return 0; } +static int vega20_smu_i2c_bus_access(struct pp_hwmgr *hwmgr, bool acquire) +{ + int res; + + /* I2C bus access can happen very early, when SMU not loaded yet */ + if (!vega20_is_smc_ram_running(hwmgr)) + return 0; + + res = smum_send_msg_to_smc_with_parameter(hwmgr, + (acquire ? + PPSMC_MSG_RequestI2CBus : + PPSMC_MSG_ReleaseI2CBus), + 0); + + PP_ASSERT_WITH_CODE(!res, "[SmuI2CAccessBus] Failed to access bus!", return res); + return res; +} + +static int vega20_set_df_cstate(struct pp_hwmgr *hwmgr, + enum pp_df_cstate state) +{ + int ret; + + /* PPSMC_MSG_DFCstateControl is supported with 40.50 and later fws */ + if (hwmgr->smu_version < 0x283200) { + pr_err("Df cstate control is supported with 40.50 and later SMC fw!\n"); + return -EINVAL; + } + + ret = smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_DFCstateControl, state); + if (ret) + pr_err("SetDfCstate failed!\n"); + + return ret; +} + +static int vega20_set_xgmi_pstate(struct pp_hwmgr *hwmgr, + uint32_t pstate) +{ + int ret; + + ret = smum_send_msg_to_smc_with_parameter(hwmgr, + PPSMC_MSG_SetXgmiMode, + pstate ? XGMI_MODE_PSTATE_D0 : XGMI_MODE_PSTATE_D3); + if (ret) + pr_err("SetXgmiPstate failed!\n"); + + return ret; +} + static const struct pp_hwmgr_func vega20_hwmgr_funcs = { /* init/fini related */ .backend_init = vega20_hwmgr_backend_init, @@ -4016,6 +4256,10 @@ static const struct pp_hwmgr_func vega20_hwmgr_funcs = { .get_asic_baco_capability = vega20_baco_get_capability, .get_asic_baco_state = vega20_baco_get_state, .set_asic_baco_state = vega20_baco_set_state, + .set_mp1_state = vega20_set_mp1_state, + .smu_i2c_bus_access = vega20_smu_i2c_bus_access, + .set_df_cstate = vega20_set_df_cstate, + .set_xgmi_pstate = vega20_set_xgmi_pstate, }; int vega20_hwmgr_init(struct pp_hwmgr *hwmgr) diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.h b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.h index ac2a3118a0ae..2c3125f82b24 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.h +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.h @@ -531,6 +531,8 @@ struct vega20_hwmgr { bool pcie_parameters_override; uint32_t pcie_gen_level1; uint32_t pcie_width_level1; + + bool is_custom_profile_set; }; #define VEGA20_DPM2_NEAR_TDP_DEC 10 diff --git a/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h b/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h new file mode 100644 index 000000000000..031e0c22fcc7 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h @@ -0,0 +1,723 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#ifndef __AMDGPU_SMU_H__ +#define __AMDGPU_SMU_H__ + +#include "amdgpu.h" +#include "kgd_pp_interface.h" +#include "dm_pp_interface.h" +#include "dm_pp_smu.h" +#include "smu_types.h" + +#define SMU_THERMAL_MINIMUM_ALERT_TEMP 0 +#define SMU_THERMAL_MAXIMUM_ALERT_TEMP 255 +#define SMU_TEMPERATURE_UNITS_PER_CENTIGRADES 1000 + +struct smu_hw_power_state { + unsigned int magic; +}; + +struct smu_power_state; + +enum smu_state_ui_label { + SMU_STATE_UI_LABEL_NONE, + SMU_STATE_UI_LABEL_BATTERY, + SMU_STATE_UI_TABEL_MIDDLE_LOW, + SMU_STATE_UI_LABEL_BALLANCED, + SMU_STATE_UI_LABEL_MIDDLE_HIGHT, + SMU_STATE_UI_LABEL_PERFORMANCE, + SMU_STATE_UI_LABEL_BACO, +}; + +enum smu_state_classification_flag { + SMU_STATE_CLASSIFICATION_FLAG_BOOT = 0x0001, + SMU_STATE_CLASSIFICATION_FLAG_THERMAL = 0x0002, + SMU_STATE_CLASSIFICATIN_FLAG_LIMITED_POWER_SOURCE = 0x0004, + SMU_STATE_CLASSIFICATION_FLAG_RESET = 0x0008, + SMU_STATE_CLASSIFICATION_FLAG_FORCED = 0x0010, + SMU_STATE_CLASSIFICATION_FLAG_USER_3D_PERFORMANCE = 0x0020, + SMU_STATE_CLASSIFICATION_FLAG_USER_2D_PERFORMANCE = 0x0040, + SMU_STATE_CLASSIFICATION_FLAG_3D_PERFORMANCE = 0x0080, + SMU_STATE_CLASSIFICATION_FLAG_AC_OVERDIRVER_TEMPLATE = 0x0100, + SMU_STATE_CLASSIFICATION_FLAG_UVD = 0x0200, + SMU_STATE_CLASSIFICATION_FLAG_3D_PERFORMANCE_LOW = 0x0400, + SMU_STATE_CLASSIFICATION_FLAG_ACPI = 0x0800, + SMU_STATE_CLASSIFICATION_FLAG_HD2 = 0x1000, + SMU_STATE_CLASSIFICATION_FLAG_UVD_HD = 0x2000, + SMU_STATE_CLASSIFICATION_FLAG_UVD_SD = 0x4000, + SMU_STATE_CLASSIFICATION_FLAG_USER_DC_PERFORMANCE = 0x8000, + SMU_STATE_CLASSIFICATION_FLAG_DC_OVERDIRVER_TEMPLATE = 0x10000, + SMU_STATE_CLASSIFICATION_FLAG_BACO = 0x20000, + SMU_STATE_CLASSIFICATIN_FLAG_LIMITED_POWER_SOURCE2 = 0x40000, + SMU_STATE_CLASSIFICATION_FLAG_ULV = 0x80000, + SMU_STATE_CLASSIFICATION_FLAG_UVD_MVC = 0x100000, +}; + +struct smu_state_classification_block { + enum smu_state_ui_label ui_label; + enum smu_state_classification_flag flags; + int bios_index; + bool temporary_state; + bool to_be_deleted; +}; + +struct smu_state_pcie_block { + unsigned int lanes; +}; + +enum smu_refreshrate_source { + SMU_REFRESHRATE_SOURCE_EDID, + SMU_REFRESHRATE_SOURCE_EXPLICIT +}; + +struct smu_state_display_block { + bool disable_frame_modulation; + bool limit_refreshrate; + enum smu_refreshrate_source refreshrate_source; + int explicit_refreshrate; + int edid_refreshrate_index; + bool enable_vari_bright; +}; + +struct smu_state_memroy_block { + bool dll_off; + uint8_t m3arb; + uint8_t unused[3]; +}; + +struct smu_state_software_algorithm_block { + bool disable_load_balancing; + bool enable_sleep_for_timestamps; +}; + +struct smu_temperature_range { + int min; + int max; + int edge_emergency_max; + int hotspot_min; + int hotspot_crit_max; + int hotspot_emergency_max; + int mem_min; + int mem_crit_max; + int mem_emergency_max; +}; + +struct smu_state_validation_block { + bool single_display_only; + bool disallow_on_dc; + uint8_t supported_power_levels; +}; + +struct smu_uvd_clocks { + uint32_t vclk; + uint32_t dclk; +}; + +/** +* Structure to hold a SMU Power State. +*/ +struct smu_power_state { + uint32_t id; + struct list_head ordered_list; + struct list_head all_states_list; + + struct smu_state_classification_block classification; + struct smu_state_validation_block validation; + struct smu_state_pcie_block pcie; + struct smu_state_display_block display; + struct smu_state_memroy_block memory; + struct smu_temperature_range temperatures; + struct smu_state_software_algorithm_block software; + struct smu_uvd_clocks uvd_clocks; + struct smu_hw_power_state hardware; +}; + +enum smu_power_src_type +{ + SMU_POWER_SOURCE_AC, + SMU_POWER_SOURCE_DC, + SMU_POWER_SOURCE_COUNT, +}; + +enum smu_memory_pool_size +{ + SMU_MEMORY_POOL_SIZE_ZERO = 0, + SMU_MEMORY_POOL_SIZE_256_MB = 0x10000000, + SMU_MEMORY_POOL_SIZE_512_MB = 0x20000000, + SMU_MEMORY_POOL_SIZE_1_GB = 0x40000000, + SMU_MEMORY_POOL_SIZE_2_GB = 0x80000000, +}; + +#define SMU_TABLE_INIT(tables, table_id, s, a, d) \ + do { \ + tables[table_id].size = s; \ + tables[table_id].align = a; \ + tables[table_id].domain = d; \ + } while (0) + +struct smu_table { + uint64_t size; + uint32_t align; + uint8_t domain; + uint64_t mc_address; + void *cpu_addr; + struct amdgpu_bo *bo; +}; + +enum smu_perf_level_designation { + PERF_LEVEL_ACTIVITY, + PERF_LEVEL_POWER_CONTAINMENT, +}; + +struct smu_performance_level { + uint32_t core_clock; + uint32_t memory_clock; + uint32_t vddc; + uint32_t vddci; + uint32_t non_local_mem_freq; + uint32_t non_local_mem_width; +}; + +struct smu_clock_info { + uint32_t min_mem_clk; + uint32_t max_mem_clk; + uint32_t min_eng_clk; + uint32_t max_eng_clk; + uint32_t min_bus_bandwidth; + uint32_t max_bus_bandwidth; +}; + +struct smu_bios_boot_up_values +{ + uint32_t revision; + uint32_t gfxclk; + uint32_t uclk; + uint32_t socclk; + uint32_t dcefclk; + uint32_t eclk; + uint32_t vclk; + uint32_t dclk; + uint16_t vddc; + uint16_t vddci; + uint16_t mvddc; + uint16_t vdd_gfx; + uint8_t cooling_id; + uint32_t pp_table_id; + uint32_t format_revision; + uint32_t content_revision; + uint32_t fclk; +}; + +enum smu_table_id +{ + SMU_TABLE_PPTABLE = 0, + SMU_TABLE_WATERMARKS, + SMU_TABLE_CUSTOM_DPM, + SMU_TABLE_DPMCLOCKS, + SMU_TABLE_AVFS, + SMU_TABLE_AVFS_PSM_DEBUG, + SMU_TABLE_AVFS_FUSE_OVERRIDE, + SMU_TABLE_PMSTATUSLOG, + SMU_TABLE_SMU_METRICS, + SMU_TABLE_DRIVER_SMU_CONFIG, + SMU_TABLE_ACTIVITY_MONITOR_COEFF, + SMU_TABLE_OVERDRIVE, + SMU_TABLE_I2C_COMMANDS, + SMU_TABLE_PACE, + SMU_TABLE_COUNT, +}; + +struct smu_table_context +{ + void *power_play_table; + uint32_t power_play_table_size; + void *hardcode_pptable; + unsigned long metrics_time; + void *metrics_table; + void *clocks_table; + + void *max_sustainable_clocks; + struct smu_bios_boot_up_values boot_values; + void *driver_pptable; + struct smu_table *tables; + struct smu_table memory_pool; + uint8_t thermal_controller_type; + + void *overdrive_table; +}; + +struct smu_dpm_context { + uint32_t dpm_context_size; + void *dpm_context; + void *golden_dpm_context; + bool enable_umd_pstate; + enum amd_dpm_forced_level dpm_level; + enum amd_dpm_forced_level saved_dpm_level; + enum amd_dpm_forced_level requested_dpm_level; + struct smu_power_state *dpm_request_power_state; + struct smu_power_state *dpm_current_power_state; + struct mclock_latency_table *mclk_latency_table; +}; + +struct smu_power_gate { + bool uvd_gated; + bool vce_gated; + bool vcn_gated; +}; + +struct smu_power_context { + void *power_context; + uint32_t power_context_size; + struct smu_power_gate power_gate; +}; + + +#define SMU_FEATURE_MAX (64) +struct smu_feature +{ + uint32_t feature_num; + DECLARE_BITMAP(supported, SMU_FEATURE_MAX); + DECLARE_BITMAP(allowed, SMU_FEATURE_MAX); + DECLARE_BITMAP(enabled, SMU_FEATURE_MAX); + struct mutex mutex; +}; + +struct smu_clocks { + uint32_t engine_clock; + uint32_t memory_clock; + uint32_t bus_bandwidth; + uint32_t engine_clock_in_sr; + uint32_t dcef_clock; + uint32_t dcef_clock_in_sr; +}; + +#define MAX_REGULAR_DPM_NUM 16 +struct mclk_latency_entries { + uint32_t frequency; + uint32_t latency; +}; +struct mclock_latency_table { + uint32_t count; + struct mclk_latency_entries entries[MAX_REGULAR_DPM_NUM]; +}; + +enum smu_reset_mode +{ + SMU_RESET_MODE_0, + SMU_RESET_MODE_1, + SMU_RESET_MODE_2, +}; + +enum smu_baco_state +{ + SMU_BACO_STATE_ENTER = 0, + SMU_BACO_STATE_EXIT, +}; + +struct smu_baco_context +{ + struct mutex mutex; + uint32_t state; + bool platform_support; +}; + +#define WORKLOAD_POLICY_MAX 7 +struct smu_context +{ + struct amdgpu_device *adev; + struct amdgpu_irq_src *irq_source; + + const struct pptable_funcs *ppt_funcs; + struct mutex mutex; + struct mutex sensor_lock; + uint64_t pool_size; + + struct smu_table_context smu_table; + struct smu_dpm_context smu_dpm; + struct smu_power_context smu_power; + struct smu_feature smu_feature; + struct amd_pp_display_configuration *display_config; + struct smu_baco_context smu_baco; + void *od_settings; + + uint32_t pstate_sclk; + uint32_t pstate_mclk; + + bool od_enabled; + uint32_t power_limit; + uint32_t default_power_limit; + + /* soft pptable */ + uint32_t ppt_offset_bytes; + uint32_t ppt_size_bytes; + uint8_t *ppt_start_addr; + + bool support_power_containment; + bool disable_watermark; + +#define WATERMARKS_EXIST (1 << 0) +#define WATERMARKS_LOADED (1 << 1) + uint32_t watermarks_bitmap; + uint32_t hard_min_uclk_req_from_dal; + bool disable_uclk_switch; + + uint32_t workload_mask; + uint32_t workload_prority[WORKLOAD_POLICY_MAX]; + uint32_t workload_setting[WORKLOAD_POLICY_MAX]; + uint32_t power_profile_mode; + uint32_t default_power_profile_mode; + bool pm_enabled; + bool is_apu; + + uint32_t smc_if_version; + + bool uploading_custom_pp_table; +}; + +struct i2c_adapter; + +struct pptable_funcs { + int (*alloc_dpm_context)(struct smu_context *smu); + int (*store_powerplay_table)(struct smu_context *smu); + int (*check_powerplay_table)(struct smu_context *smu); + int (*append_powerplay_table)(struct smu_context *smu); + int (*get_smu_msg_index)(struct smu_context *smu, uint32_t index); + int (*get_smu_clk_index)(struct smu_context *smu, uint32_t index); + int (*get_smu_feature_index)(struct smu_context *smu, uint32_t index); + int (*get_smu_table_index)(struct smu_context *smu, uint32_t index); + int (*get_smu_power_index)(struct smu_context *smu, uint32_t index); + int (*get_workload_type)(struct smu_context *smu, enum PP_SMC_POWER_PROFILE profile); + int (*run_btc)(struct smu_context *smu); + int (*get_allowed_feature_mask)(struct smu_context *smu, uint32_t *feature_mask, uint32_t num); + enum amd_pm_state_type (*get_current_power_state)(struct smu_context *smu); + int (*set_default_dpm_table)(struct smu_context *smu); + int (*set_power_state)(struct smu_context *smu); + int (*populate_umd_state_clk)(struct smu_context *smu); + int (*print_clk_levels)(struct smu_context *smu, enum smu_clk_type clk_type, char *buf); + int (*force_clk_levels)(struct smu_context *smu, enum smu_clk_type clk_type, uint32_t mask); + int (*set_default_od8_settings)(struct smu_context *smu); + int (*get_od_percentage)(struct smu_context *smu, enum smu_clk_type clk_type); + int (*set_od_percentage)(struct smu_context *smu, + enum smu_clk_type clk_type, + uint32_t value); + int (*od_edit_dpm_table)(struct smu_context *smu, + enum PP_OD_DPM_TABLE_COMMAND type, + long *input, uint32_t size); + int (*get_clock_by_type_with_latency)(struct smu_context *smu, + enum smu_clk_type clk_type, + struct + pp_clock_levels_with_latency + *clocks); + int (*get_clock_by_type_with_voltage)(struct smu_context *smu, + enum amd_pp_clock_type type, + struct + pp_clock_levels_with_voltage + *clocks); + int (*get_power_profile_mode)(struct smu_context *smu, char *buf); + 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 (*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 (*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, + enum amd_dpm_forced_level level, + uint32_t *sclk_mask, + uint32_t *mclk_mask, + uint32_t *soc_mask); + int (*set_cpu_power_state)(struct smu_context *smu); + bool (*is_dpm_running)(struct smu_context *smu); + int (*tables_init)(struct smu_context *smu, struct smu_table *tables); + int (*set_thermal_fan_table)(struct smu_context *smu); + int (*get_fan_speed_percent)(struct smu_context *smu, uint32_t *speed); + int (*get_fan_speed_rpm)(struct smu_context *smu, uint32_t *speed); + int (*set_watermarks_table)(struct smu_context *smu, void *watermarks, + struct dm_pp_wm_sets_with_clock_ranges_soc15 *clock_ranges); + int (*get_current_clk_freq_by_table)(struct smu_context *smu, + enum smu_clk_type clk_type, + uint32_t *value); + int (*get_thermal_temperature_range)(struct smu_context *smu, struct smu_temperature_range *range); + int (*get_uclk_dpm_states)(struct smu_context *smu, uint32_t *clocks_in_khz, uint32_t *num_states); + int (*set_default_od_settings)(struct smu_context *smu, bool initialize); + int (*set_performance_level)(struct smu_context *smu, enum amd_dpm_forced_level level); + int (*display_disable_memory_clock_switch)(struct smu_context *smu, bool disable_memory_clock_switch); + void (*dump_pptable)(struct smu_context *smu); + int (*get_power_limit)(struct smu_context *smu, uint32_t *limit, bool asic_default); + int (*get_dpm_clk_limited)(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t dpm_level, uint32_t *freq); + int (*set_df_cstate)(struct smu_context *smu, enum pp_df_cstate state); + int (*update_pcie_parameters)(struct smu_context *smu, uint32_t pcie_gen_cap, uint32_t pcie_width_cap); + int (*i2c_eeprom_init)(struct i2c_adapter *control); + void (*i2c_eeprom_fini)(struct i2c_adapter *control); + int (*get_dpm_clock_table)(struct smu_context *smu, struct dpm_clocks *clock_table); + int (*init_microcode)(struct smu_context *smu); + int (*load_microcode)(struct smu_context *smu); + int (*init_smc_tables)(struct smu_context *smu); + int (*fini_smc_tables)(struct smu_context *smu); + int (*init_power)(struct smu_context *smu); + int (*fini_power)(struct smu_context *smu); + int (*check_fw_status)(struct smu_context *smu); + int (*setup_pptable)(struct smu_context *smu); + int (*get_vbios_bootup_values)(struct smu_context *smu); + int (*get_clk_info_from_vbios)(struct smu_context *smu); + int (*check_pptable)(struct smu_context *smu); + int (*parse_pptable)(struct smu_context *smu); + int (*populate_smc_tables)(struct smu_context *smu); + 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 (*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_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); + int (*system_features_control)(struct smu_context *smu, bool en); + int (*send_smc_msg)(struct smu_context *smu, uint16_t msg); + int (*send_smc_msg_with_param)(struct smu_context *smu, uint16_t msg, uint32_t param); + int (*read_smc_arg)(struct smu_context *smu, uint32_t *arg); + int (*init_display_count)(struct smu_context *smu, uint32_t count); + int (*set_allowed_mask)(struct smu_context *smu); + int (*get_enabled_mask)(struct smu_context *smu, uint32_t *feature_mask, uint32_t num); + int (*notify_display_change)(struct smu_context *smu); + int (*set_power_limit)(struct smu_context *smu, uint32_t n); + int (*get_current_clk_freq)(struct smu_context *smu, enum smu_clk_type clk_id, uint32_t *value); + int (*init_max_sustainable_clocks)(struct smu_context *smu); + int (*start_thermal_control)(struct smu_context *smu); + int (*stop_thermal_control)(struct smu_context *smu); + int (*set_deep_sleep_dcefclk)(struct smu_context *smu, uint32_t clk); + int (*set_active_display_count)(struct smu_context *smu, uint32_t count); + int (*store_cc6_data)(struct smu_context *smu, uint32_t separation_time, + bool cc6_disable, bool pstate_disable, + bool pstate_switch_disable); + int (*get_clock_by_type)(struct smu_context *smu, + enum amd_pp_clock_type type, + struct amd_pp_clocks *clocks); + int (*get_max_high_clocks)(struct smu_context *smu, + struct amd_pp_simple_clock_info *clocks); + int (*display_clock_voltage_request)(struct smu_context *smu, struct + pp_display_clock_request + *clock_req); + int (*get_dal_power_level)(struct smu_context *smu, + struct amd_pp_simple_clock_info *clocks); + int (*get_perf_level)(struct smu_context *smu, + enum smu_perf_level_designation designation, + struct smu_performance_level *level); + int (*get_current_shallow_sleep_clocks)(struct smu_context *smu, + struct smu_clock_info *clocks); + int (*notify_smu_enable_pwe)(struct smu_context *smu); + int (*conv_power_profile_to_pplib_workload)(int power_profile); + uint32_t (*get_fan_control_mode)(struct smu_context *smu); + int (*set_fan_control_mode)(struct smu_context *smu, uint32_t mode); + int (*set_fan_speed_percent)(struct smu_context *smu, uint32_t speed); + int (*set_fan_speed_rpm)(struct smu_context *smu, uint32_t speed); + int (*set_xgmi_pstate)(struct smu_context *smu, uint32_t pstate); + int (*gfx_off_control)(struct smu_context *smu, bool enable); + int (*register_irq_handler)(struct smu_context *smu); + int (*set_azalia_d3_pme)(struct smu_context *smu); + int (*get_max_sustainable_clocks_by_dc)(struct smu_context *smu, struct pp_smu_nv_clock_table *max_clocks); + 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 (*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); + int (*override_pcie_parameters)(struct smu_context *smu); + uint32_t (*get_pptable_power_limit)(struct smu_context *smu); +}; + +int smu_load_microcode(struct smu_context *smu); + +int smu_check_fw_status(struct smu_context *smu); + +int smu_set_gfx_cgpg(struct smu_context *smu, bool enabled); + +#define smu_i2c_eeprom_init(smu, control) \ + ((smu)->ppt_funcs->i2c_eeprom_init ? (smu)->ppt_funcs->i2c_eeprom_init((control)) : -EINVAL) +#define smu_i2c_eeprom_fini(smu, control) \ + ((smu)->ppt_funcs->i2c_eeprom_fini ? (smu)->ppt_funcs->i2c_eeprom_fini((control)) : -EINVAL) + +int smu_set_fan_speed_rpm(struct smu_context *smu, uint32_t speed); + +int smu_get_power_limit(struct smu_context *smu, + uint32_t *limit, + bool def, + bool lock_needed); + +int smu_set_power_limit(struct smu_context *smu, uint32_t limit); +int smu_print_clk_levels(struct smu_context *smu, enum smu_clk_type clk_type, char *buf); +int smu_get_od_percentage(struct smu_context *smu, enum smu_clk_type type); +int smu_set_od_percentage(struct smu_context *smu, enum smu_clk_type type, uint32_t value); + +int smu_od_edit_dpm_table(struct smu_context *smu, + enum PP_OD_DPM_TABLE_COMMAND type, + long *input, uint32_t size); + +int smu_read_sensor(struct smu_context *smu, + enum amd_pp_sensors sensor, + void *data, uint32_t *size); +int smu_get_power_profile_mode(struct smu_context *smu, char *buf); + +int smu_set_power_profile_mode(struct smu_context *smu, + long *param, + uint32_t param_size, + bool lock_needed); +int smu_get_fan_control_mode(struct smu_context *smu); +int smu_set_fan_control_mode(struct smu_context *smu, int value); +int smu_get_fan_speed_percent(struct smu_context *smu, uint32_t *speed); +int smu_set_fan_speed_percent(struct smu_context *smu, uint32_t speed); +int smu_get_fan_speed_rpm(struct smu_context *smu, uint32_t *speed); + +int smu_set_deep_sleep_dcefclk(struct smu_context *smu, int clk); +int smu_set_active_display_count(struct smu_context *smu, uint32_t count); + +int smu_get_clock_by_type(struct smu_context *smu, + enum amd_pp_clock_type type, + struct amd_pp_clocks *clocks); + +int smu_get_max_high_clocks(struct smu_context *smu, + struct amd_pp_simple_clock_info *clocks); + +int smu_get_clock_by_type_with_latency(struct smu_context *smu, + enum smu_clk_type clk_type, + struct pp_clock_levels_with_latency *clocks); + +int smu_get_clock_by_type_with_voltage(struct smu_context *smu, + enum amd_pp_clock_type type, + struct pp_clock_levels_with_voltage *clocks); + +int smu_display_clock_voltage_request(struct smu_context *smu, + struct pp_display_clock_request *clock_req); +int smu_display_disable_memory_clock_switch(struct smu_context *smu, bool disable_memory_clock_switch); +int smu_notify_smu_enable_pwe(struct smu_context *smu); + +int smu_set_xgmi_pstate(struct smu_context *smu, + uint32_t pstate); + +int smu_set_azalia_d3_pme(struct smu_context *smu); + +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_mode2_reset(struct smu_context *smu); + +extern int smu_get_atom_data_table(struct smu_context *smu, uint32_t table, + uint16_t *size, uint8_t *frev, uint8_t *crev, + uint8_t **addr); + +extern const struct amd_ip_funcs smu_ip_funcs; + +extern const struct amdgpu_ip_block_version smu_v11_0_ip_block; +extern const struct amdgpu_ip_block_version smu_v12_0_ip_block; + +extern int smu_feature_init_dpm(struct smu_context *smu); + +extern int smu_feature_is_enabled(struct smu_context *smu, + enum smu_feature_mask mask); +extern int smu_feature_set_enabled(struct smu_context *smu, + enum smu_feature_mask mask, bool enable); +extern int smu_feature_is_supported(struct smu_context *smu, + enum smu_feature_mask mask); +extern int smu_feature_set_supported(struct smu_context *smu, + enum smu_feature_mask mask, bool enable); + +int smu_update_table(struct smu_context *smu, enum smu_table_id table_index, int argument, + void *table_data, bool drv2smu); + +bool is_support_sw_smu(struct amdgpu_device *adev); +bool is_support_sw_smu_xgmi(struct amdgpu_device *adev); +int smu_reset(struct smu_context *smu); +int smu_common_read_sensor(struct smu_context *smu, enum amd_pp_sensors sensor, + void *data, uint32_t *size); +int smu_sys_get_pp_table(struct smu_context *smu, void **table); +int smu_sys_set_pp_table(struct smu_context *smu, void *buf, size_t size); +int smu_get_power_num_states(struct smu_context *smu, struct pp_states_info *state_info); +enum amd_pm_state_type smu_get_current_power_state(struct smu_context *smu); +int smu_write_watermarks_table(struct smu_context *smu); +int smu_set_watermarks_for_clock_ranges( + struct smu_context *smu, + struct dm_pp_wm_sets_with_clock_ranges_soc15 *clock_ranges); + +/* smu to display interface */ +extern int smu_display_configuration_change(struct smu_context *smu, const + struct amd_pp_display_configuration + *display_config); +extern int smu_get_current_clocks(struct smu_context *smu, + struct amd_pp_clock_info *clocks); +extern int smu_dpm_set_power_gate(struct smu_context *smu,uint32_t block_type, bool gate); +extern int smu_handle_task(struct smu_context *smu, + enum amd_dpm_forced_level level, + enum amd_pp_task task_id, + bool lock_needed); +int smu_switch_power_profile(struct smu_context *smu, + enum PP_SMC_POWER_PROFILE type, + bool en); +int smu_get_smc_version(struct smu_context *smu, uint32_t *if_version, uint32_t *smu_version); +int smu_get_dpm_freq_by_index(struct smu_context *smu, enum smu_clk_type clk_type, + uint16_t level, uint32_t *value); +int smu_get_dpm_level_count(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t *value); +int smu_get_dpm_freq_range(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t *min, uint32_t *max, bool lock_needed); +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); +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); +bool smu_clk_dpm_is_enabled(struct smu_context *smu, enum smu_clk_type clk_type); +const char *smu_get_message_name(struct smu_context *smu, enum smu_message_type type); +const char *smu_get_feature_name(struct smu_context *smu, enum smu_feature_mask feature); +size_t smu_sys_get_pp_feature_mask(struct smu_context *smu, char *buf); +int smu_sys_set_pp_feature_mask(struct smu_context *smu, uint64_t new_mask); +int smu_force_clk_levels(struct smu_context *smu, + enum smu_clk_type clk_type, + uint32_t mask, + bool lock_needed); +int smu_set_mp1_state(struct smu_context *smu, + enum pp_mp1_state mp1_state); +int smu_set_df_cstate(struct smu_context *smu, + enum pp_df_cstate state); + +int smu_get_max_sustainable_clocks_by_dc(struct smu_context *smu, + struct pp_smu_nv_clock_table *max_clocks); + +int smu_get_uclk_dpm_states(struct smu_context *smu, + unsigned int *clock_values_in_khz, + unsigned int *num_states); + +int smu_get_dpm_clock_table(struct smu_context *smu, + struct dpm_clocks *clock_table); + +uint32_t smu_get_pptable_power_limit(struct smu_context *smu); + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/inc/arcturus_ppsmc.h b/drivers/gpu/drm/amd/powerplay/inc/arcturus_ppsmc.h new file mode 100644 index 000000000000..e3291259b249 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/inc/arcturus_ppsmc.h @@ -0,0 +1,119 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef ARCTURUS_PP_SMC_H +#define ARCTURUS_PP_SMC_H + +#pragma pack(push, 1) + +// SMU Response Codes: +#define PPSMC_Result_OK 0x1 +#define PPSMC_Result_Failed 0xFF +#define PPSMC_Result_UnknownCmd 0xFE +#define PPSMC_Result_CmdRejectedPrereq 0xFD +#define PPSMC_Result_CmdRejectedBusy 0xFC + +// Message Definitions: +// BASIC +#define PPSMC_MSG_TestMessage 0x1 +#define PPSMC_MSG_GetSmuVersion 0x2 +#define PPSMC_MSG_GetDriverIfVersion 0x3 +#define PPSMC_MSG_SetAllowedFeaturesMaskLow 0x4 +#define PPSMC_MSG_SetAllowedFeaturesMaskHigh 0x5 +#define PPSMC_MSG_EnableAllSmuFeatures 0x6 +#define PPSMC_MSG_DisableAllSmuFeatures 0x7 +#define PPSMC_MSG_EnableSmuFeaturesLow 0x8 +#define PPSMC_MSG_EnableSmuFeaturesHigh 0x9 +#define PPSMC_MSG_DisableSmuFeaturesLow 0xA +#define PPSMC_MSG_DisableSmuFeaturesHigh 0xB +#define PPSMC_MSG_GetEnabledSmuFeaturesLow 0xC +#define PPSMC_MSG_GetEnabledSmuFeaturesHigh 0xD +#define PPSMC_MSG_SetDriverDramAddrHigh 0xE +#define PPSMC_MSG_SetDriverDramAddrLow 0xF +#define PPSMC_MSG_SetToolsDramAddrHigh 0x10 +#define PPSMC_MSG_SetToolsDramAddrLow 0x11 +#define PPSMC_MSG_TransferTableSmu2Dram 0x12 +#define PPSMC_MSG_TransferTableDram2Smu 0x13 +#define PPSMC_MSG_UseDefaultPPTable 0x14 +#define PPSMC_MSG_UseBackupPPTable 0x15 +#define PPSMC_MSG_SetSystemVirtualDramAddrHigh 0x16 +#define PPSMC_MSG_SetSystemVirtualDramAddrLow 0x17 + +//BACO/BAMACO/BOMACO +#define PPSMC_MSG_EnterBaco 0x18 +#define PPSMC_MSG_ExitBaco 0x19 +#define PPSMC_MSG_ArmD3 0x1A + +//DPM +#define PPSMC_MSG_SetSoftMinByFreq 0x1B +#define PPSMC_MSG_SetSoftMaxByFreq 0x1C +#define PPSMC_MSG_SetHardMinByFreq 0x1D +#define PPSMC_MSG_SetHardMaxByFreq 0x1E +#define PPSMC_MSG_GetMinDpmFreq 0x1F +#define PPSMC_MSG_GetMaxDpmFreq 0x20 +#define PPSMC_MSG_GetDpmFreqByIndex 0x21 + +#define PPSMC_MSG_SetWorkloadMask 0x22 +#define PPSMC_MSG_SetDfSwitchType 0x23 +#define PPSMC_MSG_GetVoltageByDpm 0x24 +#define PPSMC_MSG_GetVoltageByDpmOverdrive 0x25 + +#define PPSMC_MSG_SetPptLimit 0x26 +#define PPSMC_MSG_GetPptLimit 0x27 + +//Power Gating +#define PPSMC_MSG_PowerUpVcn0 0x28 +#define PPSMC_MSG_PowerDownVcn0 0x29 +#define PPSMC_MSG_PowerUpVcn1 0x2A +#define PPSMC_MSG_PowerDownVcn1 0x2B + +//Resets and reload +#define PPSMC_MSG_PrepareMp1ForUnload 0x2C +#define PPSMC_MSG_PrepareMp1ForReset 0x2D +#define PPSMC_MSG_PrepareMp1ForShutdown 0x2E +#define PPSMC_MSG_SoftReset 0x2F + +//BTC +#define PPSMC_MSG_RunAfllBtc 0x30 +#define PPSMC_MSG_RunDcBtc 0x31 + +//Debug +#define PPSMC_MSG_DramLogSetDramAddrHigh 0x33 +#define PPSMC_MSG_DramLogSetDramAddrLow 0x34 +#define PPSMC_MSG_DramLogSetDramSize 0x35 +#define PPSMC_MSG_GetDebugData 0x36 + +//WAFL and XGMI +#define PPSMC_MSG_WaflTest 0x37 +#define PPSMC_MSG_SetXgmiMode 0x38 + +//Others +#define PPSMC_MSG_SetMemoryChannelEnable 0x39 + +#define PPSMC_Message_Count 0x3A + +typedef uint32_t PPSMC_Result; +typedef uint32_t PPSMC_Msg; +#pragma pack(pop) + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h b/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h index bac3d85e3b82..af977675fd33 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h +++ b/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h @@ -189,7 +189,16 @@ struct phm_vce_clock_voltage_dependency_table { struct phm_vce_clock_voltage_dependency_record entries[1]; }; + +enum SMU_ASIC_RESET_MODE +{ + SMU_ASIC_RESET_MODE_0, + SMU_ASIC_RESET_MODE_1, + SMU_ASIC_RESET_MODE_2, +}; + struct pp_smumgr_func { + char *name; int (*smu_init)(struct pp_hwmgr *hwmgr); int (*smu_fini)(struct pp_hwmgr *hwmgr); int (*start_smu)(struct pp_hwmgr *hwmgr); @@ -343,6 +352,13 @@ struct pp_hwmgr_func { int (*set_asic_baco_state)(struct pp_hwmgr *hwmgr, enum BACO_STATE state); int (*get_ppfeature_status)(struct pp_hwmgr *hwmgr, char *buf); int (*set_ppfeature_status)(struct pp_hwmgr *hwmgr, uint64_t ppfeature_masks); + int (*set_mp1_state)(struct pp_hwmgr *hwmgr, enum pp_mp1_state mp1_state); + int (*asic_reset)(struct pp_hwmgr *hwmgr, enum SMU_ASIC_RESET_MODE mode); + int (*smu_i2c_bus_access)(struct pp_hwmgr *hwmgr, bool aquire); + int (*set_df_cstate)(struct pp_hwmgr *hwmgr, enum pp_df_cstate state); + int (*set_xgmi_pstate)(struct pp_hwmgr *hwmgr, uint32_t pstate); + int (*disable_power_features_for_compute_performance)(struct pp_hwmgr *hwmgr, + bool disable); }; struct pp_table_func { @@ -694,6 +710,7 @@ struct pp_thermal_controller_info { uint8_t ucType; uint8_t ucI2cLine; uint8_t ucI2cAddress; + uint8_t use_hw_fan_control; struct pp_fan_info fanInfo; struct pp_advance_fan_control_parameters advanceFanControlParameters; }; @@ -782,6 +799,7 @@ struct pp_hwmgr { uint32_t workload_mask; uint32_t workload_prority[Workload_Policy_Max]; uint32_t workload_setting[Workload_Policy_Max]; + bool gfxoff_state_changed_by_workload; }; int hwmgr_early_init(struct pp_hwmgr *hwmgr); diff --git a/drivers/gpu/drm/amd/powerplay/inc/power_state.h b/drivers/gpu/drm/amd/powerplay/inc/power_state.h index a99b5cbb113e..a5f2227a3971 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/power_state.h +++ b/drivers/gpu/drm/amd/powerplay/inc/power_state.h @@ -124,6 +124,13 @@ struct PP_StateSoftwareAlgorithmBlock { struct PP_TemperatureRange { int min; int max; + int edge_emergency_max; + int hotspot_min; + int hotspot_crit_max; + int hotspot_emergency_max; + int mem_min; + int mem_crit_max; + int mem_emergency_max; }; struct PP_StateValidationBlock { diff --git a/drivers/gpu/drm/amd/powerplay/inc/pp_thermal.h b/drivers/gpu/drm/amd/powerplay/inc/pp_thermal.h index 201d2b6329ab..3e30768f9e1c 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/pp_thermal.h +++ b/drivers/gpu/drm/amd/powerplay/inc/pp_thermal.h @@ -27,14 +27,18 @@ static const struct PP_TemperatureRange SMU7ThermalWithDelayPolicy[] = { - {-273150, 99000}, - { 120000, 120000}, + {-273150, 99000, 99000, -273150, 99000, 99000, -273150, 99000, 99000}, + { 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000}, }; static const struct PP_TemperatureRange SMU7ThermalPolicy[] = { - {-273150, 99000}, - { 120000, 120000}, + {-273150, 99000, 99000, -273150, 99000, 99000, -273150, 99000, 99000}, + { 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000}, }; +#define CTF_OFFSET_EDGE 5 +#define CTF_OFFSET_HOTSPOT 5 +#define CTF_OFFSET_HBM 5 + #endif diff --git a/drivers/gpu/drm/amd/powerplay/inc/rv_ppsmc.h b/drivers/gpu/drm/amd/powerplay/inc/rv_ppsmc.h index a2991fa2e6f8..df4677da736c 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/rv_ppsmc.h +++ b/drivers/gpu/drm/amd/powerplay/inc/rv_ppsmc.h @@ -59,7 +59,7 @@ #define PPSMC_MSG_SetDriverDramAddrLow 0x1B #define PPSMC_MSG_TransferTableSmu2Dram 0x1C #define PPSMC_MSG_TransferTableDram2Smu 0x1D -#define PPSMC_MSG_ControlGfxRM 0x1E +#define PPSMC_MSG_DeviceDriverReset 0x1E #define PPSMC_MSG_SetGfxclkOverdriveByFreqVid 0x1F #define PPSMC_MSG_SetHardMinDcefclkByFreq 0x20 #define PPSMC_MSG_SetHardMinSocclkByFreq 0x21 @@ -85,7 +85,6 @@ #define PPSMC_MSG_SetRccPfcPmeRestoreRegister 0x36 #define PPSMC_Message_Count 0x37 - typedef uint16_t PPSMC_Result; typedef int PPSMC_Msg; diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu10.h b/drivers/gpu/drm/amd/powerplay/inc/smu10.h index 9e837a5014c5..b96520528240 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/smu10.h +++ b/drivers/gpu/drm/amd/powerplay/inc/smu10.h @@ -136,12 +136,14 @@ #define FEATURE_CORE_CSTATES_MASK (1 << FEATURE_CORE_CSTATES_BIT) /* Workload bits */ -#define WORKLOAD_PPLIB_FULL_SCREEN_3D_BIT 0 -#define WORKLOAD_PPLIB_VIDEO_BIT 2 -#define WORKLOAD_PPLIB_VR_BIT 3 -#define WORKLOAD_PPLIB_COMPUTE_BIT 4 -#define WORKLOAD_PPLIB_CUSTOM_BIT 5 -#define WORKLOAD_PPLIB_COUNT 6 +#define WORKLOAD_DEFAULT_BIT 0 +#define WORKLOAD_PPLIB_FULL_SCREEN_3D_BIT 1 +#define WORKLOAD_PPLIB_POWER_SAVING_BIT 2 +#define WORKLOAD_PPLIB_VIDEO_BIT 3 +#define WORKLOAD_PPLIB_VR_BIT 4 +#define WORKLOAD_PPLIB_COMPUTE_BIT 5 +#define WORKLOAD_PPLIB_CUSTOM_BIT 6 +#define WORKLOAD_PPLIB_COUNT 7 typedef struct { /* MP1_EXT_SCRATCH0 */ diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu11_driver_if.h b/drivers/gpu/drm/amd/powerplay/inc/smu11_driver_if.h index 195c4ae67058..fdc6b7a57bc9 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/smu11_driver_if.h +++ b/drivers/gpu/drm/amd/powerplay/inc/smu11_driver_if.h @@ -27,7 +27,9 @@ // *** IMPORTANT *** // SMU TEAM: Always increment the interface version if // any structure is changed in this file -#define SMU11_DRIVER_IF_VERSION 0x12 +// Be aware of that the version should be updated in +// smu_v11_0.h, rename is also needed. +// #define SMU11_DRIVER_IF_VERSION 0x13 #define PPTABLE_V20_SMU_VERSION 3 @@ -615,6 +617,7 @@ typedef struct { uint16_t UclkAverageLpfTau; uint16_t GfxActivityLpfTau; uint16_t UclkActivityLpfTau; + uint16_t SocketPowerLpfTau; uint32_t MmHubPadding[8]; @@ -665,7 +668,8 @@ typedef struct { uint32_t ThrottlerStatus ; uint8_t LinkDpmLevel; - uint8_t Padding[3]; + uint16_t AverageSocketPower; + uint8_t Padding; uint32_t MmHubPadding[7]; 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 new file mode 100644 index 000000000000..a886f0644d24 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/inc/smu11_driver_if_arcturus.h @@ -0,0 +1,908 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef SMU11_DRIVER_IF_ARCTURUS_H +#define SMU11_DRIVER_IF_ARCTURUS_H + +// *** IMPORTANT *** +// SMU TEAM: Always increment the interface version if +// any structure is changed in this file +//#define SMU11_DRIVER_IF_VERSION 0x09 + +#define PPTABLE_ARCTURUS_SMU_VERSION 4 + +#define NUM_GFXCLK_DPM_LEVELS 16 +#define NUM_VCLK_DPM_LEVELS 8 +#define NUM_DCLK_DPM_LEVELS 8 +#define NUM_MP0CLK_DPM_LEVELS 2 +#define NUM_SOCCLK_DPM_LEVELS 8 +#define NUM_UCLK_DPM_LEVELS 4 +#define NUM_FCLK_DPM_LEVELS 8 +#define NUM_XGMI_LEVELS 2 +#define NUM_XGMI_PSTATE_LEVELS 4 + +#define MAX_GFXCLK_DPM_LEVEL (NUM_GFXCLK_DPM_LEVELS - 1) +#define MAX_VCLK_DPM_LEVEL (NUM_VCLK_DPM_LEVELS - 1) +#define MAX_DCLK_DPM_LEVEL (NUM_DCLK_DPM_LEVELS - 1) +#define MAX_MP0CLK_DPM_LEVEL (NUM_MP0CLK_DPM_LEVELS - 1) +#define MAX_SOCCLK_DPM_LEVEL (NUM_SOCCLK_DPM_LEVELS - 1) +#define MAX_UCLK_DPM_LEVEL (NUM_UCLK_DPM_LEVELS - 1) +#define MAX_FCLK_DPM_LEVEL (NUM_FCLK_DPM_LEVELS - 1) +#define MAX_XGMI_LEVEL (NUM_XGMI_LEVELS - 1) +#define MAX_XGMI_PSTATE_LEVEL (NUM_XGMI_PSTATE_LEVELS - 1) + +// Feature Control Defines +// DPM +#define FEATURE_DPM_PREFETCHER_BIT 0 +#define FEATURE_DPM_GFXCLK_BIT 1 +#define FEATURE_DPM_UCLK_BIT 2 +#define FEATURE_DPM_SOCCLK_BIT 3 +#define FEATURE_DPM_FCLK_BIT 4 +#define FEATURE_DPM_MP0CLK_BIT 5 +#define FEATURE_DPM_XGMI_BIT 6 +// Idle +#define FEATURE_DS_GFXCLK_BIT 7 +#define FEATURE_DS_SOCCLK_BIT 8 +#define FEATURE_DS_LCLK_BIT 9 +#define FEATURE_DS_FCLK_BIT 10 +#define FEATURE_DS_UCLK_BIT 11 +#define FEATURE_GFX_ULV_BIT 12 +#define FEATURE_DPM_VCN_BIT 13 +#define FEATURE_RSMU_SMN_CG_BIT 14 +#define FEATURE_WAFL_CG_BIT 15 +// Throttler/Response +#define FEATURE_PPT_BIT 16 +#define FEATURE_TDC_BIT 17 +#define FEATURE_APCC_PLUS_BIT 18 +#define FEATURE_VR0HOT_BIT 19 +#define FEATURE_VR1HOT_BIT 20 +#define FEATURE_FW_CTF_BIT 21 +#define FEATURE_FAN_CONTROL_BIT 22 +#define FEATURE_THERMAL_BIT 23 +// Other +#define FEATURE_OUT_OF_BAND_MONITOR_BIT 24 +#define FEATURE_TEMP_DEPENDENT_VMIN_BIT 25 + +#define FEATURE_SPARE_26_BIT 26 +#define FEATURE_SPARE_27_BIT 27 +#define FEATURE_SPARE_28_BIT 28 +#define FEATURE_SPARE_29_BIT 29 +#define FEATURE_SPARE_30_BIT 30 +#define FEATURE_SPARE_31_BIT 31 +#define FEATURE_SPARE_32_BIT 32 +#define FEATURE_SPARE_33_BIT 33 +#define FEATURE_SPARE_34_BIT 34 +#define FEATURE_SPARE_35_BIT 35 +#define FEATURE_SPARE_36_BIT 36 +#define FEATURE_SPARE_37_BIT 37 +#define FEATURE_SPARE_38_BIT 38 +#define FEATURE_SPARE_39_BIT 39 +#define FEATURE_SPARE_40_BIT 40 +#define FEATURE_SPARE_41_BIT 41 +#define FEATURE_SPARE_42_BIT 42 +#define FEATURE_SPARE_43_BIT 43 +#define FEATURE_SPARE_44_BIT 44 +#define FEATURE_SPARE_45_BIT 45 +#define FEATURE_SPARE_46_BIT 46 +#define FEATURE_SPARE_47_BIT 47 +#define FEATURE_SPARE_48_BIT 48 +#define FEATURE_SPARE_49_BIT 49 +#define FEATURE_SPARE_50_BIT 50 +#define FEATURE_SPARE_51_BIT 51 +#define FEATURE_SPARE_52_BIT 52 +#define FEATURE_SPARE_53_BIT 53 +#define FEATURE_SPARE_54_BIT 54 +#define FEATURE_SPARE_55_BIT 55 +#define FEATURE_SPARE_56_BIT 56 +#define FEATURE_SPARE_57_BIT 57 +#define FEATURE_SPARE_58_BIT 58 +#define FEATURE_SPARE_59_BIT 59 +#define FEATURE_SPARE_60_BIT 60 +#define FEATURE_SPARE_61_BIT 61 +#define FEATURE_SPARE_62_BIT 62 +#define FEATURE_SPARE_63_BIT 63 + +#define NUM_FEATURES 64 + + +#define FEATURE_DPM_PREFETCHER_MASK (1 << FEATURE_DPM_PREFETCHER_BIT ) +#define FEATURE_DPM_GFXCLK_MASK (1 << FEATURE_DPM_GFXCLK_BIT ) +#define FEATURE_DPM_UCLK_MASK (1 << FEATURE_DPM_UCLK_BIT ) +#define FEATURE_DPM_SOCCLK_MASK (1 << FEATURE_DPM_SOCCLK_BIT ) +#define FEATURE_DPM_FCLK_MASK (1 << FEATURE_DPM_FCLK_BIT ) +#define FEATURE_DPM_MP0CLK_MASK (1 << FEATURE_DPM_MP0CLK_BIT ) +#define FEATURE_DPM_XGMI_MASK (1 << FEATURE_DPM_XGMI_BIT ) + +#define FEATURE_DS_GFXCLK_MASK (1 << FEATURE_DS_GFXCLK_BIT ) +#define FEATURE_DS_SOCCLK_MASK (1 << FEATURE_DS_SOCCLK_BIT ) +#define FEATURE_DS_LCLK_MASK (1 << FEATURE_DS_LCLK_BIT ) +#define FEATURE_DS_FCLK_MASK (1 << FEATURE_DS_FCLK_BIT ) +#define FEATURE_DS_UCLK_MASK (1 << FEATURE_DS_UCLK_BIT ) +#define FEATURE_GFX_ULV_MASK (1 << FEATURE_GFX_ULV_BIT ) +#define FEATURE_DPM_VCN_MASK (1 << FEATURE_DPM_VCN_BIT ) +#define FEATURE_RSMU_SMN_CG_MASK (1 << FEATURE_RSMU_SMN_CG_BIT ) +#define FEATURE_WAFL_CG_MASK (1 << FEATURE_WAFL_CG_BIT ) + +#define FEATURE_PPT_MASK (1 << FEATURE_PPT_BIT ) +#define FEATURE_TDC_MASK (1 << FEATURE_TDC_BIT ) +#define FEATURE_APCC_PLUS_MASK (1 << FEATURE_APCC_PLUS_BIT ) +#define FEATURE_VR0HOT_MASK (1 << FEATURE_VR0HOT_BIT ) +#define FEATURE_VR1HOT_MASK (1 << FEATURE_VR1HOT_BIT ) +#define FEATURE_FW_CTF_MASK (1 << FEATURE_FW_CTF_BIT ) +#define FEATURE_FAN_CONTROL_MASK (1 << FEATURE_FAN_CONTROL_BIT ) +#define FEATURE_THERMAL_MASK (1 << FEATURE_THERMAL_BIT ) + +#define FEATURE_OUT_OF_BAND_MONITOR_MASK (1 << FEATURE_OUT_OF_BAND_MONITOR_BIT ) +#define FEATURE_TEMP_DEPENDENT_VMIN_MASK (1 << FEATURE_TEMP_DEPENDENT_VMIN_BIT ) + + +//FIXME need updating +// Debug Overrides Bitmask +#define DPM_OVERRIDE_DISABLE_UCLK_PID 0x00000001 +#define DPM_OVERRIDE_DISABLE_VOLT_LINK_VCN_FCLK 0x00000002 + +// I2C Config Bit Defines +#define I2C_CONTROLLER_ENABLED 1 +#define I2C_CONTROLLER_DISABLED 0 + +// VR Mapping Bit Defines +#define VR_MAPPING_VR_SELECT_MASK 0x01 +#define VR_MAPPING_VR_SELECT_SHIFT 0x00 + +#define VR_MAPPING_PLANE_SELECT_MASK 0x02 +#define VR_MAPPING_PLANE_SELECT_SHIFT 0x01 + +// PSI Bit Defines +#define PSI_SEL_VR0_PLANE0_PSI0 0x01 +#define PSI_SEL_VR0_PLANE0_PSI1 0x02 +#define PSI_SEL_VR0_PLANE1_PSI0 0x04 +#define PSI_SEL_VR0_PLANE1_PSI1 0x08 +#define PSI_SEL_VR1_PLANE0_PSI0 0x10 +#define PSI_SEL_VR1_PLANE0_PSI1 0x20 +#define PSI_SEL_VR1_PLANE1_PSI0 0x40 +#define PSI_SEL_VR1_PLANE1_PSI1 0x80 + +// Throttler Control/Status Bits +#define THROTTLER_PADDING_BIT 0 +#define THROTTLER_TEMP_EDGE_BIT 1 +#define THROTTLER_TEMP_HOTSPOT_BIT 2 +#define THROTTLER_TEMP_MEM_BIT 3 +#define THROTTLER_TEMP_VR_GFX_BIT 4 +#define THROTTLER_TEMP_VR_MEM_BIT 5 +#define THROTTLER_TEMP_VR_SOC_BIT 6 +#define THROTTLER_TDC_GFX_BIT 7 +#define THROTTLER_TDC_SOC_BIT 8 +#define THROTTLER_PPT0_BIT 9 +#define THROTTLER_PPT1_BIT 10 +#define THROTTLER_PPT2_BIT 11 +#define THROTTLER_PPT3_BIT 12 +#define THROTTLER_PPM_BIT 13 +#define THROTTLER_FIT_BIT 14 +#define THROTTLER_APCC_BIT 15 + +// Table transfer status +#define TABLE_TRANSFER_OK 0x0 +#define TABLE_TRANSFER_FAILED 0xFF +#define TABLE_TRANSFER_PENDING 0xAB + +// Workload bits +#define WORKLOAD_PPLIB_DEFAULT_BIT 0 +#define WORKLOAD_PPLIB_POWER_SAVING_BIT 1 +#define WORKLOAD_PPLIB_VIDEO_BIT 2 +#define WORKLOAD_PPLIB_COMPUTE_BIT 3 +#define WORKLOAD_PPLIB_CUSTOM_BIT 4 +#define WORKLOAD_PPLIB_COUNT 5 + +//XGMI performance states +#define XGMI_STATE_D0 1 +#define XGMI_STATE_D3 0 + +#define NUM_I2C_CONTROLLERS 8 + +#define I2C_CONTROLLER_ENABLED 1 +#define I2C_CONTROLLER_DISABLED 0 + +#define MAX_SW_I2C_COMMANDS 8 + +typedef enum { + I2C_CONTROLLER_PORT_0 = 0, //CKSVII2C0 + I2C_CONTROLLER_PORT_1 = 1, //CKSVII2C1 + I2C_CONTROLLER_PORT_COUNT, +} I2cControllerPort_e; + +typedef enum { + I2C_CONTROLLER_NAME_VR_GFX = 0, + I2C_CONTROLLER_NAME_VR_SOC, + I2C_CONTROLLER_NAME_VR_MEM, + I2C_CONTROLLER_NAME_SPARE, + I2C_CONTROLLER_NAME_COUNT, +} I2cControllerName_e; + +typedef enum { + I2C_CONTROLLER_THROTTLER_TYPE_NONE = 0, + I2C_CONTROLLER_THROTTLER_VR_GFX, + I2C_CONTROLLER_THROTTLER_VR_SOC, + I2C_CONTROLLER_THROTTLER_VR_MEM, + I2C_CONTROLLER_THROTTLER_COUNT, +} I2cControllerThrottler_e; + +typedef enum { + I2C_CONTROLLER_PROTOCOL_VR_0, + I2C_CONTROLLER_PROTOCOL_VR_1, + I2C_CONTROLLER_PROTOCOL_TMP_0, + I2C_CONTROLLER_PROTOCOL_TMP_1, + I2C_CONTROLLER_PROTOCOL_SPARE_0, + I2C_CONTROLLER_PROTOCOL_SPARE_1, + I2C_CONTROLLER_PROTOCOL_COUNT, +} I2cControllerProtocol_e; + +typedef struct { + uint8_t Enabled; + uint8_t Speed; + uint8_t Padding[2]; + uint32_t SlaveAddress; + uint8_t ControllerPort; + uint8_t ControllerName; + uint8_t ThermalThrotter; + uint8_t I2cProtocol; +} I2cControllerConfig_t; + +typedef enum { + I2C_PORT_SVD_SCL = 0, + I2C_PORT_GPIO, +} I2cPort_e; + +typedef enum { + I2C_SPEED_FAST_50K = 0, //50 Kbits/s + I2C_SPEED_FAST_100K, //100 Kbits/s + I2C_SPEED_FAST_400K, //400 Kbits/s + I2C_SPEED_FAST_PLUS_1M, //1 Mbits/s (in fast mode) + I2C_SPEED_HIGH_1M, //1 Mbits/s (in high speed mode) + I2C_SPEED_HIGH_2M, //2.3 Mbits/s + I2C_SPEED_COUNT, +} I2cSpeed_e; + +typedef enum { + I2C_CMD_READ = 0, + I2C_CMD_WRITE, + I2C_CMD_COUNT, +} I2cCmdType_e; + +#define CMDCONFIG_STOP_BIT 0 +#define CMDCONFIG_RESTART_BIT 1 + +#define CMDCONFIG_STOP_MASK (1 << CMDCONFIG_STOP_BIT) +#define CMDCONFIG_RESTART_MASK (1 << CMDCONFIG_RESTART_BIT) + +typedef struct { + uint8_t RegisterAddr; ////only valid for write, ignored for read + uint8_t Cmd; //Read(0) or Write(1) + uint8_t Data; //Return data for read. Data to send for write + uint8_t CmdConfig; //Includes whether associated command should have a stop or restart command +} SwI2cCmd_t; //SW I2C Command Table + +typedef struct { + uint8_t I2CcontrollerPort; //CKSVII2C0(0) or //CKSVII2C1(1) + uint8_t I2CSpeed; //Slow(0) or Fast(1) + uint16_t SlaveAddress; + uint8_t NumCmds; //Number of commands + uint8_t Padding[3]; + + SwI2cCmd_t SwI2cCmds[MAX_SW_I2C_COMMANDS]; + + uint32_t MmHubPadding[8]; // SMU internal use + +} SwI2cRequest_t; // SW I2C Request Table + +//D3HOT sequences +typedef enum { + BACO_SEQUENCE, + MSR_SEQUENCE, + BAMACO_SEQUENCE, + ULPS_SEQUENCE, + D3HOT_SEQUENCE_COUNT, +}D3HOTSequence_e; + +//THis is aligned with RSMU PGFSM Register Mapping +typedef enum { + PG_DYNAMIC_MODE = 0, + PG_STATIC_MODE, +} PowerGatingMode_e; + +//This is aligned with RSMU PGFSM Register Mapping +typedef enum { + PG_POWER_DOWN = 0, + PG_POWER_UP, +} PowerGatingSettings_e; + +typedef struct { + uint32_t a; // store in IEEE float format in this variable + uint32_t b; // store in IEEE float format in this variable + uint32_t c; // store in IEEE float format in this variable +} QuadraticInt_t; + +typedef struct { + uint32_t m; // store in IEEE float format in this variable + uint32_t b; // store in IEEE float format in this variable +} LinearInt_t; + +typedef struct { + uint32_t a; // store in IEEE float format in this variable + uint32_t b; // store in IEEE float format in this variable + uint32_t c; // store in IEEE float format in this variable +} DroopInt_t; + +typedef enum { + GFXCLK_SOURCE_PLL = 0, + GFXCLK_SOURCE_AFLL, + GFXCLK_SOURCE_COUNT, +} GfxclkSrc_e; + +typedef enum { + PPCLK_GFXCLK, + PPCLK_VCLK, + PPCLK_DCLK, + PPCLK_SOCCLK, + PPCLK_UCLK, + PPCLK_FCLK, + PPCLK_COUNT, +} PPCLK_e; + +typedef enum { + POWER_SOURCE_AC, + POWER_SOURCE_DC, + POWER_SOURCE_COUNT, +} POWER_SOURCE_e; + +typedef enum { + TEMP_EDGE, + TEMP_HOTSPOT, + TEMP_MEM, + TEMP_VR_GFX, + TEMP_VR_SOC, + TEMP_VR_MEM, + TEMP_COUNT +} TEMP_TYPE_e; + +typedef enum { + PPT_THROTTLER_PPT0, + PPT_THROTTLER_PPT1, + PPT_THROTTLER_PPT2, + PPT_THROTTLER_PPT3, + PPT_THROTTLER_COUNT +} PPT_THROTTLER_e; + +typedef enum { + VOLTAGE_MODE_AVFS = 0, + VOLTAGE_MODE_AVFS_SS, + VOLTAGE_MODE_SS, + VOLTAGE_MODE_COUNT, +} VOLTAGE_MODE_e; + +typedef enum { + AVFS_VOLTAGE_GFX = 0, + AVFS_VOLTAGE_SOC, + AVFS_VOLTAGE_COUNT, +} AVFS_VOLTAGE_TYPE_e; + +typedef enum { + GPIO_INT_POLARITY_ACTIVE_LOW = 0, + GPIO_INT_POLARITY_ACTIVE_HIGH, +} GpioIntPolarity_e; + +typedef enum { + MEMORY_TYPE_GDDR6 = 0, + MEMORY_TYPE_HBM, +} MemoryType_e; + +typedef enum { + PWR_CONFIG_TDP = 0, + PWR_CONFIG_TGP, + PWR_CONFIG_TCP_ESTIMATED, + PWR_CONFIG_TCP_MEASURED, +} PwrConfig_e; + +typedef enum { + XGMI_LINK_RATE_2 = 2, // 2Gbps + XGMI_LINK_RATE_4 = 4, // 4Gbps + XGMI_LINK_RATE_8 = 8, // 8Gbps + XGMI_LINK_RATE_12 = 12, // 12Gbps + XGMI_LINK_RATE_16 = 16, // 16Gbps + XGMI_LINK_RATE_17 = 17, // 17Gbps + XGMI_LINK_RATE_18 = 18, // 18Gbps + XGMI_LINK_RATE_19 = 19, // 19Gbps + XGMI_LINK_RATE_20 = 20, // 20Gbps + XGMI_LINK_RATE_21 = 21, // 21Gbps + XGMI_LINK_RATE_22 = 22, // 22Gbps + XGMI_LINK_RATE_23 = 23, // 23Gbps + XGMI_LINK_RATE_24 = 24, // 24Gbps + XGMI_LINK_RATE_25 = 25, // 25Gbps + XGMI_LINK_RATE_COUNT +} XGMI_LINK_RATE_e; + +typedef enum { + XGMI_LINK_WIDTH_1 = 1, // x1 + XGMI_LINK_WIDTH_2 = 2, // x2 + XGMI_LINK_WIDTH_4 = 4, // x4 + XGMI_LINK_WIDTH_8 = 8, // x8 + XGMI_LINK_WIDTH_9 = 9, // x9 + XGMI_LINK_WIDTH_16 = 16, // x16 + XGMI_LINK_WIDTH_COUNT +} XGMI_LINK_WIDTH_e; + +typedef struct { + uint8_t VoltageMode; // 0 - AVFS only, 1- min(AVFS,SS), 2-SS only + uint8_t SnapToDiscrete; // 0 - Fine grained DPM, 1 - Discrete DPM + uint8_t NumDiscreteLevels; // Set to 2 (Fmin, Fmax) when using fine grained DPM, otherwise set to # discrete levels used + uint8_t padding; + LinearInt_t ConversionToAvfsClk; // Transfer function to AVFS Clock (GHz->GHz) + QuadraticInt_t SsCurve; // Slow-slow curve (GHz->V) + uint16_t SsFmin; // Fmin for SS curve. If SS curve is selected, will use V@SSFmin for F <= Fmin + uint16_t Padding16; +} DpmDescriptor_t; + +typedef struct { + uint32_t Version; + + // SECTION: Feature Enablement + uint32_t FeaturesToRun[2]; + + // SECTION: Infrastructure Limits + uint16_t SocketPowerLimitAc[PPT_THROTTLER_COUNT]; + uint16_t SocketPowerLimitAcTau[PPT_THROTTLER_COUNT]; + uint16_t TdcLimitSoc; // Amps + uint16_t TdcLimitSocTau; // Time constant of LPF in ms + uint16_t TdcLimitGfx; // Amps + uint16_t TdcLimitGfxTau; // Time constant of LPF in ms + + uint16_t TedgeLimit; // Celcius + uint16_t ThotspotLimit; // Celcius + uint16_t TmemLimit; // Celcius + uint16_t Tvr_gfxLimit; // Celcius + uint16_t Tvr_memLimit; // Celcius + uint16_t Tvr_socLimit; // Celcius + uint32_t FitLimit; // Failures in time (failures per million parts over the defined lifetime) + + uint16_t PpmPowerLimit; // Switch this this power limit when temperature is above PpmTempThreshold + uint16_t PpmTemperatureThreshold; + + // SECTION: Throttler settings + uint32_t ThrottlerControlMask; // See Throtter masks defines + + // SECTION: ULV Settings + uint16_t UlvVoltageOffsetGfx; // In mV(Q2) + uint16_t UlvPadding; // Padding + + uint8_t UlvGfxclkBypass; // 1 to turn off/bypass Gfxclk during ULV, 0 to leave Gfxclk on during ULV + uint8_t Padding234[3]; + + // SECTION: Voltage Control Parameters + uint16_t MinVoltageGfx; // In mV(Q2) Minimum Voltage ("Vmin") of VDD_GFX + uint16_t MinVoltageSoc; // In mV(Q2) Minimum Voltage ("Vmin") of VDD_SOC + uint16_t MaxVoltageGfx; // In mV(Q2) Maximum Voltage allowable of VDD_GFX + uint16_t MaxVoltageSoc; // In mV(Q2) Maximum Voltage allowable of VDD_SOC + + uint16_t LoadLineResistanceGfx; // In mOhms with 8 fractional bits + uint16_t LoadLineResistanceSoc; // In mOhms with 8 fractional bits + + //SECTION: DPM Config 1 + DpmDescriptor_t DpmDescriptor[PPCLK_COUNT]; + + uint16_t FreqTableGfx [NUM_GFXCLK_DPM_LEVELS ]; // In MHz + uint16_t FreqTableVclk [NUM_VCLK_DPM_LEVELS ]; // In MHz + uint16_t FreqTableDclk [NUM_DCLK_DPM_LEVELS ]; // In MHz + uint16_t FreqTableSocclk [NUM_SOCCLK_DPM_LEVELS ]; // In MHz + uint16_t FreqTableUclk [NUM_UCLK_DPM_LEVELS ]; // In MHz + uint16_t FreqTableFclk [NUM_FCLK_DPM_LEVELS ]; // In MHz + + uint32_t Paddingclks[16]; + + // SECTION: DPM Config 2 + uint16_t Mp0clkFreq [NUM_MP0CLK_DPM_LEVELS]; // in MHz + uint16_t Mp0DpmVoltage [NUM_MP0CLK_DPM_LEVELS]; // mV(Q2) + + // GFXCLK DPM + uint16_t GfxclkFidle; // In MHz + uint16_t GfxclkSlewRate; // for PLL babystepping??? + uint8_t Padding567[4]; + uint16_t GfxclkDsMaxFreq; // In MHz + uint8_t GfxclkSource; // 0 = PLL, 1 = AFLL + uint8_t Padding456; + + // GFXCLK Thermal DPM (formerly 'Boost' Settings) + uint16_t EnableTdpm; + uint16_t TdpmHighHystTemperature; + uint16_t TdpmLowHystTemperature; + uint16_t GfxclkFreqHighTempLimit; // High limit on GFXCLK when temperature is high, for reliability. + + // SECTION: Fan Control + uint16_t FanStopTemp; //Celcius + uint16_t FanStartTemp; //Celcius + + uint16_t FanGainEdge; + uint16_t FanGainHotspot; + uint16_t FanGainVrGfx; + uint16_t FanGainVrSoc; + uint16_t FanGainVrMem; + uint16_t FanGainHbm; + uint16_t FanPwmMin; + uint16_t FanAcousticLimitRpm; + uint16_t FanThrottlingRpm; + uint16_t FanMaximumRpm; + uint16_t FanTargetTemperature; + uint16_t FanTargetGfxclk; + uint8_t FanZeroRpmEnable; + uint8_t FanTachEdgePerRev; + uint8_t FanTempInputSelect; + uint8_t padding8_Fan; + + // The following are AFC override parameters. Leave at 0 to use FW defaults. + int16_t FuzzyFan_ErrorSetDelta; + int16_t FuzzyFan_ErrorRateSetDelta; + int16_t FuzzyFan_PwmSetDelta; + uint16_t FuzzyFan_Reserved; + + + // SECTION: AVFS + // Overrides + uint8_t OverrideAvfsGb[AVFS_VOLTAGE_COUNT]; + uint8_t Padding8_Avfs[2]; + + QuadraticInt_t qAvfsGb[AVFS_VOLTAGE_COUNT]; // GHz->V Override of fused curve + DroopInt_t dBtcGbGfxPll; // GHz->V BtcGb + DroopInt_t dBtcGbGfxAfll; // GHz->V BtcGb + DroopInt_t dBtcGbSoc; // GHz->V BtcGb + LinearInt_t qAgingGb[AVFS_VOLTAGE_COUNT]; // GHz->V + + QuadraticInt_t qStaticVoltageOffset[AVFS_VOLTAGE_COUNT]; // GHz->V + + uint16_t DcTol[AVFS_VOLTAGE_COUNT]; // mV Q2 + + uint8_t DcBtcEnabled[AVFS_VOLTAGE_COUNT]; + uint8_t Padding8_GfxBtc[2]; + + uint16_t DcBtcMin[AVFS_VOLTAGE_COUNT]; // mV Q2 + uint16_t DcBtcMax[AVFS_VOLTAGE_COUNT]; // mV Q2 + + uint16_t DcBtcGb[AVFS_VOLTAGE_COUNT]; // mV Q2 + + // SECTION: XGMI + uint8_t XgmiDpmPstates[NUM_XGMI_LEVELS]; // 2 DPM states, high and low. 0-P0, 1-P1, 2-P2, 3-P3. + uint8_t XgmiDpmSpare[2]; + + // Temperature Dependent Vmin + uint16_t VDDGFX_TVmin; //Celcius + uint16_t VDDSOC_TVmin; //Celcius + uint16_t VDDGFX_Vmin_HiTemp; // mV Q2 + uint16_t VDDGFX_Vmin_LoTemp; // mV Q2 + uint16_t VDDSOC_Vmin_HiTemp; // mV Q2 + uint16_t VDDSOC_Vmin_LoTemp; // mV Q2 + + uint16_t VDDGFX_TVminHystersis; // Celcius + uint16_t VDDSOC_TVminHystersis; // Celcius + + + // SECTION: Advanced Options + uint32_t DebugOverrides; + QuadraticInt_t ReservedEquation0; + QuadraticInt_t ReservedEquation1; + QuadraticInt_t ReservedEquation2; + QuadraticInt_t ReservedEquation3; + + uint16_t MinVoltageUlvGfx; // In mV(Q2) Minimum Voltage ("Vmin") of VDD_GFX in ULV mode + uint16_t PaddingUlv; // Padding + + // Total Power configuration, use defines from PwrConfig_e + uint8_t TotalPowerConfig; //0-TDP, 1-TGP, 2-TCP Estimated, 3-TCP Measured + uint8_t TotalPowerSpare1; + uint16_t TotalPowerSpare2; + + // APCC Settings + uint16_t PccThresholdLow; + uint16_t PccThresholdHigh; + uint32_t PaddingAPCC[6]; //FIXME pending SPEC + + // SECTION: Reserved + uint32_t Reserved[11]; + + // SECTION: BOARD PARAMETERS + + // SVI2 Board Parameters + uint16_t MaxVoltageStepGfx; // In mV(Q2) Max voltage step that SMU will request. Multiple steps are taken if voltage change exceeds this value. + uint16_t MaxVoltageStepSoc; // In mV(Q2) Max voltage step that SMU will request. Multiple steps are taken if voltage change exceeds this value. + + uint8_t VddGfxVrMapping; // Use VR_MAPPING* bitfields + uint8_t VddSocVrMapping; // Use VR_MAPPING* bitfields + uint8_t VddMemVrMapping; // Use VR_MAPPING* bitfields + uint8_t BoardVrMapping; // Use VR_MAPPING* bitfields + + uint8_t GfxUlvPhaseSheddingMask; // set this to 1 to set PSI0/1 to 1 in ULV mode + uint8_t ExternalSensorPresent; // External RDI connected to TMON (aka TEMP IN) + uint8_t Padding8_V[2]; + + // Telemetry Settings + uint16_t GfxMaxCurrent; // in Amps + int8_t GfxOffset; // in Amps + uint8_t Padding_TelemetryGfx; + + uint16_t SocMaxCurrent; // in Amps + int8_t SocOffset; // in Amps + uint8_t Padding_TelemetrySoc; + + uint16_t MemMaxCurrent; // in Amps + int8_t MemOffset; // in Amps + uint8_t Padding_TelemetryMem; + + uint16_t BoardMaxCurrent; // in Amps + int8_t BoardOffset; // in Amps + uint8_t Padding_TelemetryBoardInput; + + // GPIO Settings + uint8_t VR0HotGpio; // GPIO pin configured for VR0 HOT event + uint8_t VR0HotPolarity; // GPIO polarity for VR0 HOT event + uint8_t VR1HotGpio; // GPIO pin configured for VR1 HOT event + uint8_t VR1HotPolarity; // GPIO polarity for VR1 HOT event + + // GFXCLK PLL Spread Spectrum + uint8_t PllGfxclkSpreadEnabled; // on or off + uint8_t PllGfxclkSpreadPercent; // Q4.4 + uint16_t PllGfxclkSpreadFreq; // kHz + + // UCLK Spread Spectrum + uint8_t UclkSpreadEnabled; // on or off + uint8_t UclkSpreadPercent; // Q4.4 + uint16_t UclkSpreadFreq; // kHz + + // FCLK Spread Spectrum + uint8_t FclkSpreadEnabled; // on or off + uint8_t FclkSpreadPercent; // Q4.4 + uint16_t FclkSpreadFreq; // kHz + + // GFXCLK Fll Spread Spectrum + uint8_t FllGfxclkSpreadEnabled; // on or off + uint8_t FllGfxclkSpreadPercent; // Q4.4 + uint16_t FllGfxclkSpreadFreq; // kHz + + // I2C Controller Structure + I2cControllerConfig_t I2cControllers[NUM_I2C_CONTROLLERS]; + + // Memory section + uint32_t MemoryChannelEnabled; // For DRAM use only, Max 32 channels enabled bit mask. + + uint8_t DramBitWidth; // For DRAM use only. See Dram Bit width type defines + uint8_t PaddingMem[3]; + + // Total board power + uint16_t TotalBoardPower; //Only needed for TCP Estimated case, where TCP = TGP+Total Board Power + uint16_t BoardPadding; + + // SECTION: XGMI Training + uint8_t XgmiLinkSpeed [NUM_XGMI_PSTATE_LEVELS]; + uint8_t XgmiLinkWidth [NUM_XGMI_PSTATE_LEVELS]; + + uint16_t XgmiFclkFreq [NUM_XGMI_PSTATE_LEVELS]; + uint16_t XgmiSocVoltage [NUM_XGMI_PSTATE_LEVELS]; + + // GPIO pins for I2C communications with 2nd controller for Input Telemetry Sequence + uint8_t GpioI2cScl; // Serial Clock + uint8_t GpioI2cSda; // Serial Data + uint16_t GpioPadding; + + // Platform input telemetry voltage coefficient + uint32_t BoardVoltageCoeffA; // decode by /1000 + uint32_t BoardVoltageCoeffB; // decode by /1000 + + uint32_t BoardReserved[7]; + + // Padding for MMHUB - do not modify this + uint32_t MmHubPadding[8]; // SMU internal use + +} PPTable_t; + +typedef struct { + // Time constant parameters for clock averages in ms + uint16_t GfxclkAverageLpfTau; + uint16_t SocclkAverageLpfTau; + uint16_t UclkAverageLpfTau; + uint16_t GfxActivityLpfTau; + uint16_t UclkActivityLpfTau; + + uint16_t SocketPowerLpfTau; + + // Padding - ignore + uint32_t MmHubPadding[8]; // SMU internal use +} DriverSmuConfig_t; + +typedef struct { + uint16_t CurrClock[PPCLK_COUNT]; + uint16_t AverageGfxclkFrequency; + uint16_t AverageSocclkFrequency; + uint16_t AverageUclkFrequency ; + uint16_t AverageGfxActivity ; + uint16_t AverageUclkActivity ; + uint8_t CurrSocVoltageOffset ; + uint8_t CurrGfxVoltageOffset ; + uint8_t CurrMemVidOffset ; + uint8_t Padding8 ; + uint16_t AverageSocketPower ; + uint16_t TemperatureEdge ; + uint16_t TemperatureHotspot ; + uint16_t TemperatureHBM ; + uint16_t TemperatureVrGfx ; + uint16_t TemperatureVrSoc ; + uint16_t TemperatureVrMem ; + uint32_t ThrottlerStatus ; + + uint16_t CurrFanSpeed ; + uint16_t Padding16; + + uint32_t Padding[4]; + + // Padding - ignore + uint32_t MmHubPadding[8]; // SMU internal use +} SmuMetrics_t; + + +typedef struct { + uint16_t avgPsmCount[75]; + uint16_t minPsmCount[75]; + float avgPsmVoltage[75]; + float minPsmVoltage[75]; + + uint32_t MmHubPadding[8]; // SMU internal use +} AvfsDebugTable_t; + +typedef struct { + uint8_t AvfsVersion; + uint8_t Padding; + uint8_t AvfsEn[AVFS_VOLTAGE_COUNT]; + + uint8_t OverrideVFT[AVFS_VOLTAGE_COUNT]; + uint8_t OverrideAvfsGb[AVFS_VOLTAGE_COUNT]; + + uint8_t OverrideTemperatures[AVFS_VOLTAGE_COUNT]; + uint8_t OverrideVInversion[AVFS_VOLTAGE_COUNT]; + uint8_t OverrideP2V[AVFS_VOLTAGE_COUNT]; + uint8_t OverrideP2VCharzFreq[AVFS_VOLTAGE_COUNT]; + + int32_t VFT0_m1[AVFS_VOLTAGE_COUNT]; // Q8.24 + int32_t VFT0_m2[AVFS_VOLTAGE_COUNT]; // Q12.12 + int32_t VFT0_b[AVFS_VOLTAGE_COUNT]; // Q32 + + int32_t VFT1_m1[AVFS_VOLTAGE_COUNT]; // Q8.16 + int32_t VFT1_m2[AVFS_VOLTAGE_COUNT]; // Q12.12 + int32_t VFT1_b[AVFS_VOLTAGE_COUNT]; // Q32 + + int32_t VFT2_m1[AVFS_VOLTAGE_COUNT]; // Q8.16 + int32_t VFT2_m2[AVFS_VOLTAGE_COUNT]; // Q12.12 + int32_t VFT2_b[AVFS_VOLTAGE_COUNT]; // Q32 + + int32_t AvfsGb0_m1[AVFS_VOLTAGE_COUNT]; // Q8.24 + int32_t AvfsGb0_m2[AVFS_VOLTAGE_COUNT]; // Q12.12 + int32_t AvfsGb0_b[AVFS_VOLTAGE_COUNT]; // Q32 + + int32_t AcBtcGb_m1[AVFS_VOLTAGE_COUNT]; // Q8.24 + int32_t AcBtcGb_m2[AVFS_VOLTAGE_COUNT]; // Q12.12 + int32_t AcBtcGb_b[AVFS_VOLTAGE_COUNT]; // Q32 + + uint32_t AvfsTempCold[AVFS_VOLTAGE_COUNT]; + uint32_t AvfsTempMid[AVFS_VOLTAGE_COUNT]; + uint32_t AvfsTempHot[AVFS_VOLTAGE_COUNT]; + + uint32_t VInversion[AVFS_VOLTAGE_COUNT]; // in mV with 2 fractional bits + + + int32_t P2V_m1[AVFS_VOLTAGE_COUNT]; // Q8.24 + int32_t P2V_m2[AVFS_VOLTAGE_COUNT]; // Q12.12 + int32_t P2V_b[AVFS_VOLTAGE_COUNT]; // Q32 + + uint32_t P2VCharzFreq[AVFS_VOLTAGE_COUNT]; // in 10KHz units + + uint32_t EnabledAvfsModules[3]; + + uint32_t MmHubPadding[8]; // SMU internal use +} AvfsFuseOverride_t; + +/* NOT CURRENTLY USED +typedef struct { + uint8_t Gfx_ActiveHystLimit; + uint8_t Gfx_IdleHystLimit; + uint8_t Gfx_FPS; + uint8_t Gfx_MinActiveFreqType; + uint8_t Gfx_BoosterFreqType; + uint8_t Gfx_MinFreqStep; // Minimum delta between current and target frequeny in order for FW to change clock. + uint8_t Gfx_UseRlcBusy; + uint8_t PaddingGfx[3]; + uint16_t Gfx_MinActiveFreq; // MHz + uint16_t Gfx_BoosterFreq; // MHz + uint16_t Gfx_PD_Data_time_constant; // Time constant of PD controller in ms + uint32_t Gfx_PD_Data_limit_a; // Q16 + uint32_t Gfx_PD_Data_limit_b; // Q16 + uint32_t Gfx_PD_Data_limit_c; // Q16 + uint32_t Gfx_PD_Data_error_coeff; // Q16 + uint32_t Gfx_PD_Data_error_rate_coeff; // Q16 + + uint8_t Mem_ActiveHystLimit; + uint8_t Mem_IdleHystLimit; + uint8_t Mem_FPS; + uint8_t Mem_MinActiveFreqType; + uint8_t Mem_BoosterFreqType; + uint8_t Mem_MinFreqStep; // Minimum delta between current and target frequeny in order for FW to change clock. + uint8_t Mem_UseRlcBusy; + uint8_t PaddingMem[3]; + uint16_t Mem_MinActiveFreq; // MHz + uint16_t Mem_BoosterFreq; // MHz + uint16_t Mem_PD_Data_time_constant; // Time constant of PD controller in ms + uint32_t Mem_PD_Data_limit_a; // Q16 + uint32_t Mem_PD_Data_limit_b; // Q16 + uint32_t Mem_PD_Data_limit_c; // Q16 + uint32_t Mem_PD_Data_error_coeff; // Q16 + uint32_t Mem_PD_Data_error_rate_coeff; // Q16 + + uint32_t Mem_UpThreshold_Limit; // Q16 + uint8_t Mem_UpHystLimit; + uint8_t Mem_DownHystLimit; + uint16_t Mem_Fps; + + uint32_t MmHubPadding[8]; // SMU internal use +} DpmActivityMonitorCoeffInt_t; +*/ + +// These defines are used with the following messages: +// SMC_MSG_TransferTableDram2Smu +// SMC_MSG_TransferTableSmu2Dram +#define TABLE_PPTABLE 0 +#define TABLE_AVFS 1 +#define TABLE_AVFS_PSM_DEBUG 2 +#define TABLE_AVFS_FUSE_OVERRIDE 3 +#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 + +// These defines are used with the SMC_MSG_SetUclkFastSwitch message. +typedef enum { + DF_SWITCH_TYPE_FAST = 0, + DF_SWITCH_TYPE_SLOW, + DF_SWITCH_TYPE_COUNT, +} DF_SWITCH_TYPE_e; + +typedef enum { + DRAM_BIT_WIDTH_DISABLED = 0, + DRAM_BIT_WIDTH_X_8, + DRAM_BIT_WIDTH_X_16, + DRAM_BIT_WIDTH_X_32, + DRAM_BIT_WIDTH_X_64, // NOT USED. + DRAM_BIT_WIDTH_X_128, + DRAM_BIT_WIDTH_COUNT, +} DRAM_BIT_WIDTH_TYPE_e; + +#define REMOVE_FMAX_MARGIN_BIT 0x0 +#define REMOVE_DCTOL_MARGIN_BIT 0x1 +#define REMOVE_PLATFORM_MARGIN_BIT 0x2 + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu11_driver_if_navi10.h b/drivers/gpu/drm/amd/powerplay/inc/smu11_driver_if_navi10.h new file mode 100644 index 000000000000..ac0120e384be --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/inc/smu11_driver_if_navi10.h @@ -0,0 +1,1086 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __SMU11_DRIVER_IF_NAVI10_H__ +#define __SMU11_DRIVER_IF_NAVI10_H__ + +// *** IMPORTANT *** +// SMU TEAM: Always increment the interface version if +// any structure is changed in this file +// Be aware of that the version should be updated in +// smu_v11_0.h, maybe rename is also needed. +// #define SMU11_DRIVER_IF_VERSION 0x33 + +#define PPTABLE_NV10_SMU_VERSION 8 + +#define NUM_GFXCLK_DPM_LEVELS 16 +#define NUM_SMNCLK_DPM_LEVELS 2 +#define NUM_SOCCLK_DPM_LEVELS 8 +#define NUM_MP0CLK_DPM_LEVELS 2 +#define NUM_DCLK_DPM_LEVELS 8 +#define NUM_VCLK_DPM_LEVELS 8 +#define NUM_DCEFCLK_DPM_LEVELS 8 +#define NUM_PHYCLK_DPM_LEVELS 8 +#define NUM_DISPCLK_DPM_LEVELS 8 +#define NUM_PIXCLK_DPM_LEVELS 8 +#define NUM_UCLK_DPM_LEVELS 4 +#define NUM_MP1CLK_DPM_LEVELS 2 +#define NUM_LINK_LEVELS 2 + + +#define MAX_GFXCLK_DPM_LEVEL (NUM_GFXCLK_DPM_LEVELS - 1) +#define MAX_SMNCLK_DPM_LEVEL (NUM_SMNCLK_DPM_LEVELS - 1) +#define MAX_SOCCLK_DPM_LEVEL (NUM_SOCCLK_DPM_LEVELS - 1) +#define MAX_MP0CLK_DPM_LEVEL (NUM_MP0CLK_DPM_LEVELS - 1) +#define MAX_DCLK_DPM_LEVEL (NUM_DCLK_DPM_LEVELS - 1) +#define MAX_VCLK_DPM_LEVEL (NUM_VCLK_DPM_LEVELS - 1) +#define MAX_DCEFCLK_DPM_LEVEL (NUM_DCEFCLK_DPM_LEVELS - 1) +#define MAX_DISPCLK_DPM_LEVEL (NUM_DISPCLK_DPM_LEVELS - 1) +#define MAX_PIXCLK_DPM_LEVEL (NUM_PIXCLK_DPM_LEVELS - 1) +#define MAX_PHYCLK_DPM_LEVEL (NUM_PHYCLK_DPM_LEVELS - 1) +#define MAX_UCLK_DPM_LEVEL (NUM_UCLK_DPM_LEVELS - 1) +#define MAX_MP1CLK_DPM_LEVEL (NUM_MP1CLK_DPM_LEVELS - 1) +#define MAX_LINK_LEVEL (NUM_LINK_LEVELS - 1) + +//Gemini Modes +#define PPSMC_GeminiModeNone 0 //Single GPU board +#define PPSMC_GeminiModeMaster 1 //Master GPU on a Gemini board +#define PPSMC_GeminiModeSlave 2 //Slave GPU on a Gemini board + +// Feature Control Defines +// DPM +#define FEATURE_DPM_PREFETCHER_BIT 0 +#define FEATURE_DPM_GFXCLK_BIT 1 +#define FEATURE_DPM_GFX_PACE_BIT 2 +#define FEATURE_DPM_UCLK_BIT 3 +#define FEATURE_DPM_SOCCLK_BIT 4 +#define FEATURE_DPM_MP0CLK_BIT 5 +#define FEATURE_DPM_LINK_BIT 6 +#define FEATURE_DPM_DCEFCLK_BIT 7 +#define FEATURE_MEM_VDDCI_SCALING_BIT 8 +#define FEATURE_MEM_MVDD_SCALING_BIT 9 + +//Idle +#define FEATURE_DS_GFXCLK_BIT 10 +#define FEATURE_DS_SOCCLK_BIT 11 +#define FEATURE_DS_LCLK_BIT 12 +#define FEATURE_DS_DCEFCLK_BIT 13 +#define FEATURE_DS_UCLK_BIT 14 +#define FEATURE_GFX_ULV_BIT 15 +#define FEATURE_FW_DSTATE_BIT 16 +#define FEATURE_GFXOFF_BIT 17 +#define FEATURE_BACO_BIT 18 +#define FEATURE_VCN_PG_BIT 19 +#define FEATURE_JPEG_PG_BIT 20 +#define FEATURE_USB_PG_BIT 21 +#define FEATURE_RSMU_SMN_CG_BIT 22 +//Throttler/Response +#define FEATURE_PPT_BIT 23 +#define FEATURE_TDC_BIT 24 +#define FEATURE_GFX_EDC_BIT 25 +#define FEATURE_APCC_PLUS_BIT 26 +#define FEATURE_GTHR_BIT 27 +#define FEATURE_ACDC_BIT 28 +#define FEATURE_VR0HOT_BIT 29 +#define FEATURE_VR1HOT_BIT 30 +#define FEATURE_FW_CTF_BIT 31 +#define FEATURE_FAN_CONTROL_BIT 32 +#define FEATURE_THERMAL_BIT 33 +#define FEATURE_GFX_DCS_BIT 34 +//VF +#define FEATURE_RM_BIT 35 +#define FEATURE_LED_DISPLAY_BIT 36 +//Other +#define FEATURE_GFX_SS_BIT 37 +#define FEATURE_OUT_OF_BAND_MONITOR_BIT 38 +#define FEATURE_TEMP_DEPENDENT_VMIN_BIT 39 + +#define FEATURE_MMHUB_PG_BIT 40 +#define FEATURE_ATHUB_PG_BIT 41 +#define FEATURE_APCC_DFLL_BIT 42 +#define FEATURE_SPARE_43_BIT 43 +#define FEATURE_SPARE_44_BIT 44 +#define FEATURE_SPARE_45_BIT 45 +#define FEATURE_SPARE_46_BIT 46 +#define FEATURE_SPARE_47_BIT 47 +#define FEATURE_SPARE_48_BIT 48 +#define FEATURE_SPARE_49_BIT 49 +#define FEATURE_SPARE_50_BIT 50 +#define FEATURE_SPARE_51_BIT 51 +#define FEATURE_SPARE_52_BIT 52 +#define FEATURE_SPARE_53_BIT 53 +#define FEATURE_SPARE_54_BIT 54 +#define FEATURE_SPARE_55_BIT 55 +#define FEATURE_SPARE_56_BIT 56 +#define FEATURE_SPARE_57_BIT 57 +#define FEATURE_SPARE_58_BIT 58 +#define FEATURE_SPARE_59_BIT 59 +#define FEATURE_SPARE_60_BIT 60 +#define FEATURE_SPARE_61_BIT 61 +#define FEATURE_SPARE_62_BIT 62 +#define FEATURE_SPARE_63_BIT 63 +#define NUM_FEATURES 64 + +// Debug Overrides Bitmask +#define DPM_OVERRIDE_DISABLE_SOCCLK_PID 0x00000001 +#define DPM_OVERRIDE_DISABLE_UCLK_PID 0x00000002 +#define DPM_OVERRIDE_DISABLE_VOLT_LINK_VCN_SOCCLK 0x00000004 +#define DPM_OVERRIDE_ENABLE_FREQ_LINK_VCLK_SOCCLK 0x00000008 +#define DPM_OVERRIDE_ENABLE_FREQ_LINK_DCLK_SOCCLK 0x00000010 +#define DPM_OVERRIDE_ENABLE_FREQ_LINK_GFXCLK_SOCCLK 0x00000020 +#define DPM_OVERRIDE_ENABLE_FREQ_LINK_GFXCLK_UCLK 0x00000040 +#define DPM_OVERRIDE_DISABLE_VOLT_LINK_DCE_SOCCLK 0x00000080 +#define DPM_OVERRIDE_DISABLE_VOLT_LINK_MP0_SOCCLK 0x00000100 +#define DPM_OVERRIDE_DISABLE_DFLL_PLL_SHUTDOWN 0x00000200 +#define DPM_OVERRIDE_DISABLE_MEMORY_TEMPERATURE_READ 0x00000400 + +// VR Mapping Bit Defines +#define VR_MAPPING_VR_SELECT_MASK 0x01 +#define VR_MAPPING_VR_SELECT_SHIFT 0x00 + +#define VR_MAPPING_PLANE_SELECT_MASK 0x02 +#define VR_MAPPING_PLANE_SELECT_SHIFT 0x01 + +// PSI Bit Defines +#define PSI_SEL_VR0_PLANE0_PSI0 0x01 +#define PSI_SEL_VR0_PLANE0_PSI1 0x02 +#define PSI_SEL_VR0_PLANE1_PSI0 0x04 +#define PSI_SEL_VR0_PLANE1_PSI1 0x08 +#define PSI_SEL_VR1_PLANE0_PSI0 0x10 +#define PSI_SEL_VR1_PLANE0_PSI1 0x20 +#define PSI_SEL_VR1_PLANE1_PSI0 0x40 +#define PSI_SEL_VR1_PLANE1_PSI1 0x80 + +// Throttler Control/Status Bits +#define THROTTLER_PADDING_BIT 0 +#define THROTTLER_TEMP_EDGE_BIT 1 +#define THROTTLER_TEMP_HOTSPOT_BIT 2 +#define THROTTLER_TEMP_MEM_BIT 3 +#define THROTTLER_TEMP_VR_GFX_BIT 4 +#define THROTTLER_TEMP_VR_MEM0_BIT 5 +#define THROTTLER_TEMP_VR_MEM1_BIT 6 +#define THROTTLER_TEMP_VR_SOC_BIT 7 +#define THROTTLER_TEMP_LIQUID0_BIT 8 +#define THROTTLER_TEMP_LIQUID1_BIT 9 +#define THROTTLER_TEMP_PLX_BIT 10 +#define THROTTLER_TEMP_SKIN_BIT 11 +#define THROTTLER_TDC_GFX_BIT 12 +#define THROTTLER_TDC_SOC_BIT 13 +#define THROTTLER_PPT0_BIT 14 +#define THROTTLER_PPT1_BIT 15 +#define THROTTLER_PPT2_BIT 16 +#define THROTTLER_PPT3_BIT 17 +#define THROTTLER_FIT_BIT 18 +#define THROTTLER_PPM_BIT 19 +#define THROTTLER_APCC_BIT 20 + +// FW DState Features Control Bits +#define FW_DSTATE_SOC_ULV_BIT 0 +#define FW_DSTATE_G6_HSR_BIT 1 +#define FW_DSTATE_G6_PHY_VDDCI_OFF_BIT 2 +#define FW_DSTATE_MP0_DS_BIT 3 +#define FW_DSTATE_SMN_DS_BIT 4 +#define FW_DSTATE_MP1_DS_BIT 5 +#define FW_DSTATE_MP1_WHISPER_MODE_BIT 6 +#define FW_DSTATE_LIV_MIN_BIT 7 +#define FW_DSTATE_SOC_PLL_PWRDN_BIT 8 + +#define FW_DSTATE_SOC_ULV_MASK (1 << FW_DSTATE_SOC_ULV_BIT ) +#define FW_DSTATE_G6_HSR_MASK (1 << FW_DSTATE_G6_HSR_BIT ) +#define FW_DSTATE_G6_PHY_VDDCI_OFF_MASK (1 << FW_DSTATE_G6_PHY_VDDCI_OFF_BIT ) +#define FW_DSTATE_MP1_DS_MASK (1 << FW_DSTATE_MP1_DS_BIT ) +#define FW_DSTATE_MP0_DS_MASK (1 << FW_DSTATE_MP0_DS_BIT ) +#define FW_DSTATE_SMN_DS_MASK (1 << FW_DSTATE_SMN_DS_BIT ) +#define FW_DSTATE_MP1_WHISPER_MODE_MASK (1 << FW_DSTATE_MP1_WHISPER_MODE_BIT ) +#define FW_DSTATE_LIV_MIN_MASK (1 << FW_DSTATE_LIV_MIN_BIT ) +#define FW_DSTATE_SOC_PLL_PWRDN_MASK (1 << FW_DSTATE_SOC_PLL_PWRDN_BIT ) + +//I2C Interface + +#define NUM_I2C_CONTROLLERS 8 + +#define I2C_CONTROLLER_ENABLED 1 +#define I2C_CONTROLLER_DISABLED 0 + +#define MAX_SW_I2C_COMMANDS 8 + +typedef enum { + I2C_CONTROLLER_PORT_0 = 0, //CKSVII2C0 + I2C_CONTROLLER_PORT_1 = 1, //CKSVII2C1 + I2C_CONTROLLER_PORT_COUNT, +} I2cControllerPort_e; + +typedef enum { + I2C_CONTROLLER_NAME_VR_GFX = 0, + I2C_CONTROLLER_NAME_VR_SOC, + I2C_CONTROLLER_NAME_VR_VDDCI, + I2C_CONTROLLER_NAME_VR_MVDD, + I2C_CONTROLLER_NAME_LIQUID0, + I2C_CONTROLLER_NAME_LIQUID1, + I2C_CONTROLLER_NAME_PLX, + I2C_CONTROLLER_NAME_SPARE, + I2C_CONTROLLER_NAME_COUNT, +} I2cControllerName_e; + +typedef enum { + I2C_CONTROLLER_THROTTLER_TYPE_NONE = 0, + I2C_CONTROLLER_THROTTLER_VR_GFX, + I2C_CONTROLLER_THROTTLER_VR_SOC, + I2C_CONTROLLER_THROTTLER_VR_VDDCI, + I2C_CONTROLLER_THROTTLER_VR_MVDD, + I2C_CONTROLLER_THROTTLER_LIQUID0, + I2C_CONTROLLER_THROTTLER_LIQUID1, + I2C_CONTROLLER_THROTTLER_PLX, + I2C_CONTROLLER_THROTTLER_COUNT, +} I2cControllerThrottler_e; + +typedef enum { + I2C_CONTROLLER_PROTOCOL_VR_0, + I2C_CONTROLLER_PROTOCOL_VR_1, + I2C_CONTROLLER_PROTOCOL_TMP_0, + I2C_CONTROLLER_PROTOCOL_TMP_1, + I2C_CONTROLLER_PROTOCOL_SPARE_0, + I2C_CONTROLLER_PROTOCOL_SPARE_1, + I2C_CONTROLLER_PROTOCOL_COUNT, +} I2cControllerProtocol_e; + +typedef struct { + uint8_t Enabled; + uint8_t Speed; + uint8_t Padding[2]; + uint32_t SlaveAddress; + uint8_t ControllerPort; + uint8_t ControllerName; + uint8_t ThermalThrotter; + uint8_t I2cProtocol; +} I2cControllerConfig_t; + +typedef enum { + I2C_PORT_SVD_SCL = 0, + I2C_PORT_GPIO, +} I2cPort_e; + +typedef enum { + I2C_SPEED_FAST_50K = 0, //50 Kbits/s + I2C_SPEED_FAST_100K, //100 Kbits/s + I2C_SPEED_FAST_400K, //400 Kbits/s + I2C_SPEED_FAST_PLUS_1M, //1 Mbits/s (in fast mode) + I2C_SPEED_HIGH_1M, //1 Mbits/s (in high speed mode) + I2C_SPEED_HIGH_2M, //2.3 Mbits/s + I2C_SPEED_COUNT, +} I2cSpeed_e; + +typedef enum { + I2C_CMD_READ = 0, + I2C_CMD_WRITE, + I2C_CMD_COUNT, +} I2cCmdType_e; + +#define CMDCONFIG_STOP_BIT 0 +#define CMDCONFIG_RESTART_BIT 1 + +#define CMDCONFIG_STOP_MASK (1 << CMDCONFIG_STOP_BIT) +#define CMDCONFIG_RESTART_MASK (1 << CMDCONFIG_RESTART_BIT) + +typedef struct { + uint8_t RegisterAddr; ////only valid for write, ignored for read + uint8_t Cmd; //Read(0) or Write(1) + uint8_t Data; //Return data for read. Data to send for write + uint8_t CmdConfig; //Includes whether associated command should have a stop or restart command +} SwI2cCmd_t; //SW I2C Command Table + +typedef struct { + uint8_t I2CcontrollerPort; //CKSVII2C0(0) or //CKSVII2C1(1) + uint8_t I2CSpeed; //Slow(0) or Fast(1) + uint16_t SlaveAddress; + uint8_t NumCmds; //Number of commands + uint8_t Padding[3]; + + SwI2cCmd_t SwI2cCmds[MAX_SW_I2C_COMMANDS]; + + uint32_t MmHubPadding[8]; // SMU internal use + +} SwI2cRequest_t; // SW I2C Request Table + +//D3HOT sequences +typedef enum { + BACO_SEQUENCE, + MSR_SEQUENCE, + BAMACO_SEQUENCE, + ULPS_SEQUENCE, + D3HOT_SEQUENCE_COUNT, +}D3HOTSequence_e; + +//THis is aligned with RSMU PGFSM Register Mapping +typedef enum { + PG_DYNAMIC_MODE = 0, + PG_STATIC_MODE, +} PowerGatingMode_e; + +//This is aligned with RSMU PGFSM Register Mapping +typedef enum { + PG_POWER_DOWN = 0, + PG_POWER_UP, +} PowerGatingSettings_e; + +typedef struct { + uint32_t a; // store in IEEE float format in this variable + uint32_t b; // store in IEEE float format in this variable + uint32_t c; // store in IEEE float format in this variable +} QuadraticInt_t; + +typedef struct { + uint32_t m; // store in IEEE float format in this variable + uint32_t b; // store in IEEE float format in this variable +} LinearInt_t; + +typedef struct { + uint32_t a; // store in IEEE float format in this variable + uint32_t b; // store in IEEE float format in this variable + uint32_t c; // store in IEEE float format in this variable +} DroopInt_t; + +typedef enum { + GFXCLK_SOURCE_PLL = 0, + GFXCLK_SOURCE_DFLL, + GFXCLK_SOURCE_COUNT, +} GfxclkSrc_e; + +//Only Clks that have DPM descriptors are listed here +typedef enum { + PPCLK_GFXCLK = 0, + PPCLK_SOCCLK, + PPCLK_UCLK, + PPCLK_DCLK, + PPCLK_VCLK, + PPCLK_DCEFCLK, + PPCLK_DISPCLK, + PPCLK_PIXCLK, + PPCLK_PHYCLK, + PPCLK_COUNT, +} PPCLK_e; + +typedef enum { + POWER_SOURCE_AC, + POWER_SOURCE_DC, + POWER_SOURCE_COUNT, +} POWER_SOURCE_e; + +typedef enum { + PPT_THROTTLER_PPT0, + PPT_THROTTLER_PPT1, + PPT_THROTTLER_PPT2, + PPT_THROTTLER_PPT3, + PPT_THROTTLER_COUNT +} PPT_THROTTLER_e; + +typedef enum { + VOLTAGE_MODE_AVFS = 0, + VOLTAGE_MODE_AVFS_SS, + VOLTAGE_MODE_SS, + VOLTAGE_MODE_COUNT, +} VOLTAGE_MODE_e; + + +typedef enum { + AVFS_VOLTAGE_GFX = 0, + AVFS_VOLTAGE_SOC, + AVFS_VOLTAGE_COUNT, +} AVFS_VOLTAGE_TYPE_e; + +typedef enum { + UCLK_DIV_BY_1 = 0, + UCLK_DIV_BY_2, + UCLK_DIV_BY_4, + UCLK_DIV_BY_8, +} UCLK_DIV_e; + +typedef enum { + GPIO_INT_POLARITY_ACTIVE_LOW = 0, + GPIO_INT_POLARITY_ACTIVE_HIGH, +} GpioIntPolarity_e; + +typedef enum { + MEMORY_TYPE_GDDR6 = 0, + MEMORY_TYPE_HBM, +} MemoryType_e; + +typedef enum { + PWR_CONFIG_TDP = 0, + PWR_CONFIG_TGP, + PWR_CONFIG_TCP_ESTIMATED, + PWR_CONFIG_TCP_MEASURED, +} PwrConfig_e; + +typedef struct { + uint8_t VoltageMode; // 0 - AVFS only, 1- min(AVFS,SS), 2-SS only + uint8_t SnapToDiscrete; // 0 - Fine grained DPM, 1 - Discrete DPM + uint8_t NumDiscreteLevels; // Set to 2 (Fmin, Fmax) when using fine grained DPM, otherwise set to # discrete levels used + uint8_t Padding; + LinearInt_t ConversionToAvfsClk; // Transfer function to AVFS Clock (GHz->GHz) + QuadraticInt_t SsCurve; // Slow-slow curve (GHz->V) +} DpmDescriptor_t; + +typedef enum { + TEMP_EDGE, + TEMP_HOTSPOT, + TEMP_MEM, + TEMP_VR_GFX, + TEMP_VR_MEM0, + TEMP_VR_MEM1, + TEMP_VR_SOC, + TEMP_LIQUID0, + TEMP_LIQUID1, + TEMP_PLX, + TEMP_COUNT +} TEMP_e; + +//Out of band monitor status defines +//see SPEC //gpu/doc/soc_arch/spec/feature/SMBUS/SMBUS.xlsx +#define POWER_MANAGER_CONTROLLER_NOT_RUNNING 0 +#define POWER_MANAGER_CONTROLLER_RUNNING 1 + +#define POWER_MANAGER_CONTROLLER_BIT 0 +#define MAXIMUM_DPM_STATE_GFX_ENGINE_RESTRICTED_BIT 8 +#define GPU_DIE_TEMPERATURE_THROTTLING_BIT 9 +#define HBM_DIE_TEMPERATURE_THROTTLING_BIT 10 +#define TGP_THROTTLING_BIT 11 +#define PCC_THROTTLING_BIT 12 +#define HBM_TEMPERATURE_EXCEEDING_TEMPERATURE_LIMIT_BIT 13 +#define HBM_TEMPERATURE_EXCEEDING_MAX_MEMORY_TEMPERATURE_BIT 14 + +#define POWER_MANAGER_CONTROLLER_MASK (1 << POWER_MANAGER_CONTROLLER_BIT ) +#define MAXIMUM_DPM_STATE_GFX_ENGINE_RESTRICTED_MASK (1 << MAXIMUM_DPM_STATE_GFX_ENGINE_RESTRICTED_BIT ) +#define GPU_DIE_TEMPERATURE_THROTTLING_MASK (1 << GPU_DIE_TEMPERATURE_THROTTLING_BIT ) +#define HBM_DIE_TEMPERATURE_THROTTLING_MASK (1 << HBM_DIE_TEMPERATURE_THROTTLING_BIT ) +#define TGP_THROTTLING_MASK (1 << TGP_THROTTLING_BIT ) +#define PCC_THROTTLING_MASK (1 << PCC_THROTTLING_BIT ) +#define HBM_TEMPERATURE_EXCEEDING_TEMPERATURE_LIMIT_MASK (1 << HBM_TEMPERATURE_EXCEEDING_TEMPERATURE_LIMIT_BIT ) +#define HBM_TEMPERATURE_EXCEEDING_MAX_MEMORY_TEMPERATURE_MASK (1 << HBM_TEMPERATURE_EXCEEDING_MAX_MEMORY_TEMPERATURE_BIT) + +//This structure to be DMA to SMBUS Config register space +typedef struct { + uint8_t MinorInfoVersion; + uint8_t MajorInfoVersion; + uint8_t TableSize; + uint8_t Reserved; + + uint8_t Reserved1; + uint8_t RevID; + uint16_t DeviceID; + + uint16_t DieTemperatureLimit; + uint16_t FanTargetTemperature; + + uint16_t MemoryTemperatureLimit; + uint16_t MemoryTemperatureLimit1; + + uint16_t TGP; + uint16_t CardPower; + + uint32_t DieTemperatureRegisterOffset; + + uint32_t Reserved2; + + uint32_t Reserved3; + + uint32_t Status; + + uint16_t DieTemperature; + uint16_t CurrentMemoryTemperature; + + uint16_t MemoryTemperature; + uint8_t MemoryHotspotPosition; + uint8_t Reserved4; + + uint32_t BoardLevelEnergyAccumulator; +} OutOfBandMonitor_t; + +typedef struct { + uint32_t Version; + + // SECTION: Feature Enablement + uint32_t FeaturesToRun[2]; + + // SECTION: Infrastructure Limits + uint16_t SocketPowerLimitAc[PPT_THROTTLER_COUNT]; + uint16_t SocketPowerLimitAcTau[PPT_THROTTLER_COUNT]; + uint16_t SocketPowerLimitDc[PPT_THROTTLER_COUNT]; + uint16_t SocketPowerLimitDcTau[PPT_THROTTLER_COUNT]; + + uint16_t TdcLimitSoc; // Amps + uint16_t TdcLimitSocTau; // Time constant of LPF in ms + uint16_t TdcLimitGfx; // Amps + uint16_t TdcLimitGfxTau; // Time constant of LPF in ms + + uint16_t TedgeLimit; // Celcius + uint16_t ThotspotLimit; // Celcius + uint16_t TmemLimit; // Celcius + uint16_t Tvr_gfxLimit; // Celcius + uint16_t Tvr_mem0Limit; // Celcius + uint16_t Tvr_mem1Limit; // Celcius + uint16_t Tvr_socLimit; // Celcius + uint16_t Tliquid0Limit; // Celcius + uint16_t Tliquid1Limit; // Celcius + uint16_t TplxLimit; // Celcius + uint32_t FitLimit; // Failures in time (failures per million parts over the defined lifetime) + + uint16_t PpmPowerLimit; // Switch this this power limit when temperature is above PpmTempThreshold + uint16_t PpmTemperatureThreshold; + + // SECTION: Throttler settings + uint32_t ThrottlerControlMask; // See Throtter masks defines + + // SECTION: FW DSTATE Settings + uint32_t FwDStateMask; // See FW DState masks defines + + // SECTION: ULV Settings + uint16_t UlvVoltageOffsetSoc; // In mV(Q2) + uint16_t UlvVoltageOffsetGfx; // In mV(Q2) + + uint8_t GceaLinkMgrIdleThreshold; //Set by SMU FW during enablment of SOC_ULV. Controls delay for GFX SDP port disconnection during idle events + uint8_t paddingRlcUlvParams[3]; + + uint8_t UlvSmnclkDid; //DID for ULV mode. 0 means CLK will not be modified in ULV. + uint8_t UlvMp1clkDid; //DID for ULV mode. 0 means CLK will not be modified in ULV. + uint8_t UlvGfxclkBypass; // 1 to turn off/bypass Gfxclk during ULV, 0 to leave Gfxclk on during ULV + uint8_t Padding234; + + uint16_t MinVoltageUlvGfx; // In mV(Q2) Minimum Voltage ("Vmin") of VDD_GFX in ULV mode + uint16_t MinVoltageUlvSoc; // In mV(Q2) Minimum Voltage ("Vmin") of VDD_SOC in ULV mode + + + // SECTION: Voltage Control Parameters + uint16_t MinVoltageGfx; // In mV(Q2) Minimum Voltage ("Vmin") of VDD_GFX + uint16_t MinVoltageSoc; // In mV(Q2) Minimum Voltage ("Vmin") of VDD_SOC + uint16_t MaxVoltageGfx; // In mV(Q2) Maximum Voltage allowable of VDD_GFX + uint16_t MaxVoltageSoc; // In mV(Q2) Maximum Voltage allowable of VDD_SOC + + uint16_t LoadLineResistanceGfx; // In mOhms with 8 fractional bits + uint16_t LoadLineResistanceSoc; // In mOhms with 8 fractional bits + + //SECTION: DPM Config 1 + DpmDescriptor_t DpmDescriptor[PPCLK_COUNT]; + + uint16_t FreqTableGfx [NUM_GFXCLK_DPM_LEVELS ]; // In MHz + uint16_t FreqTableVclk [NUM_VCLK_DPM_LEVELS ]; // In MHz + uint16_t FreqTableDclk [NUM_DCLK_DPM_LEVELS ]; // In MHz + uint16_t FreqTableSocclk [NUM_SOCCLK_DPM_LEVELS ]; // In MHz + uint16_t FreqTableUclk [NUM_UCLK_DPM_LEVELS ]; // In MHz + uint16_t FreqTableDcefclk [NUM_DCEFCLK_DPM_LEVELS ]; // In MHz + uint16_t FreqTableDispclk [NUM_DISPCLK_DPM_LEVELS ]; // In MHz + uint16_t FreqTablePixclk [NUM_PIXCLK_DPM_LEVELS ]; // In MHz + uint16_t FreqTablePhyclk [NUM_PHYCLK_DPM_LEVELS ]; // In MHz + uint32_t Paddingclks[16]; + + uint16_t DcModeMaxFreq [PPCLK_COUNT ]; // In MHz + uint16_t Padding8_Clks; + + uint8_t FreqTableUclkDiv [NUM_UCLK_DPM_LEVELS ]; // 0:Div-1, 1:Div-1/2, 2:Div-1/4, 3:Div-1/8 + + // SECTION: DPM Config 2 + uint16_t Mp0clkFreq [NUM_MP0CLK_DPM_LEVELS]; // in MHz + uint16_t Mp0DpmVoltage [NUM_MP0CLK_DPM_LEVELS]; // mV(Q2) + uint16_t MemVddciVoltage [NUM_UCLK_DPM_LEVELS]; // mV(Q2) + uint16_t MemMvddVoltage [NUM_UCLK_DPM_LEVELS]; // mV(Q2) + // GFXCLK DPM + uint16_t GfxclkFgfxoffEntry; // in Mhz + uint16_t GfxclkFinit; // in Mhz + uint16_t GfxclkFidle; // in MHz + uint16_t GfxclkSlewRate; // for PLL babystepping??? + uint16_t GfxclkFopt; // in Mhz + uint8_t Padding567[2]; + uint16_t GfxclkDsMaxFreq; // in MHz + uint8_t GfxclkSource; // 0 = PLL, 1 = DFLL + uint8_t Padding456; + + // UCLK section + uint8_t LowestUclkReservedForUlv; // Set this to 1 if UCLK DPM0 is reserved for ULV-mode only + uint8_t paddingUclk[3]; + + uint8_t MemoryType; // 0-GDDR6, 1-HBM + uint8_t MemoryChannels; + uint8_t PaddingMem[2]; + + // Link DPM Settings + uint8_t PcieGenSpeed[NUM_LINK_LEVELS]; ///< 0:PciE-gen1 1:PciE-gen2 2:PciE-gen3 3:PciE-gen4 + uint8_t PcieLaneCount[NUM_LINK_LEVELS]; ///< 1=x1, 2=x2, 3=x4, 4=x8, 5=x12, 6=x16 + uint16_t LclkFreq[NUM_LINK_LEVELS]; + + // GFXCLK Thermal DPM (formerly 'Boost' Settings) + uint16_t EnableTdpm; + uint16_t TdpmHighHystTemperature; + uint16_t TdpmLowHystTemperature; + uint16_t GfxclkFreqHighTempLimit; // High limit on GFXCLK when temperature is high, for reliability. + + // SECTION: Fan Control + uint16_t FanStopTemp; //Celcius + uint16_t FanStartTemp; //Celcius + + uint16_t FanGainEdge; + uint16_t FanGainHotspot; + uint16_t FanGainLiquid0; + uint16_t FanGainLiquid1; + uint16_t FanGainVrGfx; + uint16_t FanGainVrSoc; + uint16_t FanGainVrMem0; + uint16_t FanGainVrMem1; + uint16_t FanGainPlx; + uint16_t FanGainMem; + uint16_t FanPwmMin; + uint16_t FanAcousticLimitRpm; + uint16_t FanThrottlingRpm; + uint16_t FanMaximumRpm; + uint16_t FanTargetTemperature; + uint16_t FanTargetGfxclk; + uint8_t FanTempInputSelect; + uint8_t FanPadding; + uint8_t FanZeroRpmEnable; + uint8_t FanTachEdgePerRev; + //uint8_t padding8_Fan[2]; + + // The following are AFC override parameters. Leave at 0 to use FW defaults. + int16_t FuzzyFan_ErrorSetDelta; + int16_t FuzzyFan_ErrorRateSetDelta; + int16_t FuzzyFan_PwmSetDelta; + uint16_t FuzzyFan_Reserved; + + + // SECTION: AVFS + // Overrides + uint8_t OverrideAvfsGb[AVFS_VOLTAGE_COUNT]; + uint8_t Padding8_Avfs[2]; + + QuadraticInt_t qAvfsGb[AVFS_VOLTAGE_COUNT]; // GHz->V Override of fused curve + DroopInt_t dBtcGbGfxPll; // GHz->V BtcGb + DroopInt_t dBtcGbGfxDfll; // GHz->V BtcGb + DroopInt_t dBtcGbSoc; // GHz->V BtcGb + LinearInt_t qAgingGb[AVFS_VOLTAGE_COUNT]; // GHz->V + + QuadraticInt_t qStaticVoltageOffset[AVFS_VOLTAGE_COUNT]; // GHz->V + + uint16_t DcTol[AVFS_VOLTAGE_COUNT]; // mV Q2 + + uint8_t DcBtcEnabled[AVFS_VOLTAGE_COUNT]; + uint8_t Padding8_GfxBtc[2]; + + uint16_t DcBtcMin[AVFS_VOLTAGE_COUNT]; // mV Q2 + uint16_t DcBtcMax[AVFS_VOLTAGE_COUNT]; // mV Q2 + + // SECTION: Advanced Options + uint32_t DebugOverrides; + QuadraticInt_t ReservedEquation0; + QuadraticInt_t ReservedEquation1; + QuadraticInt_t ReservedEquation2; + QuadraticInt_t ReservedEquation3; + + // Total Power configuration, use defines from PwrConfig_e + uint8_t TotalPowerConfig; //0-TDP, 1-TGP, 2-TCP Estimated, 3-TCP Measured + uint8_t TotalPowerSpare1; + uint16_t TotalPowerSpare2; + + // APCC Settings + uint16_t PccThresholdLow; + uint16_t PccThresholdHigh; + uint32_t PaddingAPCC[6]; //FIXME pending SPEC + + // Temperature Dependent Vmin + uint16_t VDDGFX_TVmin; //Celcius + uint16_t VDDSOC_TVmin; //Celcius + uint16_t VDDGFX_Vmin_HiTemp; // mV Q2 + uint16_t VDDGFX_Vmin_LoTemp; // mV Q2 + uint16_t VDDSOC_Vmin_HiTemp; // mV Q2 + uint16_t VDDSOC_Vmin_LoTemp; // mV Q2 + + uint16_t VDDGFX_TVminHystersis; // Celcius + uint16_t VDDSOC_TVminHystersis; // Celcius + + // BTC Setting + uint32_t BtcConfig; + + uint16_t SsFmin[10]; // PPtable value to function similar to VFTFmin for SS Curve; Size is PPCLK_COUNT rounded to nearest multiple of 2 + uint16_t DcBtcGb[AVFS_VOLTAGE_COUNT]; + + // SECTION: Board Reserved + uint32_t Reserved[8]; + + // SECTION: BOARD PARAMETERS + // I2C Control + I2cControllerConfig_t I2cControllers[NUM_I2C_CONTROLLERS]; + + // SVI2 Board Parameters + uint16_t MaxVoltageStepGfx; // In mV(Q2) Max voltage step that SMU will request. Multiple steps are taken if voltage change exceeds this value. + uint16_t MaxVoltageStepSoc; // In mV(Q2) Max voltage step that SMU will request. Multiple steps are taken if voltage change exceeds this value. + + uint8_t VddGfxVrMapping; // Use VR_MAPPING* bitfields + uint8_t VddSocVrMapping; // Use VR_MAPPING* bitfields + uint8_t VddMem0VrMapping; // Use VR_MAPPING* bitfields + uint8_t VddMem1VrMapping; // Use VR_MAPPING* bitfields + + uint8_t GfxUlvPhaseSheddingMask; // set this to 1 to set PSI0/1 to 1 in ULV mode + uint8_t SocUlvPhaseSheddingMask; // set this to 1 to set PSI0/1 to 1 in ULV mode + uint8_t ExternalSensorPresent; // External RDI connected to TMON (aka TEMP IN) + uint8_t Padding8_V; + + // Telemetry Settings + uint16_t GfxMaxCurrent; // in Amps + int8_t GfxOffset; // in Amps + uint8_t Padding_TelemetryGfx; + + uint16_t SocMaxCurrent; // in Amps + int8_t SocOffset; // in Amps + uint8_t Padding_TelemetrySoc; + + uint16_t Mem0MaxCurrent; // in Amps + int8_t Mem0Offset; // in Amps + uint8_t Padding_TelemetryMem0; + + uint16_t Mem1MaxCurrent; // in Amps + int8_t Mem1Offset; // in Amps + uint8_t Padding_TelemetryMem1; + + // GPIO Settings + uint8_t AcDcGpio; // GPIO pin configured for AC/DC switching + uint8_t AcDcPolarity; // GPIO polarity for AC/DC switching + uint8_t VR0HotGpio; // GPIO pin configured for VR0 HOT event + uint8_t VR0HotPolarity; // GPIO polarity for VR0 HOT event + + uint8_t VR1HotGpio; // GPIO pin configured for VR1 HOT event + uint8_t VR1HotPolarity; // GPIO polarity for VR1 HOT event + uint8_t GthrGpio; // GPIO pin configured for GTHR Event + uint8_t GthrPolarity; // replace GPIO polarity for GTHR + + // LED Display Settings + uint8_t LedPin0; // GPIO number for LedPin[0] + uint8_t LedPin1; // GPIO number for LedPin[1] + uint8_t LedPin2; // GPIO number for LedPin[2] + uint8_t padding8_4; + + // GFXCLK PLL Spread Spectrum + uint8_t PllGfxclkSpreadEnabled; // on or off + uint8_t PllGfxclkSpreadPercent; // Q4.4 + uint16_t PllGfxclkSpreadFreq; // kHz + + // GFXCLK DFLL Spread Spectrum + uint8_t DfllGfxclkSpreadEnabled; // on or off + uint8_t DfllGfxclkSpreadPercent; // Q4.4 + uint16_t DfllGfxclkSpreadFreq; // kHz + + // UCLK Spread Spectrum + uint8_t UclkSpreadEnabled; // on or off + uint8_t UclkSpreadPercent; // Q4.4 + uint16_t UclkSpreadFreq; // kHz + + // SOCCLK Spread Spectrum + uint8_t SoclkSpreadEnabled; // on or off + uint8_t SocclkSpreadPercent; // Q4.4 + uint16_t SocclkSpreadFreq; // kHz + + // Total board power + uint16_t TotalBoardPower; //Only needed for TCP Estimated case, where TCP = TGP+Total Board Power + uint16_t BoardPadding; + + // Mvdd Svi2 Div Ratio Setting + uint32_t MvddRatio; // This is used for MVDD Vid workaround. It has 16 fractional bits (Q16.16) + + uint8_t RenesesLoadLineEnabled; + uint8_t GfxLoadlineResistance; + uint8_t SocLoadlineResistance; + uint8_t Padding8_Loadline; + + uint32_t BoardReserved[8]; + + // Padding for MMHUB - do not modify this + uint32_t MmHubPadding[8]; // SMU internal use + +} PPTable_t; + +typedef struct { + // Time constant parameters for clock averages in ms + uint16_t GfxclkAverageLpfTau; + uint16_t SocclkAverageLpfTau; + uint16_t UclkAverageLpfTau; + uint16_t GfxActivityLpfTau; + uint16_t UclkActivityLpfTau; + uint16_t SocketPowerLpfTau; + + // Padding - ignore + uint32_t MmHubPadding[8]; // SMU internal use +} DriverSmuConfig_t; + +typedef struct { + + uint16_t GfxclkFmin; // MHz + uint16_t GfxclkFmax; // MHz + uint16_t GfxclkFreq1; // MHz + uint16_t GfxclkVolt1; // mV (Q2) + uint16_t GfxclkFreq2; // MHz + uint16_t GfxclkVolt2; // mV (Q2) + uint16_t GfxclkFreq3; // MHz + uint16_t GfxclkVolt3; // mV (Q2) + uint16_t UclkFmax; // MHz + int16_t OverDrivePct; // % + uint16_t FanMaximumRpm; + uint16_t FanMinimumPwm; + uint16_t FanTargetTemperature; // Degree Celcius + uint16_t MaxOpTemp; // Degree Celcius + uint16_t FanZeroRpmEnable; + uint16_t Padding; + + uint32_t MmHubPadding[8]; // SMU internal use + +} OverDriveTable_t; + +typedef struct { + uint16_t CurrClock[PPCLK_COUNT]; + uint16_t AverageGfxclkFrequency; + uint16_t AverageSocclkFrequency; + uint16_t AverageUclkFrequency ; + uint16_t AverageGfxActivity ; + uint16_t AverageUclkActivity ; + uint8_t CurrSocVoltageOffset ; + uint8_t CurrGfxVoltageOffset ; + uint8_t CurrMemVidOffset ; + uint8_t Padding8 ; + uint16_t AverageSocketPower ; + uint16_t TemperatureEdge ; + uint16_t TemperatureHotspot ; + uint16_t TemperatureMem ; + uint16_t TemperatureVrGfx ; + uint16_t TemperatureVrMem0 ; + uint16_t TemperatureVrMem1 ; + uint16_t TemperatureVrSoc ; + uint16_t TemperatureLiquid0 ; + uint16_t TemperatureLiquid1 ; + uint16_t TemperaturePlx ; + uint16_t Padding16 ; + uint32_t ThrottlerStatus ; + + uint8_t LinkDpmLevel; + uint8_t Padding8_2; + uint16_t CurrFanSpeed; + + // Padding - ignore + uint32_t MmHubPadding[8]; // SMU internal use +} SmuMetrics_t; + +typedef struct { + uint16_t MinClock; // This is either DCEFCLK or SOCCLK (in MHz) + uint16_t MaxClock; // This is either DCEFCLK or SOCCLK (in MHz) + uint16_t MinUclk; + uint16_t MaxUclk; + + uint8_t WmSetting; + uint8_t Padding[3]; + + uint32_t MmHubPadding[8]; // SMU internal use +} WatermarkRowGeneric_t; + +#define NUM_WM_RANGES 4 + +typedef enum { + WM_SOCCLK = 0, + WM_DCEFCLK, + WM_COUNT, +} WM_CLOCK_e; + +typedef struct { + // Watermarks + WatermarkRowGeneric_t WatermarkRow[WM_COUNT][NUM_WM_RANGES]; + + uint32_t MmHubPadding[8]; // SMU internal use +} Watermarks_t; + +typedef struct { + uint16_t avgPsmCount[28]; + uint16_t minPsmCount[28]; + float avgPsmVoltage[28]; + float minPsmVoltage[28]; + + uint32_t MmHubPadding[32]; // SMU internal use +} AvfsDebugTable_t_NV14; + +typedef struct { + uint16_t avgPsmCount[36]; + uint16_t minPsmCount[36]; + float avgPsmVoltage[36]; + float minPsmVoltage[36]; + + uint32_t MmHubPadding[8]; // SMU internal use +} AvfsDebugTable_t_NV10; + +typedef struct { + uint8_t AvfsVersion; + uint8_t Padding; + + uint8_t AvfsEn[AVFS_VOLTAGE_COUNT]; + + uint8_t OverrideVFT[AVFS_VOLTAGE_COUNT]; + uint8_t OverrideAvfsGb[AVFS_VOLTAGE_COUNT]; + + uint8_t OverrideTemperatures[AVFS_VOLTAGE_COUNT]; + uint8_t OverrideVInversion[AVFS_VOLTAGE_COUNT]; + uint8_t OverrideP2V[AVFS_VOLTAGE_COUNT]; + uint8_t OverrideP2VCharzFreq[AVFS_VOLTAGE_COUNT]; + + int32_t VFT0_m1[AVFS_VOLTAGE_COUNT]; // Q8.24 + int32_t VFT0_m2[AVFS_VOLTAGE_COUNT]; // Q12.12 + int32_t VFT0_b[AVFS_VOLTAGE_COUNT]; // Q32 + + int32_t VFT1_m1[AVFS_VOLTAGE_COUNT]; // Q8.16 + int32_t VFT1_m2[AVFS_VOLTAGE_COUNT]; // Q12.12 + int32_t VFT1_b[AVFS_VOLTAGE_COUNT]; // Q32 + + int32_t VFT2_m1[AVFS_VOLTAGE_COUNT]; // Q8.16 + int32_t VFT2_m2[AVFS_VOLTAGE_COUNT]; // Q12.12 + int32_t VFT2_b[AVFS_VOLTAGE_COUNT]; // Q32 + + int32_t AvfsGb0_m1[AVFS_VOLTAGE_COUNT]; // Q8.24 + int32_t AvfsGb0_m2[AVFS_VOLTAGE_COUNT]; // Q12.12 + int32_t AvfsGb0_b[AVFS_VOLTAGE_COUNT]; // Q32 + + int32_t AcBtcGb_m1[AVFS_VOLTAGE_COUNT]; // Q8.24 + int32_t AcBtcGb_m2[AVFS_VOLTAGE_COUNT]; // Q12.12 + int32_t AcBtcGb_b[AVFS_VOLTAGE_COUNT]; // Q32 + + uint32_t AvfsTempCold[AVFS_VOLTAGE_COUNT]; + uint32_t AvfsTempMid[AVFS_VOLTAGE_COUNT]; + uint32_t AvfsTempHot[AVFS_VOLTAGE_COUNT]; + + uint32_t VInversion[AVFS_VOLTAGE_COUNT]; // in mV with 2 fractional bits + + + int32_t P2V_m1[AVFS_VOLTAGE_COUNT]; // Q8.24 + int32_t P2V_m2[AVFS_VOLTAGE_COUNT]; // Q12.12 + int32_t P2V_b[AVFS_VOLTAGE_COUNT]; // Q32 + + uint32_t P2VCharzFreq[AVFS_VOLTAGE_COUNT]; // in 10KHz units + + uint32_t EnabledAvfsModules[2]; //NV10 - 36 AVFS modules + + uint32_t MmHubPadding[8]; // SMU internal use +} AvfsFuseOverride_t; + +typedef struct { + + uint8_t Gfx_ActiveHystLimit; + uint8_t Gfx_IdleHystLimit; + uint8_t Gfx_FPS; + uint8_t Gfx_MinActiveFreqType; + uint8_t Gfx_BoosterFreqType; + uint8_t Gfx_MinFreqStep; // Minimum delta between current and target frequeny in order for FW to change clock. + uint16_t Gfx_MinActiveFreq; // MHz + uint16_t Gfx_BoosterFreq; // MHz + uint16_t Gfx_PD_Data_time_constant; // Time constant of PD controller in ms + uint32_t Gfx_PD_Data_limit_a; // Q16 + uint32_t Gfx_PD_Data_limit_b; // Q16 + uint32_t Gfx_PD_Data_limit_c; // Q16 + uint32_t Gfx_PD_Data_error_coeff; // Q16 + uint32_t Gfx_PD_Data_error_rate_coeff; // Q16 + + uint8_t Soc_ActiveHystLimit; + uint8_t Soc_IdleHystLimit; + uint8_t Soc_FPS; + uint8_t Soc_MinActiveFreqType; + uint8_t Soc_BoosterFreqType; + uint8_t Soc_MinFreqStep; // Minimum delta between current and target frequeny in order for FW to change clock. + uint16_t Soc_MinActiveFreq; // MHz + uint16_t Soc_BoosterFreq; // MHz + uint16_t Soc_PD_Data_time_constant; // Time constant of PD controller in ms + uint32_t Soc_PD_Data_limit_a; // Q16 + uint32_t Soc_PD_Data_limit_b; // Q16 + uint32_t Soc_PD_Data_limit_c; // Q16 + uint32_t Soc_PD_Data_error_coeff; // Q16 + uint32_t Soc_PD_Data_error_rate_coeff; // Q16 + + uint8_t Mem_ActiveHystLimit; + uint8_t Mem_IdleHystLimit; + uint8_t Mem_FPS; + uint8_t Mem_MinActiveFreqType; + uint8_t Mem_BoosterFreqType; + uint8_t Mem_MinFreqStep; // Minimum delta between current and target frequeny in order for FW to change clock. + uint16_t Mem_MinActiveFreq; // MHz + uint16_t Mem_BoosterFreq; // MHz + uint16_t Mem_PD_Data_time_constant; // Time constant of PD controller in ms + uint32_t Mem_PD_Data_limit_a; // Q16 + uint32_t Mem_PD_Data_limit_b; // Q16 + uint32_t Mem_PD_Data_limit_c; // Q16 + uint32_t Mem_PD_Data_error_coeff; // Q16 + uint32_t Mem_PD_Data_error_rate_coeff; // Q16 + + uint32_t Mem_UpThreshold_Limit; // Q16 + uint8_t Mem_UpHystLimit; + uint8_t Mem_DownHystLimit; + uint16_t Mem_Fps; + + uint32_t MmHubPadding[8]; // SMU internal use + +} DpmActivityMonitorCoeffInt_t; + + +// Workload bits +#define WORKLOAD_PPLIB_DEFAULT_BIT 0 +#define WORKLOAD_PPLIB_FULL_SCREEN_3D_BIT 1 +#define WORKLOAD_PPLIB_POWER_SAVING_BIT 2 +#define WORKLOAD_PPLIB_VIDEO_BIT 3 +#define WORKLOAD_PPLIB_VR_BIT 4 +#define WORKLOAD_PPLIB_COMPUTE_BIT 5 +#define WORKLOAD_PPLIB_CUSTOM_BIT 6 +#define WORKLOAD_PPLIB_COUNT 7 + + +// These defines are used with the following messages: +// SMC_MSG_TransferTableDram2Smu +// SMC_MSG_TransferTableSmu2Dram + +// Table transfer status +#define TABLE_TRANSFER_OK 0x0 +#define TABLE_TRANSFER_FAILED 0xFF + +// Table types +#define TABLE_PPTABLE 0 +#define TABLE_WATERMARKS 1 +#define TABLE_AVFS 2 +#define TABLE_AVFS_PSM_DEBUG 3 +#define TABLE_AVFS_FUSE_OVERRIDE 4 +#define TABLE_PMSTATUSLOG 5 +#define TABLE_SMU_METRICS 6 +#define TABLE_DRIVER_SMU_CONFIG 7 +#define TABLE_ACTIVITY_MONITOR_COEFF 8 +#define TABLE_OVERDRIVE 9 +#define TABLE_I2C_COMMANDS 10 +#define TABLE_PACE 11 +#define TABLE_COUNT 12 + +//RLC Pace Table total number of levels +#define RLC_PACE_TABLE_NUM_LEVELS 16 + +typedef struct { + float FlopsPerByteTable[RLC_PACE_TABLE_NUM_LEVELS]; + + uint32_t MmHubPadding[8]; // SMU internal use +} RlcPaceFlopsPerByteOverride_t; + +// These defines are used with the SMC_MSG_SetUclkFastSwitch message. +#define UCLK_SWITCH_SLOW 0 +#define UCLK_SWITCH_FAST 1 +#endif diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu12_driver_if.h b/drivers/gpu/drm/amd/powerplay/inc/smu12_driver_if.h new file mode 100644 index 000000000000..c27c82851468 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/inc/smu12_driver_if.h @@ -0,0 +1,217 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef SMU12_DRIVER_IF_H +#define SMU12_DRIVER_IF_H + +// *** IMPORTANT *** +// SMU TEAM: Always increment the interface version if +// any structure is changed in this file +#define SMU12_DRIVER_IF_VERSION 10 + +typedef struct { + int32_t value; + uint32_t numFractionalBits; +} FloatInIntFormat_t; + +typedef enum { + DSPCLK_DCFCLK = 0, + DSPCLK_DISPCLK, + DSPCLK_PIXCLK, + DSPCLK_PHYCLK, + DSPCLK_COUNT, +} DSPCLK_e; + +typedef struct { + uint16_t Freq; // in MHz + uint16_t Vid; // min voltage in SVI2 VID +} DisplayClockTable_t; + +typedef struct { + uint16_t MinClock; // This is either DCFCLK or SOCCLK (in MHz) + uint16_t MaxClock; // This is either DCFCLK or SOCCLK (in MHz) + uint16_t MinMclk; + uint16_t MaxMclk; + + uint8_t WmSetting; + uint8_t WmType; // Used for normal pstate change or memory retraining + uint8_t Padding[2]; +} WatermarkRowGeneric_t; + +#define NUM_WM_RANGES 4 +#define WM_PSTATE_CHG 0 +#define WM_RETRAINING 1 + +typedef enum { + WM_SOCCLK = 0, + WM_DCFCLK, + WM_COUNT, +} WM_CLOCK_e; + +typedef struct { + // Watermarks + WatermarkRowGeneric_t WatermarkRow[WM_COUNT][NUM_WM_RANGES]; + + uint32_t MmHubPadding[7]; // SMU internal use +} Watermarks_t; + +typedef enum { + CUSTOM_DPM_SETTING_GFXCLK, + CUSTOM_DPM_SETTING_CCLK, + CUSTOM_DPM_SETTING_FCLK_CCX, + CUSTOM_DPM_SETTING_FCLK_GFX, + CUSTOM_DPM_SETTING_FCLK_STALLS, + CUSTOM_DPM_SETTING_LCLK, + CUSTOM_DPM_SETTING_COUNT, +} CUSTOM_DPM_SETTING_e; + +typedef struct { + uint8_t ActiveHystLimit; + uint8_t IdleHystLimit; + uint8_t FPS; + uint8_t MinActiveFreqType; + FloatInIntFormat_t MinActiveFreq; + FloatInIntFormat_t PD_Data_limit; + FloatInIntFormat_t PD_Data_time_constant; + FloatInIntFormat_t PD_Data_error_coeff; + FloatInIntFormat_t PD_Data_error_rate_coeff; +} DpmActivityMonitorCoeffExt_t; + +typedef struct { + DpmActivityMonitorCoeffExt_t DpmActivityMonitorCoeff[CUSTOM_DPM_SETTING_COUNT]; +} CustomDpmSettings_t; + + +#define NUM_DCFCLK_DPM_LEVELS 8 +#define NUM_SOCCLK_DPM_LEVELS 8 +#define NUM_FCLK_DPM_LEVELS 4 +#define NUM_MEMCLK_DPM_LEVELS 4 +#define NUM_VCN_DPM_LEVELS 8 + +typedef struct { + uint32_t Freq; // In MHz + uint32_t Vol; // Millivolts with 2 fractional bits +} DpmClock_t; + +typedef struct { + DpmClock_t DcfClocks[NUM_DCFCLK_DPM_LEVELS]; + DpmClock_t SocClocks[NUM_SOCCLK_DPM_LEVELS]; + DpmClock_t FClocks[NUM_FCLK_DPM_LEVELS]; + DpmClock_t MemClocks[NUM_MEMCLK_DPM_LEVELS]; + DpmClock_t VClocks[NUM_VCN_DPM_LEVELS]; + DpmClock_t DClocks[NUM_VCN_DPM_LEVELS]; + + uint8_t NumDcfClkDpmEnabled; + uint8_t NumSocClkDpmEnabled; + uint8_t NumFClkDpmEnabled; + uint8_t NumMemClkDpmEnabled; + uint8_t NumVClkDpmEnabled; + uint8_t NumDClkDpmEnabled; + uint8_t spare[2]; +} DpmClocks_t; + + +typedef enum { + CLOCK_SMNCLK = 0, + CLOCK_SOCCLK, + CLOCK_MP0CLK, + CLOCK_MP1CLK, + CLOCK_MP2CLK, + CLOCK_VCLK, + CLOCK_LCLK, + CLOCK_DCLK, + CLOCK_ACLK, + CLOCK_ISPCLK, + CLOCK_SHUBCLK, + CLOCK_DISPCLK, + CLOCK_DPPCLK, + CLOCK_DPREFCLK, + CLOCK_DCFCLK, + CLOCK_FCLK, + CLOCK_UMCCLK, + CLOCK_GFXCLK, + CLOCK_COUNT, +} CLOCK_IDs_e; + +// Throttler Status Bitmask +#define THROTTLER_STATUS_BIT_SPL 0 +#define THROTTLER_STATUS_BIT_FPPT 1 +#define THROTTLER_STATUS_BIT_SPPT 2 +#define THROTTLER_STATUS_BIT_SPPT_APU 3 +#define THROTTLER_STATUS_BIT_THM_CORE 4 +#define THROTTLER_STATUS_BIT_THM_GFX 5 +#define THROTTLER_STATUS_BIT_THM_SOC 6 +#define THROTTLER_STATUS_BIT_TDC_VDD 7 +#define THROTTLER_STATUS_BIT_TDC_SOC 8 + +typedef struct { + uint16_t ClockFrequency[CLOCK_COUNT]; //[MHz] + + uint16_t AverageGfxclkFrequency; //[MHz] + uint16_t AverageSocclkFrequency; //[MHz] + uint16_t AverageVclkFrequency; //[MHz] + uint16_t AverageFclkFrequency; //[MHz] + + uint16_t AverageGfxActivity; //[centi] + uint16_t AverageUvdActivity; //[centi] + + uint16_t Voltage[2]; //[mV] indices: VDDCR_VDD, VDDCR_SOC + uint16_t Current[2]; //[mA] indices: VDDCR_VDD, VDDCR_SOC + uint16_t Power[2]; //[mW] indices: VDDCR_VDD, VDDCR_SOC + + uint16_t FanPwm; //[milli] + uint16_t CurrentSocketPower; //[mW] + + uint16_t CoreFrequency[8]; //[MHz] + uint16_t CorePower[8]; //[mW] + uint16_t CoreTemperature[8]; //[centi-Celsius] + uint16_t L3Frequency[2]; //[MHz] + uint16_t L3Temperature[2]; //[centi-Celsius] + + uint16_t GfxTemperature; //[centi-Celsius] + uint16_t SocTemperature; //[centi-Celsius] + uint16_t ThrottlerStatus; + uint16_t spare; +} SmuMetrics_t; + + +// Workload bits +#define WORKLOAD_PPLIB_FULL_SCREEN_3D_BIT 0 +#define WORKLOAD_PPLIB_VIDEO_BIT 2 +#define WORKLOAD_PPLIB_VR_BIT 3 +#define WORKLOAD_PPLIB_COMPUTE_BIT 4 +#define WORKLOAD_PPLIB_CUSTOM_BIT 5 +#define WORKLOAD_PPLIB_COUNT 6 + +#define TABLE_BIOS_IF 0 // Called by BIOS +#define TABLE_WATERMARKS 1 // Called by Driver +#define TABLE_CUSTOM_DPM 2 // Called by Driver +#define TABLE_SPARE1 3 +#define TABLE_DPMCLOCKS 4 // Called by Driver +#define TABLE_MOMENTARY_PM 5 // Called by Tools +#define TABLE_MODERN_STDBY 6 // Called by Tools for Modern Standby Log +#define TABLE_SMU_METRICS 7 // Called by Driver +#define TABLE_COUNT 8 + + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu_types.h b/drivers/gpu/drm/amd/powerplay/inc/smu_types.h new file mode 100644 index 000000000000..d8c9b7f91fcc --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/inc/smu_types.h @@ -0,0 +1,266 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#ifndef __SMU_TYPES_H__ +#define __SMU_TYPES_H__ + +#define SMU_MESSAGE_TYPES \ + __SMU_DUMMY_MAP(TestMessage), \ + __SMU_DUMMY_MAP(GetSmuVersion), \ + __SMU_DUMMY_MAP(GetDriverIfVersion), \ + __SMU_DUMMY_MAP(SetAllowedFeaturesMaskLow), \ + __SMU_DUMMY_MAP(SetAllowedFeaturesMaskHigh), \ + __SMU_DUMMY_MAP(EnableAllSmuFeatures), \ + __SMU_DUMMY_MAP(DisableAllSmuFeatures), \ + __SMU_DUMMY_MAP(EnableSmuFeaturesLow), \ + __SMU_DUMMY_MAP(EnableSmuFeaturesHigh), \ + __SMU_DUMMY_MAP(DisableSmuFeaturesLow), \ + __SMU_DUMMY_MAP(DisableSmuFeaturesHigh), \ + __SMU_DUMMY_MAP(GetEnabledSmuFeaturesLow), \ + __SMU_DUMMY_MAP(GetEnabledSmuFeaturesHigh), \ + __SMU_DUMMY_MAP(SetWorkloadMask), \ + __SMU_DUMMY_MAP(SetPptLimit), \ + __SMU_DUMMY_MAP(SetDriverDramAddrHigh), \ + __SMU_DUMMY_MAP(SetDriverDramAddrLow), \ + __SMU_DUMMY_MAP(SetToolsDramAddrHigh), \ + __SMU_DUMMY_MAP(SetToolsDramAddrLow), \ + __SMU_DUMMY_MAP(TransferTableSmu2Dram), \ + __SMU_DUMMY_MAP(TransferTableDram2Smu), \ + __SMU_DUMMY_MAP(UseDefaultPPTable), \ + __SMU_DUMMY_MAP(UseBackupPPTable), \ + __SMU_DUMMY_MAP(RunBtc), \ + __SMU_DUMMY_MAP(RequestI2CBus), \ + __SMU_DUMMY_MAP(ReleaseI2CBus), \ + __SMU_DUMMY_MAP(SetFloorSocVoltage), \ + __SMU_DUMMY_MAP(SoftReset), \ + __SMU_DUMMY_MAP(StartBacoMonitor), \ + __SMU_DUMMY_MAP(CancelBacoMonitor), \ + __SMU_DUMMY_MAP(EnterBaco), \ + __SMU_DUMMY_MAP(SetSoftMinByFreq), \ + __SMU_DUMMY_MAP(SetSoftMaxByFreq), \ + __SMU_DUMMY_MAP(SetHardMinByFreq), \ + __SMU_DUMMY_MAP(SetHardMaxByFreq), \ + __SMU_DUMMY_MAP(GetMinDpmFreq), \ + __SMU_DUMMY_MAP(GetMaxDpmFreq), \ + __SMU_DUMMY_MAP(GetDpmFreqByIndex), \ + __SMU_DUMMY_MAP(GetDpmClockFreq), \ + __SMU_DUMMY_MAP(GetSsVoltageByDpm), \ + __SMU_DUMMY_MAP(SetMemoryChannelConfig), \ + __SMU_DUMMY_MAP(SetGeminiMode), \ + __SMU_DUMMY_MAP(SetGeminiApertureHigh), \ + __SMU_DUMMY_MAP(SetGeminiApertureLow), \ + __SMU_DUMMY_MAP(SetMinLinkDpmByIndex), \ + __SMU_DUMMY_MAP(OverridePcieParameters), \ + __SMU_DUMMY_MAP(OverDriveSetPercentage), \ + __SMU_DUMMY_MAP(SetMinDeepSleepDcefclk), \ + __SMU_DUMMY_MAP(ReenableAcDcInterrupt), \ + __SMU_DUMMY_MAP(NotifyPowerSource), \ + __SMU_DUMMY_MAP(SetUclkFastSwitch), \ + __SMU_DUMMY_MAP(SetUclkDownHyst), \ + __SMU_DUMMY_MAP(GfxDeviceDriverReset), \ + __SMU_DUMMY_MAP(GetCurrentRpm), \ + __SMU_DUMMY_MAP(SetVideoFps), \ + __SMU_DUMMY_MAP(SetTjMax), \ + __SMU_DUMMY_MAP(SetFanTemperatureTarget), \ + __SMU_DUMMY_MAP(PrepareMp1ForUnload), \ + __SMU_DUMMY_MAP(DramLogSetDramAddrHigh), \ + __SMU_DUMMY_MAP(DramLogSetDramAddrLow), \ + __SMU_DUMMY_MAP(DramLogSetDramSize), \ + __SMU_DUMMY_MAP(SetFanMaxRpm), \ + __SMU_DUMMY_MAP(SetFanMinPwm), \ + __SMU_DUMMY_MAP(ConfigureGfxDidt), \ + __SMU_DUMMY_MAP(NumOfDisplays), \ + __SMU_DUMMY_MAP(RemoveMargins), \ + __SMU_DUMMY_MAP(ReadSerialNumTop32), \ + __SMU_DUMMY_MAP(ReadSerialNumBottom32), \ + __SMU_DUMMY_MAP(SetSystemVirtualDramAddrHigh), \ + __SMU_DUMMY_MAP(SetSystemVirtualDramAddrLow), \ + __SMU_DUMMY_MAP(WaflTest), \ + __SMU_DUMMY_MAP(SetFclkGfxClkRatio), \ + __SMU_DUMMY_MAP(AllowGfxOff), \ + __SMU_DUMMY_MAP(DisallowGfxOff), \ + __SMU_DUMMY_MAP(GetPptLimit), \ + __SMU_DUMMY_MAP(GetDcModeMaxDpmFreq), \ + __SMU_DUMMY_MAP(GetDebugData), \ + __SMU_DUMMY_MAP(SetXgmiMode), \ + __SMU_DUMMY_MAP(RunAfllBtc), \ + __SMU_DUMMY_MAP(ExitBaco), \ + __SMU_DUMMY_MAP(PrepareMp1ForReset), \ + __SMU_DUMMY_MAP(PrepareMp1ForShutdown), \ + __SMU_DUMMY_MAP(SetMGpuFanBoostLimitRpm), \ + __SMU_DUMMY_MAP(GetAVFSVoltageByDpm), \ + __SMU_DUMMY_MAP(PowerUpVcn), \ + __SMU_DUMMY_MAP(PowerDownVcn), \ + __SMU_DUMMY_MAP(PowerUpJpeg), \ + __SMU_DUMMY_MAP(PowerDownJpeg), \ + __SMU_DUMMY_MAP(BacoAudioD3PME), \ + __SMU_DUMMY_MAP(ArmD3), \ + __SMU_DUMMY_MAP(RunDcBtc), \ + __SMU_DUMMY_MAP(RunGfxDcBtc), \ + __SMU_DUMMY_MAP(RunSocDcBtc), \ + __SMU_DUMMY_MAP(SetMemoryChannelEnable), \ + __SMU_DUMMY_MAP(SetDfSwitchType), \ + __SMU_DUMMY_MAP(GetVoltageByDpm), \ + __SMU_DUMMY_MAP(GetVoltageByDpmOverdrive), \ + __SMU_DUMMY_MAP(PowerUpVcn0), \ + __SMU_DUMMY_MAP(PowerDownVcn0), \ + __SMU_DUMMY_MAP(PowerUpVcn1), \ + __SMU_DUMMY_MAP(PowerDownVcn1), \ + __SMU_DUMMY_MAP(PowerUpGfx), \ + __SMU_DUMMY_MAP(PowerDownIspByTile), \ + __SMU_DUMMY_MAP(PowerUpIspByTile), \ + __SMU_DUMMY_MAP(PowerDownSdma), \ + __SMU_DUMMY_MAP(PowerUpSdma), \ + __SMU_DUMMY_MAP(SetHardMinIspclkByFreq), \ + __SMU_DUMMY_MAP(SetHardMinVcn), \ + __SMU_DUMMY_MAP(Spare1), \ + __SMU_DUMMY_MAP(Spare2), \ + __SMU_DUMMY_MAP(SetAllowFclkSwitch), \ + __SMU_DUMMY_MAP(SetMinVideoGfxclkFreq), \ + __SMU_DUMMY_MAP(ActiveProcessNotify), \ + __SMU_DUMMY_MAP(SetCustomPolicy), \ + __SMU_DUMMY_MAP(QueryPowerLimit), \ + __SMU_DUMMY_MAP(SetGfxclkOverdriveByFreqVid), \ + __SMU_DUMMY_MAP(SetHardMinDcfclkByFreq), \ + __SMU_DUMMY_MAP(SetHardMinSocclkByFreq), \ + __SMU_DUMMY_MAP(ControlIgpuATS), \ + __SMU_DUMMY_MAP(SetMinVideoFclkFreq), \ + __SMU_DUMMY_MAP(SetMinDeepSleepDcfclk), \ + __SMU_DUMMY_MAP(ForcePowerDownGfx), \ + __SMU_DUMMY_MAP(SetPhyclkVoltageByFreq), \ + __SMU_DUMMY_MAP(SetDppclkVoltageByFreq), \ + __SMU_DUMMY_MAP(SetSoftMinVcn), \ + __SMU_DUMMY_MAP(EnablePostCode), \ + __SMU_DUMMY_MAP(GetGfxclkFrequency), \ + __SMU_DUMMY_MAP(GetFclkFrequency), \ + __SMU_DUMMY_MAP(GetMinGfxclkFrequency), \ + __SMU_DUMMY_MAP(GetMaxGfxclkFrequency), \ + __SMU_DUMMY_MAP(SetGfxCGPG), \ + __SMU_DUMMY_MAP(SetSoftMaxGfxClk), \ + __SMU_DUMMY_MAP(SetHardMinGfxClk), \ + __SMU_DUMMY_MAP(SetSoftMaxSocclkByFreq), \ + __SMU_DUMMY_MAP(SetSoftMaxFclkByFreq), \ + __SMU_DUMMY_MAP(SetSoftMaxVcn), \ + __SMU_DUMMY_MAP(PowerGateMmHub), \ + __SMU_DUMMY_MAP(UpdatePmeRestore), \ + __SMU_DUMMY_MAP(GpuChangeState), \ + __SMU_DUMMY_MAP(SetPowerLimitPercentage), \ + __SMU_DUMMY_MAP(ForceGfxContentSave), \ + __SMU_DUMMY_MAP(EnableTmdp48MHzRefclkPwrDown), \ + __SMU_DUMMY_MAP(PowerGateAtHub), \ + __SMU_DUMMY_MAP(SetSoftMinJpeg), \ + __SMU_DUMMY_MAP(SetHardMinFclkByFreq), \ + __SMU_DUMMY_MAP(DFCstateControl), \ + +#undef __SMU_DUMMY_MAP +#define __SMU_DUMMY_MAP(type) SMU_MSG_##type +enum smu_message_type { + SMU_MESSAGE_TYPES + SMU_MSG_MAX_COUNT, +}; + +enum smu_clk_type { + SMU_GFXCLK, + SMU_VCLK, + SMU_DCLK, + SMU_ECLK, + SMU_SOCCLK, + SMU_UCLK, + SMU_DCEFCLK, + SMU_DISPCLK, + SMU_PIXCLK, + SMU_PHYCLK, + SMU_FCLK, + SMU_SCLK, + SMU_MCLK, + SMU_PCIE, + SMU_OD_SCLK, + SMU_OD_MCLK, + SMU_OD_VDDC_CURVE, + SMU_OD_RANGE, + SMU_CLK_COUNT, +}; + +#define SMU_FEATURE_MASKS \ + __SMU_DUMMY_MAP(DPM_PREFETCHER), \ + __SMU_DUMMY_MAP(DPM_GFXCLK), \ + __SMU_DUMMY_MAP(DPM_UCLK), \ + __SMU_DUMMY_MAP(DPM_SOCCLK), \ + __SMU_DUMMY_MAP(DPM_UVD), \ + __SMU_DUMMY_MAP(DPM_VCE), \ + __SMU_DUMMY_MAP(ULV), \ + __SMU_DUMMY_MAP(DPM_MP0CLK), \ + __SMU_DUMMY_MAP(DPM_LINK), \ + __SMU_DUMMY_MAP(DPM_DCEFCLK), \ + __SMU_DUMMY_MAP(DS_GFXCLK), \ + __SMU_DUMMY_MAP(DS_SOCCLK), \ + __SMU_DUMMY_MAP(DS_LCLK), \ + __SMU_DUMMY_MAP(PPT), \ + __SMU_DUMMY_MAP(TDC), \ + __SMU_DUMMY_MAP(THERMAL), \ + __SMU_DUMMY_MAP(GFX_PER_CU_CG), \ + __SMU_DUMMY_MAP(RM), \ + __SMU_DUMMY_MAP(DS_DCEFCLK), \ + __SMU_DUMMY_MAP(ACDC), \ + __SMU_DUMMY_MAP(VR0HOT), \ + __SMU_DUMMY_MAP(VR1HOT), \ + __SMU_DUMMY_MAP(FW_CTF), \ + __SMU_DUMMY_MAP(LED_DISPLAY), \ + __SMU_DUMMY_MAP(FAN_CONTROL), \ + __SMU_DUMMY_MAP(GFX_EDC), \ + __SMU_DUMMY_MAP(GFXOFF), \ + __SMU_DUMMY_MAP(CG), \ + __SMU_DUMMY_MAP(DPM_FCLK), \ + __SMU_DUMMY_MAP(DS_FCLK), \ + __SMU_DUMMY_MAP(DS_MP1CLK), \ + __SMU_DUMMY_MAP(DS_MP0CLK), \ + __SMU_DUMMY_MAP(XGMI), \ + __SMU_DUMMY_MAP(DPM_GFX_PACE), \ + __SMU_DUMMY_MAP(MEM_VDDCI_SCALING), \ + __SMU_DUMMY_MAP(MEM_MVDD_SCALING), \ + __SMU_DUMMY_MAP(DS_UCLK), \ + __SMU_DUMMY_MAP(GFX_ULV), \ + __SMU_DUMMY_MAP(FW_DSTATE), \ + __SMU_DUMMY_MAP(BACO), \ + __SMU_DUMMY_MAP(VCN_PG), \ + __SMU_DUMMY_MAP(JPEG_PG), \ + __SMU_DUMMY_MAP(USB_PG), \ + __SMU_DUMMY_MAP(RSMU_SMN_CG), \ + __SMU_DUMMY_MAP(APCC_PLUS), \ + __SMU_DUMMY_MAP(GTHR), \ + __SMU_DUMMY_MAP(GFX_DCS), \ + __SMU_DUMMY_MAP(GFX_SS), \ + __SMU_DUMMY_MAP(OUT_OF_BAND_MONITOR), \ + __SMU_DUMMY_MAP(TEMP_DEPENDENT_VMIN), \ + __SMU_DUMMY_MAP(MMHUB_PG), \ + __SMU_DUMMY_MAP(ATHUB_PG), \ + __SMU_DUMMY_MAP(APCC_DFLL), \ + __SMU_DUMMY_MAP(WAFL_CG), + +#undef __SMU_DUMMY_MAP +#define __SMU_DUMMY_MAP(feature) SMU_FEATURE_##feature##_BIT +enum smu_feature_mask { + SMU_FEATURE_MASKS + SMU_FEATURE_COUNT, +}; + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h b/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h new file mode 100644 index 000000000000..606149085683 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h @@ -0,0 +1,265 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __SMU_V11_0_H__ +#define __SMU_V11_0_H__ + +#include "amdgpu_smu.h" + +#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_NV10 0x33 +#define SMU11_DRIVER_IF_VERSION_NV14 0x34 + +/* MP Apertures */ +#define MP0_Public 0x03800000 +#define MP0_SRAM 0x03900000 +#define MP1_Public 0x03b00000 +#define MP1_SRAM 0x03c00004 +#define MP1_SMC_SIZE 0x40000 + +/* address block */ +#define smnMP1_FIRMWARE_FLAGS 0x3010024 +#define smnMP0_FW_INTF 0x30101c0 +#define smnMP1_PUB_CTRL 0x3010b14 + +#define TEMP_RANGE_MIN (0) +#define TEMP_RANGE_MAX (80 * 1000) + +#define SMU11_TOOL_SIZE 0x19000 + +#define MAX_PCIE_CONF 2 + +#define CLK_MAP(clk, index) \ + [SMU_##clk] = {1, (index)} + +#define FEA_MAP(fea) \ + [SMU_FEATURE_##fea##_BIT] = {1, FEATURE_##fea##_BIT} + +#define TAB_MAP(tab) \ + [SMU_TABLE_##tab] = {1, TABLE_##tab} + +#define PWR_MAP(tab) \ + [SMU_POWER_SOURCE_##tab] = {1, POWER_SOURCE_##tab} + +#define WORKLOAD_MAP(profile, workload) \ + [profile] = {1, (workload)} + +static const struct smu_temperature_range smu11_thermal_policy[] = +{ + {-273150, 99000, 99000, -273150, 99000, 99000, -273150, 99000, 99000}, + { 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000}, +}; + +struct smu_11_0_cmn2aisc_mapping { + int valid_mapping; + int map_to; +}; + +struct smu_11_0_max_sustainable_clocks { + uint32_t display_clock; + uint32_t phy_clock; + uint32_t pixel_clock; + uint32_t uclock; + uint32_t dcef_clock; + uint32_t soc_clock; +}; + +struct smu_11_0_dpm_table { + uint32_t min; /* MHz */ + uint32_t max; /* MHz */ +}; + +struct smu_11_0_pcie_table { + uint8_t pcie_gen[MAX_PCIE_CONF]; + uint8_t pcie_lane[MAX_PCIE_CONF]; +}; + +struct smu_11_0_dpm_tables { + struct smu_11_0_dpm_table soc_table; + struct smu_11_0_dpm_table gfx_table; + struct smu_11_0_dpm_table uclk_table; + struct smu_11_0_dpm_table eclk_table; + struct smu_11_0_dpm_table vclk_table; + struct smu_11_0_dpm_table dclk_table; + struct smu_11_0_dpm_table dcef_table; + struct smu_11_0_dpm_table pixel_table; + struct smu_11_0_dpm_table display_table; + struct smu_11_0_dpm_table phy_table; + struct smu_11_0_dpm_table fclk_table; + struct smu_11_0_pcie_table pcie_table; +}; + +struct smu_11_0_dpm_context { + struct smu_11_0_dpm_tables dpm_tables; + uint32_t workload_policy_mask; + uint32_t dcef_min_ds_clk; +}; + +enum smu_11_0_power_state { + SMU_11_0_POWER_STATE__D0 = 0, + SMU_11_0_POWER_STATE__D1, + SMU_11_0_POWER_STATE__D3, /* Sleep*/ + SMU_11_0_POWER_STATE__D4, /* Hibernate*/ + SMU_11_0_POWER_STATE__D5, /* Power off*/ +}; + +struct smu_11_0_power_context { + uint32_t power_source; + uint8_t in_power_limit_boost_mode; + enum smu_11_0_power_state power_state; +}; + +enum smu_v11_0_baco_seq { + BACO_SEQ_BACO = 0, + BACO_SEQ_MSR, + BACO_SEQ_BAMACO, + BACO_SEQ_ULPS, + BACO_SEQ_COUNT, +}; + +int smu_v11_0_init_microcode(struct smu_context *smu); + +int smu_v11_0_load_microcode(struct smu_context *smu); + +int smu_v11_0_init_smc_tables(struct smu_context *smu); + +int smu_v11_0_fini_smc_tables(struct smu_context *smu); + +int smu_v11_0_init_power(struct smu_context *smu); + +int smu_v11_0_fini_power(struct smu_context *smu); + +int smu_v11_0_check_fw_status(struct smu_context *smu); + +int smu_v11_0_setup_pptable(struct smu_context *smu); + +int smu_v11_0_get_vbios_bootup_values(struct smu_context *smu); + +int smu_v11_0_get_clk_info_from_vbios(struct smu_context *smu); + +int smu_v11_0_check_pptable(struct smu_context *smu); + +int smu_v11_0_parse_pptable(struct smu_context *smu); + +int smu_v11_0_populate_smc_pptable(struct smu_context *smu); + +int smu_v11_0_check_fw_version(struct smu_context *smu); + +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_tool_table_location(struct smu_context *smu); + +int smu_v11_0_notify_memory_pool_location(struct smu_context *smu); + +int smu_v11_0_system_features_control(struct smu_context *smu, + bool en); + +int smu_v11_0_send_msg(struct smu_context *smu, uint16_t msg); + +int +smu_v11_0_send_msg_with_param(struct smu_context *smu, uint16_t msg, + uint32_t param); + +int smu_v11_0_read_arg(struct smu_context *smu, uint32_t *arg); + +int smu_v11_0_init_display_count(struct smu_context *smu, uint32_t count); + +int smu_v11_0_set_allowed_mask(struct smu_context *smu); + +int smu_v11_0_get_enabled_mask(struct smu_context *smu, + uint32_t *feature_mask, uint32_t num); + +int smu_v11_0_notify_display_change(struct smu_context *smu); + +int smu_v11_0_set_power_limit(struct smu_context *smu, uint32_t n); + +int smu_v11_0_get_current_clk_freq(struct smu_context *smu, + enum smu_clk_type clk_id, + uint32_t *value); + +int smu_v11_0_init_max_sustainable_clocks(struct smu_context *smu); + +int smu_v11_0_start_thermal_control(struct smu_context *smu); + +int smu_v11_0_stop_thermal_control(struct smu_context *smu); + +int smu_v11_0_read_sensor(struct smu_context *smu, + enum amd_pp_sensors sensor, + void *data, uint32_t *size); + +int smu_v11_0_set_deep_sleep_dcefclk(struct smu_context *smu, uint32_t clk); + +int +smu_v11_0_display_clock_voltage_request(struct smu_context *smu, + struct pp_display_clock_request + *clock_req); + +uint32_t +smu_v11_0_get_fan_control_mode(struct smu_context *smu); + +int +smu_v11_0_set_fan_control_mode(struct smu_context *smu, + uint32_t mode); + +int +smu_v11_0_set_fan_speed_percent(struct smu_context *smu, uint32_t speed); + +int smu_v11_0_set_fan_speed_rpm(struct smu_context *smu, + uint32_t speed); + +int smu_v11_0_set_xgmi_pstate(struct smu_context *smu, + uint32_t pstate); + +int smu_v11_0_gfx_off_control(struct smu_context *smu, bool enable); + +int smu_v11_0_register_irq_handler(struct smu_context *smu); + +int smu_v11_0_set_azalia_d3_pme(struct smu_context *smu); + +int smu_v11_0_get_max_sustainable_clocks_by_dc(struct smu_context *smu, + struct pp_smu_nv_clock_table *max_clocks); + +bool smu_v11_0_baco_is_support(struct smu_context *smu); + +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_get_dpm_ultimate_freq(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t *min, uint32_t *max); + +int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t min, uint32_t max); + +int smu_v11_0_override_pcie_parameters(struct smu_context *smu); + +int smu_v11_0_set_default_od_settings(struct smu_context *smu, bool initialize, size_t overdrive_table_size); + +uint32_t smu_v11_0_get_max_power_limit(struct smu_context *smu); + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0_ppsmc.h b/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0_ppsmc.h new file mode 100644 index 000000000000..373861ddccd0 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0_ppsmc.h @@ -0,0 +1,131 @@ +/* + * Copyright 2018 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef SMU_V11_0_PPSMC_H +#define SMU_V11_0_PPSMC_H + +// SMU Response Codes: +#define PPSMC_Result_OK 0x1 +#define PPSMC_Result_Failed 0xFF +#define PPSMC_Result_UnknownCmd 0xFE +#define PPSMC_Result_CmdRejectedPrereq 0xFD +#define PPSMC_Result_CmdRejectedBusy 0xFC + +// Message Definitions: +// BASIC +#define PPSMC_MSG_TestMessage 0x1 +#define PPSMC_MSG_GetSmuVersion 0x2 +#define PPSMC_MSG_GetDriverIfVersion 0x3 +#define PPSMC_MSG_SetAllowedFeaturesMaskLow 0x4 +#define PPSMC_MSG_SetAllowedFeaturesMaskHigh 0x5 +#define PPSMC_MSG_EnableAllSmuFeatures 0x6 +#define PPSMC_MSG_DisableAllSmuFeatures 0x7 +#define PPSMC_MSG_EnableSmuFeaturesLow 0x8 +#define PPSMC_MSG_EnableSmuFeaturesHigh 0x9 +#define PPSMC_MSG_DisableSmuFeaturesLow 0xA +#define PPSMC_MSG_DisableSmuFeaturesHigh 0xB +#define PPSMC_MSG_GetEnabledSmuFeaturesLow 0xC +#define PPSMC_MSG_GetEnabledSmuFeaturesHigh 0xD +#define PPSMC_MSG_SetDriverDramAddrHigh 0xE +#define PPSMC_MSG_SetDriverDramAddrLow 0xF +#define PPSMC_MSG_SetToolsDramAddrHigh 0x10 +#define PPSMC_MSG_SetToolsDramAddrLow 0x11 +#define PPSMC_MSG_TransferTableSmu2Dram 0x12 +#define PPSMC_MSG_TransferTableDram2Smu 0x13 +#define PPSMC_MSG_UseDefaultPPTable 0x14 +#define PPSMC_MSG_UseBackupPPTable 0x15 +#define PPSMC_MSG_SetSystemVirtualDramAddrHigh 0x16 +#define PPSMC_MSG_SetSystemVirtualDramAddrLow 0x17 + +//BACO/BAMACO/BOMACO +#define PPSMC_MSG_EnterBaco 0x18 +#define PPSMC_MSG_ExitBaco 0x19 +#define PPSMC_MSG_ArmD3 0x46 + +//DPM +#define PPSMC_MSG_SetSoftMinByFreq 0x1A +#define PPSMC_MSG_SetSoftMaxByFreq 0x1B +#define PPSMC_MSG_SetHardMinByFreq 0x1C +#define PPSMC_MSG_SetHardMaxByFreq 0x1D +#define PPSMC_MSG_GetMinDpmFreq 0x1E +#define PPSMC_MSG_GetMaxDpmFreq 0x1F +#define PPSMC_MSG_GetDpmFreqByIndex 0x20 +#define PPSMC_MSG_OverridePcieParameters 0x21 +#define PPSMC_MSG_SetMinDeepSleepDcefclk 0x22 + +#define PPSMC_MSG_SetWorkloadMask 0x24 +#define PPSMC_MSG_SetUclkFastSwitch 0x25 +#define PPSMC_MSG_GetVoltageByDpm 0x26 +#define PPSMC_MSG_SetVideoFps 0x27 +#define PPSMC_MSG_GetDcModeMaxDpmFreq 0x28 + +//Power Gating +#define PPSMC_MSG_AllowGfxOff 0x29 +#define PPSMC_MSG_DisallowGfxOff 0x2A +#define PPSMC_MSG_PowerUpVcn 0x2B +#define PPSMC_MSG_PowerDownVcn 0x2C +#define PPSMC_MSG_PowerUpJpeg 0x2D +#define PPSMC_MSG_PowerDownJpeg 0x2E +//reserve 0x29 to 0x30 for PG harvesting TBD + +//Resets +#define PPSMC_MSG_PrepareMp1ForUnload 0x32 +#define PPSMC_MSG_PrepareMp1ForReset 0x33 +#define PPSMC_MSG_PrepareMp1ForShutdown 0x34 + +//ACDC Power Source +#define PPSMC_MSG_SetPptLimit 0x35 +#define PPSMC_MSG_GetPptLimit 0x36 +#define PPSMC_MSG_ReenableAcDcInterrupt 0x37 +#define PPSMC_MSG_NotifyPowerSource 0x38 + +//BTC +#define PPSMC_MSG_RunBtc 0x3A + +//Debug +#define PPSMC_MSG_DramLogSetDramAddrHigh 0x3B +#define PPSMC_MSG_DramLogSetDramAddrLow 0x3C +#define PPSMC_MSG_DramLogSetDramSize 0x3D +#define PPSMC_MSG_GetDebugData 0x3E + +//Others +#define PPSMC_MSG_ConfigureGfxDidt 0x3F +#define PPSMC_MSG_NumOfDisplays 0x40 + +#define PPSMC_MSG_SetMemoryChannelConfig 0x41 +#define PPSMC_MSG_SetGeminiMode 0x42 +#define PPSMC_MSG_SetGeminiApertureHigh 0x43 +#define PPSMC_MSG_SetGeminiApertureLow 0x44 + +#define PPSMC_MSG_GetVoltageByDpmOverdrive 0x45 +#define PPSMC_MSG_BacoAudioD3PME 0x48 + +#define PPSMC_Message_Count 0x49 + +typedef uint32_t PPSMC_Result; +typedef uint32_t PPSMC_Msg; + +//for use with PPSMC_MSG_GetVoltageByDpmOverdrive +#define PPSMC_GET_AVFS_CURVE 0 +#define PPSMC_GET_OVERDRIVE_CURVE 1 +#endif diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0_pptable.h b/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0_pptable.h new file mode 100644 index 000000000000..b2f96a101124 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0_pptable.h @@ -0,0 +1,149 @@ +/* + * Copyright 2018 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#ifndef SMU_11_0_PPTABLE_H +#define SMU_11_0_PPTABLE_H + + +#define SMU_11_0_TABLE_FORMAT_REVISION 12 + +//// POWERPLAYTABLE::ulPlatformCaps +#define SMU_11_0_PP_PLATFORM_CAP_POWERPLAY 0x1 +#define SMU_11_0_PP_PLATFORM_CAP_SBIOSPOWERSOURCE 0x2 +#define SMU_11_0_PP_PLATFORM_CAP_HARDWAREDC 0x4 +#define SMU_11_0_PP_PLATFORM_CAP_BACO 0x8 +#define SMU_11_0_PP_PLATFORM_CAP_MACO 0x10 +#define SMU_11_0_PP_PLATFORM_CAP_SHADOWPSTATE 0x20 + +// SMU_11_0_PP_THERMALCONTROLLER - Thermal Controller Type +#define SMU_11_0_PP_THERMALCONTROLLER_NONE 0 + +#define SMU_11_0_PP_OVERDRIVE_VERSION 0x0800 +#define SMU_11_0_PP_POWERSAVINGCLOCK_VERSION 0x0100 + +enum SMU_11_0_ODFEATURE_ID { + SMU_11_0_ODFEATURE_GFXCLK_LIMITS = 1 << 0, //GFXCLK Limit feature + SMU_11_0_ODFEATURE_GFXCLK_CURVE = 1 << 1, //GFXCLK Curve feature + SMU_11_0_ODFEATURE_UCLK_MAX = 1 << 2, //UCLK Limit feature + SMU_11_0_ODFEATURE_POWER_LIMIT = 1 << 3, //Power Limit feature + SMU_11_0_ODFEATURE_FAN_ACOUSTIC_LIMIT = 1 << 4, //Fan Acoustic RPM feature + SMU_11_0_ODFEATURE_FAN_SPEED_MIN = 1 << 5, //Minimum Fan Speed feature + SMU_11_0_ODFEATURE_TEMPERATURE_FAN = 1 << 6, //Fan Target Temperature Limit feature + SMU_11_0_ODFEATURE_TEMPERATURE_SYSTEM = 1 << 7, //Operating Temperature Limit feature + SMU_11_0_ODFEATURE_MEMORY_TIMING_TUNE = 1 << 8, //AC Timing Tuning feature + SMU_11_0_ODFEATURE_FAN_ZERO_RPM_CONTROL = 1 << 9, //Zero RPM feature + SMU_11_0_ODFEATURE_AUTO_UV_ENGINE = 1 << 10, //Auto Under Volt GFXCLK feature + SMU_11_0_ODFEATURE_AUTO_OC_ENGINE = 1 << 11, //Auto Over Clock GFXCLK feature + SMU_11_0_ODFEATURE_AUTO_OC_MEMORY = 1 << 12, //Auto Over Clock MCLK feature + SMU_11_0_ODFEATURE_FAN_CURVE = 1 << 13, //VICTOR TODO + SMU_11_0_ODFEATURE_COUNT = 14, +}; +#define SMU_11_0_MAX_ODFEATURE 32 //Maximum Number of OD Features + +enum SMU_11_0_ODSETTING_ID { + SMU_11_0_ODSETTING_GFXCLKFMAX = 0, + SMU_11_0_ODSETTING_GFXCLKFMIN, + SMU_11_0_ODSETTING_VDDGFXCURVEFREQ_P1, + SMU_11_0_ODSETTING_VDDGFXCURVEVOLTAGE_P1, + SMU_11_0_ODSETTING_VDDGFXCURVEFREQ_P2, + SMU_11_0_ODSETTING_VDDGFXCURVEVOLTAGE_P2, + SMU_11_0_ODSETTING_VDDGFXCURVEFREQ_P3, + SMU_11_0_ODSETTING_VDDGFXCURVEVOLTAGE_P3, + SMU_11_0_ODSETTING_UCLKFMAX, + SMU_11_0_ODSETTING_POWERPERCENTAGE, + SMU_11_0_ODSETTING_FANRPMMIN, + SMU_11_0_ODSETTING_FANRPMACOUSTICLIMIT, + SMU_11_0_ODSETTING_FANTARGETTEMPERATURE, + SMU_11_0_ODSETTING_OPERATINGTEMPMAX, + SMU_11_0_ODSETTING_ACTIMING, + SMU_11_0_ODSETTING_FAN_ZERO_RPM_CONTROL, + SMU_11_0_ODSETTING_AUTOUVENGINE, + SMU_11_0_ODSETTING_AUTOOCENGINE, + SMU_11_0_ODSETTING_AUTOOCMEMORY, + SMU_11_0_ODSETTING_COUNT, +}; +#define SMU_11_0_MAX_ODSETTING 32 //Maximum Number of ODSettings + +struct smu_11_0_overdrive_table +{ + uint8_t revision; //Revision = SMU_11_0_PP_OVERDRIVE_VERSION + uint8_t reserve[3]; //Zero filled field reserved for future use + uint32_t feature_count; //Total number of supported features + uint32_t setting_count; //Total number of supported settings + uint8_t cap[SMU_11_0_MAX_ODFEATURE]; //OD feature support flags + uint32_t max[SMU_11_0_MAX_ODSETTING]; //default maximum settings + uint32_t min[SMU_11_0_MAX_ODSETTING]; //default minimum settings +} __attribute__((packed)); + +enum SMU_11_0_PPCLOCK_ID { + SMU_11_0_PPCLOCK_GFXCLK = 0, + SMU_11_0_PPCLOCK_VCLK, + SMU_11_0_PPCLOCK_DCLK, + SMU_11_0_PPCLOCK_ECLK, + SMU_11_0_PPCLOCK_SOCCLK, + SMU_11_0_PPCLOCK_UCLK, + SMU_11_0_PPCLOCK_DCEFCLK, + SMU_11_0_PPCLOCK_DISPCLK, + SMU_11_0_PPCLOCK_PIXCLK, + SMU_11_0_PPCLOCK_PHYCLK, + SMU_11_0_PPCLOCK_COUNT, +}; +#define SMU_11_0_MAX_PPCLOCK 16 //Maximum Number of PP Clocks + +struct smu_11_0_power_saving_clock_table +{ + uint8_t revision; //Revision = SMU_11_0_PP_POWERSAVINGCLOCK_VERSION + uint8_t reserve[3]; //Zero filled field reserved for future use + uint32_t count; //power_saving_clock_count = SMU_11_0_PPCLOCK_COUNT + uint32_t max[SMU_11_0_MAX_PPCLOCK]; //PowerSavingClock Mode Clock Maximum array In MHz + uint32_t min[SMU_11_0_MAX_PPCLOCK]; //PowerSavingClock Mode Clock Minimum array In MHz +} __attribute__((packed)); + +struct smu_11_0_powerplay_table +{ + struct atom_common_table_header header; + uint8_t table_revision; + uint16_t table_size; //Driver portion table size. The offset to smc_pptable including header size + uint32_t golden_pp_id; + uint32_t golden_revision; + uint16_t format_id; + uint32_t platform_caps; //POWERPLAYABLE::ulPlatformCaps + + uint8_t thermal_controller_type; //one of SMU_11_0_PP_THERMALCONTROLLER + + uint16_t small_power_limit1; + uint16_t small_power_limit2; + uint16_t boost_power_limit; + uint16_t od_turbo_power_limit; //Power limit setting for Turbo mode in Performance UI Tuning. + uint16_t od_power_save_power_limit; //Power limit setting for PowerSave/Optimal mode in Performance UI Tuning. + uint16_t software_shutdown_temp; + + uint16_t reserve[6]; //Zero filled field reserved for future use + + struct smu_11_0_power_saving_clock_table power_saving_clock; + struct smu_11_0_overdrive_table overdrive_table; + +#ifndef SMU_11_0_PARTIAL_PPTABLE + PPTable_t smc_pptable; //PPTable_t in smu11_driver_if.h +#endif +} __attribute__((packed)); + +#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 new file mode 100644 index 000000000000..9b9f5df0911c --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/inc/smu_v12_0.h @@ -0,0 +1,81 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __SMU_V12_0_H__ +#define __SMU_V12_0_H__ + +#include "amdgpu_smu.h" + +/* MP Apertures */ +#define MP0_Public 0x03800000 +#define MP0_SRAM 0x03900000 +#define MP1_Public 0x03b00000 +#define MP1_SRAM 0x03c00004 + + +struct smu_12_0_cmn2aisc_mapping { + int valid_mapping; + int map_to; +}; + +int smu_v12_0_send_msg_without_waiting(struct smu_context *smu, + uint16_t msg); + +int smu_v12_0_read_arg(struct smu_context *smu, uint32_t *arg); + +int smu_v12_0_wait_for_response(struct smu_context *smu); + +int smu_v12_0_send_msg(struct smu_context *smu, uint16_t msg); + +int +smu_v12_0_send_msg_with_param(struct smu_context *smu, uint16_t msg, + uint32_t param); + +int smu_v12_0_check_fw_status(struct smu_context *smu); + +int smu_v12_0_check_fw_version(struct smu_context *smu); + +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_set_gfx_cgpg(struct smu_context *smu, bool enable); + +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); + +int smu_v12_0_init_smc_tables(struct smu_context *smu); + +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_dpm_ultimate_freq(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t *min, uint32_t *max); + +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); + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu_v12_0_ppsmc.h b/drivers/gpu/drm/amd/powerplay/inc/smu_v12_0_ppsmc.h new file mode 100644 index 000000000000..9ac9f3bd3664 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/inc/smu_v12_0_ppsmc.h @@ -0,0 +1,106 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef SMU_12_0_PPSMC_H +#define SMU_12_0_PPSMC_H + +// SMU Response Codes: +#define PPSMC_Result_OK 0x1 +#define PPSMC_Result_Failed 0xFF +#define PPSMC_Result_UnknownCmd 0xFE +#define PPSMC_Result_CmdRejectedPrereq 0xFD +#define PPSMC_Result_CmdRejectedBusy 0xFC + + +// Message Definitions: +#define PPSMC_MSG_TestMessage 0x1 +#define PPSMC_MSG_GetSmuVersion 0x2 +#define PPSMC_MSG_GetDriverIfVersion 0x3 +#define PPSMC_MSG_PowerUpGfx 0x6 +#define PPSMC_MSG_EnableGfxOff 0x7 +#define PPSMC_MSG_DisableGfxOff 0x8 +#define PPSMC_MSG_PowerDownIspByTile 0x9 // ISP is power gated by default +#define PPSMC_MSG_PowerUpIspByTile 0xA +#define PPSMC_MSG_PowerDownVcn 0xB // VCN is power gated by default +#define PPSMC_MSG_PowerUpVcn 0xC +#define PPSMC_MSG_PowerDownSdma 0xD // SDMA is power gated by default +#define PPSMC_MSG_PowerUpSdma 0xE +#define PPSMC_MSG_SetHardMinIspclkByFreq 0xF +#define PPSMC_MSG_SetHardMinVcn 0x10 // For wireless display +#define PPSMC_MSG_spare1 0x11 +#define PPSMC_MSG_spare2 0x12 +#define PPSMC_MSG_SetAllowFclkSwitch 0x13 +#define PPSMC_MSG_SetMinVideoGfxclkFreq 0x14 +#define PPSMC_MSG_ActiveProcessNotify 0x15 +#define PPSMC_MSG_SetCustomPolicy 0x16 +#define PPSMC_MSG_SetVideoFps 0x17 +#define PPSMC_MSG_SetDisplayCount 0x18 // Moved to VBIOS +#define PPSMC_MSG_QueryPowerLimit 0x19 //Driver to look up sustainable clocks for VQ +#define PPSMC_MSG_SetDriverDramAddrHigh 0x1A +#define PPSMC_MSG_SetDriverDramAddrLow 0x1B +#define PPSMC_MSG_TransferTableSmu2Dram 0x1C +#define PPSMC_MSG_TransferTableDram2Smu 0x1D +#define PPSMC_MSG_GfxDeviceDriverReset 0x1E +#define PPSMC_MSG_SetGfxclkOverdriveByFreqVid 0x1F +#define PPSMC_MSG_SetHardMinDcfclkByFreq 0x20 // Moved to VBIOS +#define PPSMC_MSG_SetHardMinSocclkByFreq 0x21 +#define PPSMC_MSG_ControlIgpuATS 0x22 +#define PPSMC_MSG_SetMinVideoFclkFreq 0x23 +#define PPSMC_MSG_SetMinDeepSleepDcfclk 0x24 // Moved to VBIOS +#define PPSMC_MSG_ForcePowerDownGfx 0x25 +#define PPSMC_MSG_SetPhyclkVoltageByFreq 0x26 // Moved to VBIOS +#define PPSMC_MSG_SetDppclkVoltageByFreq 0x27 // Moved to VBIOS and is SetDppclkFreq +#define PPSMC_MSG_SetSoftMinVcn 0x28 +#define PPSMC_MSG_EnablePostCode 0x29 +#define PPSMC_MSG_GetGfxclkFrequency 0x2A +#define PPSMC_MSG_GetFclkFrequency 0x2B +#define PPSMC_MSG_GetMinGfxclkFrequency 0x2C +#define PPSMC_MSG_GetMaxGfxclkFrequency 0x2D +#define PPSMC_MSG_SoftReset 0x2E // Not supported +#define PPSMC_MSG_SetGfxCGPG 0x2F +#define PPSMC_MSG_SetSoftMaxGfxClk 0x30 +#define PPSMC_MSG_SetHardMinGfxClk 0x31 +#define PPSMC_MSG_SetSoftMaxSocclkByFreq 0x32 +#define PPSMC_MSG_SetSoftMaxFclkByFreq 0x33 +#define PPSMC_MSG_SetSoftMaxVcn 0x34 +#define PPSMC_MSG_PowerGateMmHub 0x35 +#define PPSMC_MSG_UpdatePmeRestore 0x36 // Moved to VBIOS +#define PPSMC_MSG_GpuChangeState 0x37 +#define PPSMC_MSG_SetPowerLimitPercentage 0x38 +#define PPSMC_MSG_ForceGfxContentSave 0x39 +#define PPSMC_MSG_EnableTmdp48MHzRefclkPwrDown 0x3A // Moved to VBIOS +#define PPSMC_MSG_PowerDownJpeg 0x3B +#define PPSMC_MSG_PowerUpJpeg 0x3C +#define PPSMC_MSG_PowerGateAtHub 0x3D +#define PPSMC_MSG_SetSoftMinJpeg 0x3E +#define PPSMC_MSG_SetHardMinFclkByFreq 0x3F +#define PPSMC_Message_Count 0x40 + + +//Argument for PPSMC_MSG_GpuChangeState +enum { + eGpuChangeState_D0Entry = 1, + eGpuChangeState_D3Entry, +}; + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/inc/smumgr.h b/drivers/gpu/drm/amd/powerplay/inc/smumgr.h index 82550a8a3a3f..c5288831aa15 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/smumgr.h +++ b/drivers/gpu/drm/amd/powerplay/inc/smumgr.h @@ -41,6 +41,7 @@ enum SMU_MEMBER { HandshakeDisables = 0, VoltageChangeTimeout, AverageGraphicsActivity, + AverageMemoryActivity, PreVBlankGap, VBlankTimeout, UcodeLoadStatus, diff --git a/drivers/gpu/drm/amd/powerplay/inc/vega20_ppsmc.h b/drivers/gpu/drm/amd/powerplay/inc/vega20_ppsmc.h index 4f63a736ea0e..0c66f0fe1aaf 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/vega20_ppsmc.h +++ b/drivers/gpu/drm/amd/powerplay/inc/vega20_ppsmc.h @@ -119,7 +119,9 @@ #define PPSMC_MSG_PrepareMp1ForShutdown 0x5A #define PPSMC_MSG_SetMGpuFanBoostLimitRpm 0x5D #define PPSMC_MSG_GetAVFSVoltageByDpm 0x5F -#define PPSMC_Message_Count 0x60 +#define PPSMC_MSG_BacoWorkAroundFlushVDCI 0x60 +#define PPSMC_MSG_DFCstateControl 0x63 +#define PPSMC_Message_Count 0x64 typedef uint32_t PPSMC_Result; typedef uint32_t PPSMC_Msg; diff --git a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c new file mode 100644 index 000000000000..aaec884d63ed --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c @@ -0,0 +1,2097 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "pp_debug.h" +#include <linux/firmware.h> +#include <linux/pci.h> +#include "amdgpu.h" +#include "amdgpu_smu.h" +#include "smu_internal.h" +#include "atomfirmware.h" +#include "amdgpu_atomfirmware.h" +#include "smu_v11_0.h" +#include "smu11_driver_if_navi10.h" +#include "soc15_common.h" +#include "atom.h" +#include "navi10_ppt.h" +#include "smu_v11_0_pptable.h" +#include "smu_v11_0_ppsmc.h" +#include "nbio/nbio_7_4_sh_mask.h" + +#include "asic_reg/mp/mp_11_0_sh_mask.h" + +#define FEATURE_MASK(feature) (1ULL << feature) +#define SMC_DPM_FEATURE ( \ + FEATURE_MASK(FEATURE_DPM_PREFETCHER_BIT) | \ + FEATURE_MASK(FEATURE_DPM_GFXCLK_BIT) | \ + FEATURE_MASK(FEATURE_DPM_GFX_PACE_BIT) | \ + FEATURE_MASK(FEATURE_DPM_UCLK_BIT) | \ + FEATURE_MASK(FEATURE_DPM_SOCCLK_BIT) | \ + FEATURE_MASK(FEATURE_DPM_MP0CLK_BIT) | \ + FEATURE_MASK(FEATURE_DPM_LINK_BIT) | \ + FEATURE_MASK(FEATURE_DPM_DCEFCLK_BIT)) + +#define MSG_MAP(msg, index) \ + [SMU_MSG_##msg] = {1, (index)} + +static struct smu_11_0_cmn2aisc_mapping navi10_message_map[SMU_MSG_MAX_COUNT] = { + MSG_MAP(TestMessage, PPSMC_MSG_TestMessage), + MSG_MAP(GetSmuVersion, PPSMC_MSG_GetSmuVersion), + MSG_MAP(GetDriverIfVersion, PPSMC_MSG_GetDriverIfVersion), + MSG_MAP(SetAllowedFeaturesMaskLow, PPSMC_MSG_SetAllowedFeaturesMaskLow), + MSG_MAP(SetAllowedFeaturesMaskHigh, PPSMC_MSG_SetAllowedFeaturesMaskHigh), + MSG_MAP(EnableAllSmuFeatures, PPSMC_MSG_EnableAllSmuFeatures), + MSG_MAP(DisableAllSmuFeatures, PPSMC_MSG_DisableAllSmuFeatures), + MSG_MAP(EnableSmuFeaturesLow, PPSMC_MSG_EnableSmuFeaturesLow), + MSG_MAP(EnableSmuFeaturesHigh, PPSMC_MSG_EnableSmuFeaturesHigh), + MSG_MAP(DisableSmuFeaturesLow, PPSMC_MSG_DisableSmuFeaturesLow), + MSG_MAP(DisableSmuFeaturesHigh, PPSMC_MSG_DisableSmuFeaturesHigh), + MSG_MAP(GetEnabledSmuFeaturesLow, PPSMC_MSG_GetEnabledSmuFeaturesLow), + MSG_MAP(GetEnabledSmuFeaturesHigh, PPSMC_MSG_GetEnabledSmuFeaturesHigh), + MSG_MAP(SetWorkloadMask, PPSMC_MSG_SetWorkloadMask), + MSG_MAP(SetPptLimit, PPSMC_MSG_SetPptLimit), + MSG_MAP(SetDriverDramAddrHigh, PPSMC_MSG_SetDriverDramAddrHigh), + MSG_MAP(SetDriverDramAddrLow, PPSMC_MSG_SetDriverDramAddrLow), + MSG_MAP(SetToolsDramAddrHigh, PPSMC_MSG_SetToolsDramAddrHigh), + MSG_MAP(SetToolsDramAddrLow, PPSMC_MSG_SetToolsDramAddrLow), + MSG_MAP(TransferTableSmu2Dram, PPSMC_MSG_TransferTableSmu2Dram), + MSG_MAP(TransferTableDram2Smu, PPSMC_MSG_TransferTableDram2Smu), + MSG_MAP(UseDefaultPPTable, PPSMC_MSG_UseDefaultPPTable), + MSG_MAP(UseBackupPPTable, PPSMC_MSG_UseBackupPPTable), + MSG_MAP(RunBtc, PPSMC_MSG_RunBtc), + MSG_MAP(EnterBaco, PPSMC_MSG_EnterBaco), + MSG_MAP(SetSoftMinByFreq, PPSMC_MSG_SetSoftMinByFreq), + MSG_MAP(SetSoftMaxByFreq, PPSMC_MSG_SetSoftMaxByFreq), + MSG_MAP(SetHardMinByFreq, PPSMC_MSG_SetHardMinByFreq), + MSG_MAP(SetHardMaxByFreq, PPSMC_MSG_SetHardMaxByFreq), + MSG_MAP(GetMinDpmFreq, PPSMC_MSG_GetMinDpmFreq), + MSG_MAP(GetMaxDpmFreq, PPSMC_MSG_GetMaxDpmFreq), + MSG_MAP(GetDpmFreqByIndex, PPSMC_MSG_GetDpmFreqByIndex), + MSG_MAP(SetMemoryChannelConfig, PPSMC_MSG_SetMemoryChannelConfig), + MSG_MAP(SetGeminiMode, PPSMC_MSG_SetGeminiMode), + MSG_MAP(SetGeminiApertureHigh, PPSMC_MSG_SetGeminiApertureHigh), + MSG_MAP(SetGeminiApertureLow, PPSMC_MSG_SetGeminiApertureLow), + MSG_MAP(OverridePcieParameters, PPSMC_MSG_OverridePcieParameters), + MSG_MAP(SetMinDeepSleepDcefclk, PPSMC_MSG_SetMinDeepSleepDcefclk), + MSG_MAP(ReenableAcDcInterrupt, PPSMC_MSG_ReenableAcDcInterrupt), + MSG_MAP(NotifyPowerSource, PPSMC_MSG_NotifyPowerSource), + MSG_MAP(SetUclkFastSwitch, PPSMC_MSG_SetUclkFastSwitch), + MSG_MAP(SetVideoFps, PPSMC_MSG_SetVideoFps), + MSG_MAP(PrepareMp1ForUnload, PPSMC_MSG_PrepareMp1ForUnload), + MSG_MAP(DramLogSetDramAddrHigh, PPSMC_MSG_DramLogSetDramAddrHigh), + MSG_MAP(DramLogSetDramAddrLow, PPSMC_MSG_DramLogSetDramAddrLow), + MSG_MAP(DramLogSetDramSize, PPSMC_MSG_DramLogSetDramSize), + MSG_MAP(ConfigureGfxDidt, PPSMC_MSG_ConfigureGfxDidt), + MSG_MAP(NumOfDisplays, PPSMC_MSG_NumOfDisplays), + MSG_MAP(SetSystemVirtualDramAddrHigh, PPSMC_MSG_SetSystemVirtualDramAddrHigh), + MSG_MAP(SetSystemVirtualDramAddrLow, PPSMC_MSG_SetSystemVirtualDramAddrLow), + MSG_MAP(AllowGfxOff, PPSMC_MSG_AllowGfxOff), + MSG_MAP(DisallowGfxOff, PPSMC_MSG_DisallowGfxOff), + MSG_MAP(GetPptLimit, PPSMC_MSG_GetPptLimit), + MSG_MAP(GetDcModeMaxDpmFreq, PPSMC_MSG_GetDcModeMaxDpmFreq), + MSG_MAP(GetDebugData, PPSMC_MSG_GetDebugData), + MSG_MAP(ExitBaco, PPSMC_MSG_ExitBaco), + MSG_MAP(PrepareMp1ForReset, PPSMC_MSG_PrepareMp1ForReset), + MSG_MAP(PrepareMp1ForShutdown, PPSMC_MSG_PrepareMp1ForShutdown), + MSG_MAP(PowerUpVcn, PPSMC_MSG_PowerUpVcn), + MSG_MAP(PowerDownVcn, PPSMC_MSG_PowerDownVcn), + MSG_MAP(PowerUpJpeg, PPSMC_MSG_PowerUpJpeg), + MSG_MAP(PowerDownJpeg, PPSMC_MSG_PowerDownJpeg), + MSG_MAP(BacoAudioD3PME, PPSMC_MSG_BacoAudioD3PME), + MSG_MAP(ArmD3, PPSMC_MSG_ArmD3), +}; + +static struct smu_11_0_cmn2aisc_mapping navi10_clk_map[SMU_CLK_COUNT] = { + CLK_MAP(GFXCLK, PPCLK_GFXCLK), + CLK_MAP(SCLK, PPCLK_GFXCLK), + CLK_MAP(SOCCLK, PPCLK_SOCCLK), + CLK_MAP(FCLK, PPCLK_SOCCLK), + CLK_MAP(UCLK, PPCLK_UCLK), + CLK_MAP(MCLK, PPCLK_UCLK), + CLK_MAP(DCLK, PPCLK_DCLK), + CLK_MAP(VCLK, PPCLK_VCLK), + CLK_MAP(DCEFCLK, PPCLK_DCEFCLK), + CLK_MAP(DISPCLK, PPCLK_DISPCLK), + CLK_MAP(PIXCLK, PPCLK_PIXCLK), + CLK_MAP(PHYCLK, PPCLK_PHYCLK), +}; + +static struct smu_11_0_cmn2aisc_mapping navi10_feature_mask_map[SMU_FEATURE_COUNT] = { + FEA_MAP(DPM_PREFETCHER), + FEA_MAP(DPM_GFXCLK), + FEA_MAP(DPM_GFX_PACE), + FEA_MAP(DPM_UCLK), + FEA_MAP(DPM_SOCCLK), + FEA_MAP(DPM_MP0CLK), + FEA_MAP(DPM_LINK), + FEA_MAP(DPM_DCEFCLK), + FEA_MAP(MEM_VDDCI_SCALING), + FEA_MAP(MEM_MVDD_SCALING), + FEA_MAP(DS_GFXCLK), + FEA_MAP(DS_SOCCLK), + FEA_MAP(DS_LCLK), + FEA_MAP(DS_DCEFCLK), + FEA_MAP(DS_UCLK), + FEA_MAP(GFX_ULV), + FEA_MAP(FW_DSTATE), + FEA_MAP(GFXOFF), + FEA_MAP(BACO), + FEA_MAP(VCN_PG), + FEA_MAP(JPEG_PG), + FEA_MAP(USB_PG), + FEA_MAP(RSMU_SMN_CG), + FEA_MAP(PPT), + FEA_MAP(TDC), + FEA_MAP(GFX_EDC), + FEA_MAP(APCC_PLUS), + FEA_MAP(GTHR), + FEA_MAP(ACDC), + FEA_MAP(VR0HOT), + FEA_MAP(VR1HOT), + FEA_MAP(FW_CTF), + FEA_MAP(FAN_CONTROL), + FEA_MAP(THERMAL), + FEA_MAP(GFX_DCS), + FEA_MAP(RM), + FEA_MAP(LED_DISPLAY), + FEA_MAP(GFX_SS), + FEA_MAP(OUT_OF_BAND_MONITOR), + FEA_MAP(TEMP_DEPENDENT_VMIN), + FEA_MAP(MMHUB_PG), + FEA_MAP(ATHUB_PG), + FEA_MAP(APCC_DFLL), +}; + +static struct smu_11_0_cmn2aisc_mapping navi10_table_map[SMU_TABLE_COUNT] = { + TAB_MAP(PPTABLE), + TAB_MAP(WATERMARKS), + TAB_MAP(AVFS), + TAB_MAP(AVFS_PSM_DEBUG), + TAB_MAP(AVFS_FUSE_OVERRIDE), + TAB_MAP(PMSTATUSLOG), + TAB_MAP(SMU_METRICS), + TAB_MAP(DRIVER_SMU_CONFIG), + TAB_MAP(ACTIVITY_MONITOR_COEFF), + TAB_MAP(OVERDRIVE), + TAB_MAP(I2C_COMMANDS), + TAB_MAP(PACE), +}; + +static struct smu_11_0_cmn2aisc_mapping navi10_pwr_src_map[SMU_POWER_SOURCE_COUNT] = { + PWR_MAP(AC), + PWR_MAP(DC), +}; + +static struct smu_11_0_cmn2aisc_mapping navi10_workload_map[PP_SMC_POWER_PROFILE_COUNT] = { + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT, WORKLOAD_PPLIB_DEFAULT_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_FULLSCREEN3D, WORKLOAD_PPLIB_FULL_SCREEN_3D_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_POWERSAVING, WORKLOAD_PPLIB_POWER_SAVING_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_VIDEO, WORKLOAD_PPLIB_VIDEO_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_VR, WORKLOAD_PPLIB_VR_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_COMPUTE, WORKLOAD_PPLIB_COMPUTE_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_CUSTOM, WORKLOAD_PPLIB_CUSTOM_BIT), +}; + +static int navi10_get_smu_msg_index(struct smu_context *smc, uint32_t index) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (index >= SMU_MSG_MAX_COUNT) + return -EINVAL; + + mapping = navi10_message_map[index]; + if (!(mapping.valid_mapping)) { + return -EINVAL; + } + + return mapping.map_to; +} + +static int navi10_get_smu_clk_index(struct smu_context *smc, uint32_t index) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (index >= SMU_CLK_COUNT) + return -EINVAL; + + mapping = navi10_clk_map[index]; + if (!(mapping.valid_mapping)) { + return -EINVAL; + } + + return mapping.map_to; +} + +static int navi10_get_smu_feature_index(struct smu_context *smc, uint32_t index) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (index >= SMU_FEATURE_COUNT) + return -EINVAL; + + mapping = navi10_feature_mask_map[index]; + if (!(mapping.valid_mapping)) { + return -EINVAL; + } + + return mapping.map_to; +} + +static int navi10_get_smu_table_index(struct smu_context *smc, uint32_t index) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (index >= SMU_TABLE_COUNT) + return -EINVAL; + + mapping = navi10_table_map[index]; + if (!(mapping.valid_mapping)) { + return -EINVAL; + } + + return mapping.map_to; +} + +static int navi10_get_pwr_src_index(struct smu_context *smc, uint32_t index) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (index >= SMU_POWER_SOURCE_COUNT) + return -EINVAL; + + mapping = navi10_pwr_src_map[index]; + if (!(mapping.valid_mapping)) { + return -EINVAL; + } + + return mapping.map_to; +} + + +static int navi10_get_workload_type(struct smu_context *smu, enum PP_SMC_POWER_PROFILE profile) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (profile > PP_SMC_POWER_PROFILE_CUSTOM) + return -EINVAL; + + mapping = navi10_workload_map[profile]; + if (!(mapping.valid_mapping)) { + return -EINVAL; + } + + return mapping.map_to; +} + +static bool is_asic_secure(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + bool is_secure = true; + uint32_t mp0_fw_intf; + + mp0_fw_intf = RREG32_PCIE(MP0_Public | + (smnMP0_FW_INTF & 0xffffffff)); + + if (!(mp0_fw_intf & (1 << 19))) + is_secure = false; + + return is_secure; +} + +static int +navi10_get_allowed_feature_mask(struct smu_context *smu, + uint32_t *feature_mask, uint32_t num) +{ + struct amdgpu_device *adev = smu->adev; + + if (num > 2) + return -EINVAL; + + memset(feature_mask, 0, sizeof(uint32_t) * num); + + *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_DPM_PREFETCHER_BIT) + | FEATURE_MASK(FEATURE_DPM_MP0CLK_BIT) + | FEATURE_MASK(FEATURE_RSMU_SMN_CG_BIT) + | FEATURE_MASK(FEATURE_DS_SOCCLK_BIT) + | FEATURE_MASK(FEATURE_PPT_BIT) + | FEATURE_MASK(FEATURE_TDC_BIT) + | FEATURE_MASK(FEATURE_GFX_EDC_BIT) + | FEATURE_MASK(FEATURE_APCC_PLUS_BIT) + | FEATURE_MASK(FEATURE_VR0HOT_BIT) + | FEATURE_MASK(FEATURE_FAN_CONTROL_BIT) + | FEATURE_MASK(FEATURE_THERMAL_BIT) + | FEATURE_MASK(FEATURE_LED_DISPLAY_BIT) + | FEATURE_MASK(FEATURE_DS_LCLK_BIT) + | FEATURE_MASK(FEATURE_DS_DCEFCLK_BIT) + | FEATURE_MASK(FEATURE_FW_DSTATE_BIT) + | FEATURE_MASK(FEATURE_BACO_BIT) + | FEATURE_MASK(FEATURE_ACDC_BIT) + | FEATURE_MASK(FEATURE_GFX_SS_BIT) + | FEATURE_MASK(FEATURE_APCC_DFLL_BIT) + | FEATURE_MASK(FEATURE_FW_CTF_BIT) + | FEATURE_MASK(FEATURE_OUT_OF_BAND_MONITOR_BIT); + + if (adev->pm.pp_feature & PP_SOCCLK_DPM_MASK) + *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_DPM_SOCCLK_BIT); + + if (adev->pm.pp_feature & PP_SCLK_DPM_MASK) + *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_DPM_GFXCLK_BIT); + + if (adev->pm.pp_feature & PP_PCIE_DPM_MASK) + *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_DPM_LINK_BIT); + + if (adev->pm.pp_feature & PP_DCEFCLK_DPM_MASK) + *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_DPM_DCEFCLK_BIT); + + if (adev->pm.pp_feature & PP_MCLK_DPM_MASK) + *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_DPM_UCLK_BIT) + | FEATURE_MASK(FEATURE_MEM_VDDCI_SCALING_BIT) + | FEATURE_MASK(FEATURE_MEM_MVDD_SCALING_BIT); + + if (adev->pm.pp_feature & PP_ULV_MASK) + *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_GFX_ULV_BIT); + + if (adev->pm.pp_feature & PP_SCLK_DEEP_SLEEP_MASK) + *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_DS_GFXCLK_BIT); + + if (adev->pm.pp_feature & PP_GFXOFF_MASK) + *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_GFXOFF_BIT); + + if (smu->adev->pg_flags & AMD_PG_SUPPORT_MMHUB) + *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_MMHUB_PG_BIT); + + if (smu->adev->pg_flags & AMD_PG_SUPPORT_ATHUB) + *(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); + + /* disable DPM UCLK and DS SOCCLK on navi10 A0 secure board */ + if (is_asic_secure(smu)) { + /* only for navi10 A0 */ + if ((adev->asic_type == CHIP_NAVI10) && + (adev->rev_id == 0)) { + *(uint64_t *)feature_mask &= + ~(FEATURE_MASK(FEATURE_DPM_UCLK_BIT) + | FEATURE_MASK(FEATURE_MEM_VDDCI_SCALING_BIT) + | FEATURE_MASK(FEATURE_MEM_MVDD_SCALING_BIT)); + *(uint64_t *)feature_mask &= + ~FEATURE_MASK(FEATURE_DS_SOCCLK_BIT); + } + } + + return 0; +} + +static int navi10_check_powerplay_table(struct smu_context *smu) +{ + return 0; +} + +static int navi10_append_powerplay_table(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + struct smu_table_context *table_context = &smu->smu_table; + PPTable_t *smc_pptable = table_context->driver_pptable; + struct atom_smc_dpm_info_v4_5 *smc_dpm_table; + int index, ret; + + index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1, + smc_dpm_info); + + ret = smu_get_atom_data_table(smu, index, NULL, NULL, NULL, + (uint8_t **)&smc_dpm_table); + if (ret) + return ret; + + memcpy(smc_pptable->I2cControllers, smc_dpm_table->I2cControllers, + sizeof(I2cControllerConfig_t) * NUM_I2C_CONTROLLERS); + + /* SVI2 Board Parameters */ + smc_pptable->MaxVoltageStepGfx = smc_dpm_table->MaxVoltageStepGfx; + smc_pptable->MaxVoltageStepSoc = smc_dpm_table->MaxVoltageStepSoc; + smc_pptable->VddGfxVrMapping = smc_dpm_table->VddGfxVrMapping; + smc_pptable->VddSocVrMapping = smc_dpm_table->VddSocVrMapping; + smc_pptable->VddMem0VrMapping = smc_dpm_table->VddMem0VrMapping; + smc_pptable->VddMem1VrMapping = smc_dpm_table->VddMem1VrMapping; + smc_pptable->GfxUlvPhaseSheddingMask = smc_dpm_table->GfxUlvPhaseSheddingMask; + smc_pptable->SocUlvPhaseSheddingMask = smc_dpm_table->SocUlvPhaseSheddingMask; + smc_pptable->ExternalSensorPresent = smc_dpm_table->ExternalSensorPresent; + smc_pptable->Padding8_V = smc_dpm_table->Padding8_V; + + /* Telemetry Settings */ + smc_pptable->GfxMaxCurrent = smc_dpm_table->GfxMaxCurrent; + smc_pptable->GfxOffset = smc_dpm_table->GfxOffset; + smc_pptable->Padding_TelemetryGfx = smc_dpm_table->Padding_TelemetryGfx; + smc_pptable->SocMaxCurrent = smc_dpm_table->SocMaxCurrent; + smc_pptable->SocOffset = smc_dpm_table->SocOffset; + smc_pptable->Padding_TelemetrySoc = smc_dpm_table->Padding_TelemetrySoc; + smc_pptable->Mem0MaxCurrent = smc_dpm_table->Mem0MaxCurrent; + smc_pptable->Mem0Offset = smc_dpm_table->Mem0Offset; + smc_pptable->Padding_TelemetryMem0 = smc_dpm_table->Padding_TelemetryMem0; + smc_pptable->Mem1MaxCurrent = smc_dpm_table->Mem1MaxCurrent; + smc_pptable->Mem1Offset = smc_dpm_table->Mem1Offset; + smc_pptable->Padding_TelemetryMem1 = smc_dpm_table->Padding_TelemetryMem1; + + /* GPIO Settings */ + smc_pptable->AcDcGpio = smc_dpm_table->AcDcGpio; + smc_pptable->AcDcPolarity = smc_dpm_table->AcDcPolarity; + smc_pptable->VR0HotGpio = smc_dpm_table->VR0HotGpio; + smc_pptable->VR0HotPolarity = smc_dpm_table->VR0HotPolarity; + smc_pptable->VR1HotGpio = smc_dpm_table->VR1HotGpio; + smc_pptable->VR1HotPolarity = smc_dpm_table->VR1HotPolarity; + smc_pptable->GthrGpio = smc_dpm_table->GthrGpio; + smc_pptable->GthrPolarity = smc_dpm_table->GthrPolarity; + + /* LED Display Settings */ + smc_pptable->LedPin0 = smc_dpm_table->LedPin0; + smc_pptable->LedPin1 = smc_dpm_table->LedPin1; + smc_pptable->LedPin2 = smc_dpm_table->LedPin2; + smc_pptable->padding8_4 = smc_dpm_table->padding8_4; + + /* GFXCLK PLL Spread Spectrum */ + smc_pptable->PllGfxclkSpreadEnabled = smc_dpm_table->PllGfxclkSpreadEnabled; + smc_pptable->PllGfxclkSpreadPercent = smc_dpm_table->PllGfxclkSpreadPercent; + smc_pptable->PllGfxclkSpreadFreq = smc_dpm_table->PllGfxclkSpreadFreq; + + /* GFXCLK DFLL Spread Spectrum */ + smc_pptable->DfllGfxclkSpreadEnabled = smc_dpm_table->DfllGfxclkSpreadEnabled; + smc_pptable->DfllGfxclkSpreadPercent = smc_dpm_table->DfllGfxclkSpreadPercent; + smc_pptable->DfllGfxclkSpreadFreq = smc_dpm_table->DfllGfxclkSpreadFreq; + + /* UCLK Spread Spectrum */ + smc_pptable->UclkSpreadEnabled = smc_dpm_table->UclkSpreadEnabled; + smc_pptable->UclkSpreadPercent = smc_dpm_table->UclkSpreadPercent; + smc_pptable->UclkSpreadFreq = smc_dpm_table->UclkSpreadFreq; + + /* SOCCLK Spread Spectrum */ + smc_pptable->SoclkSpreadEnabled = smc_dpm_table->SoclkSpreadEnabled; + smc_pptable->SocclkSpreadPercent = smc_dpm_table->SocclkSpreadPercent; + smc_pptable->SocclkSpreadFreq = smc_dpm_table->SocclkSpreadFreq; + + /* Total board power */ + smc_pptable->TotalBoardPower = smc_dpm_table->TotalBoardPower; + smc_pptable->BoardPadding = smc_dpm_table->BoardPadding; + + /* Mvdd Svi2 Div Ratio Setting */ + smc_pptable->MvddRatio = smc_dpm_table->MvddRatio; + + if (adev->pm.pp_feature & PP_GFXOFF_MASK) { + /* TODO: remove it once SMU fw fix it */ + smc_pptable->DebugOverrides |= DPM_OVERRIDE_DISABLE_DFLL_PLL_SHUTDOWN; + } + + return 0; +} + +static int navi10_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; + + if (!table_context->power_play_table) + return -EINVAL; + + powerplay_table = table_context->power_play_table; + + memcpy(table_context->driver_pptable, &powerplay_table->smc_pptable, + sizeof(PPTable_t)); + + 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 0; +} + +static int navi10_tables_init(struct smu_context *smu, struct smu_table *tables) +{ + struct smu_table_context *smu_table = &smu->smu_table; + + SMU_TABLE_INIT(tables, SMU_TABLE_PPTABLE, sizeof(PPTable_t), + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + SMU_TABLE_INIT(tables, SMU_TABLE_WATERMARKS, sizeof(Watermarks_t), + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + SMU_TABLE_INIT(tables, SMU_TABLE_SMU_METRICS, sizeof(SmuMetrics_t), + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + SMU_TABLE_INIT(tables, SMU_TABLE_OVERDRIVE, sizeof(OverDriveTable_t), + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + SMU_TABLE_INIT(tables, SMU_TABLE_PMSTATUSLOG, SMU11_TOOL_SIZE, + 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; + smu_table->metrics_time = 0; + + return 0; +} + +static int navi10_get_metrics_table(struct smu_context *smu, + SmuMetrics_t *metrics_table) +{ + struct smu_table_context *smu_table= &smu->smu_table; + int ret = 0; + + 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"); + return ret; + } + smu_table->metrics_time = jiffies; + } + + memcpy(metrics_table, smu_table->metrics_table, sizeof(SmuMetrics_t)); + + return ret; +} + +static int navi10_allocate_dpm_context(struct smu_context *smu) +{ + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + + if (smu_dpm->dpm_context) + return -EINVAL; + + smu_dpm->dpm_context = kzalloc(sizeof(struct smu_11_0_dpm_context), + GFP_KERNEL); + if (!smu_dpm->dpm_context) + return -ENOMEM; + + smu_dpm->dpm_context_size = sizeof(struct smu_11_0_dpm_context); + + return 0; +} + +static int navi10_set_default_dpm_table(struct smu_context *smu) +{ + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + struct smu_table_context *table_context = &smu->smu_table; + struct smu_11_0_dpm_context *dpm_context = smu_dpm->dpm_context; + PPTable_t *driver_ppt = NULL; + int i; + + driver_ppt = table_context->driver_pptable; + + dpm_context->dpm_tables.soc_table.min = driver_ppt->FreqTableSocclk[0]; + dpm_context->dpm_tables.soc_table.max = driver_ppt->FreqTableSocclk[NUM_SOCCLK_DPM_LEVELS - 1]; + + dpm_context->dpm_tables.gfx_table.min = driver_ppt->FreqTableGfx[0]; + dpm_context->dpm_tables.gfx_table.max = driver_ppt->FreqTableGfx[NUM_GFXCLK_DPM_LEVELS - 1]; + + dpm_context->dpm_tables.uclk_table.min = driver_ppt->FreqTableUclk[0]; + dpm_context->dpm_tables.uclk_table.max = driver_ppt->FreqTableUclk[NUM_UCLK_DPM_LEVELS - 1]; + + dpm_context->dpm_tables.vclk_table.min = driver_ppt->FreqTableVclk[0]; + dpm_context->dpm_tables.vclk_table.max = driver_ppt->FreqTableVclk[NUM_VCLK_DPM_LEVELS - 1]; + + dpm_context->dpm_tables.dclk_table.min = driver_ppt->FreqTableDclk[0]; + dpm_context->dpm_tables.dclk_table.max = driver_ppt->FreqTableDclk[NUM_DCLK_DPM_LEVELS - 1]; + + dpm_context->dpm_tables.dcef_table.min = driver_ppt->FreqTableDcefclk[0]; + dpm_context->dpm_tables.dcef_table.max = driver_ppt->FreqTableDcefclk[NUM_DCEFCLK_DPM_LEVELS - 1]; + + dpm_context->dpm_tables.pixel_table.min = driver_ppt->FreqTablePixclk[0]; + dpm_context->dpm_tables.pixel_table.max = driver_ppt->FreqTablePixclk[NUM_PIXCLK_DPM_LEVELS - 1]; + + dpm_context->dpm_tables.display_table.min = driver_ppt->FreqTableDispclk[0]; + dpm_context->dpm_tables.display_table.max = driver_ppt->FreqTableDispclk[NUM_DISPCLK_DPM_LEVELS - 1]; + + dpm_context->dpm_tables.phy_table.min = driver_ppt->FreqTablePhyclk[0]; + dpm_context->dpm_tables.phy_table.max = driver_ppt->FreqTablePhyclk[NUM_PHYCLK_DPM_LEVELS - 1]; + + for (i = 0; i < MAX_PCIE_CONF; i++) { + dpm_context->dpm_tables.pcie_table.pcie_gen[i] = driver_ppt->PcieGenSpeed[i]; + dpm_context->dpm_tables.pcie_table.pcie_lane[i] = driver_ppt->PcieLaneCount[i]; + } + + return 0; +} + +static int navi10_dpm_set_uvd_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) { + /* vcn dpm on is a prerequisite for vcn power gate messages */ + if (smu_feature_is_enabled(smu, SMU_FEATURE_VCN_PG_BIT)) { + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_PowerUpVcn, 1); + if (ret) + return ret; + } + power_gate->vcn_gated = false; + } else { + if (smu_feature_is_enabled(smu, SMU_FEATURE_VCN_PG_BIT)) { + ret = smu_send_smc_msg(smu, SMU_MSG_PowerDownVcn); + if (ret) + return ret; + } + power_gate->vcn_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) +{ + int ret = 0, clk_id = 0; + SmuMetrics_t metrics; + + ret = navi10_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.CurrClock[clk_id]; + + return ret; +} + +static bool navi10_is_support_fine_grained_dpm(struct smu_context *smu, enum smu_clk_type clk_type) +{ + PPTable_t *pptable = smu->smu_table.driver_pptable; + DpmDescriptor_t *dpm_desc = NULL; + uint32_t clk_index = 0; + + clk_index = smu_clk_get_index(smu, clk_type); + dpm_desc = &pptable->DpmDescriptor[clk_index]; + + /* 0 - Fine grained DPM, 1 - Discrete DPM */ + return dpm_desc->SnapToDiscrete == 0 ? true : false; +} + +static inline bool navi10_od_feature_is_supported(struct smu_11_0_overdrive_table *od_table, enum SMU_11_0_ODFEATURE_ID feature) +{ + return od_table->cap[feature]; +} + + +static int navi10_print_clk_levels(struct smu_context *smu, + enum smu_clk_type clk_type, char *buf) +{ + uint16_t *curve_settings; + int i, size = 0, ret = 0; + uint32_t cur_value = 0, value = 0, count = 0; + uint32_t freq_values[3] = {0}; + uint32_t mark_index = 0; + struct smu_table_context *table_context = &smu->smu_table; + uint32_t gen_speed, lane_width; + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + struct smu_11_0_dpm_context *dpm_context = smu_dpm->dpm_context; + struct amdgpu_device *adev = smu->adev; + PPTable_t *pptable = (PPTable_t *)table_context->driver_pptable; + OverDriveTable_t *od_table = + (OverDriveTable_t *)table_context->overdrive_table; + struct smu_11_0_overdrive_table *od_settings = smu->od_settings; + + switch (clk_type) { + case SMU_GFXCLK: + case SMU_SCLK: + case SMU_SOCCLK: + case SMU_MCLK: + case SMU_UCLK: + case SMU_FCLK: + case SMU_DCEFCLK: + ret = smu_get_current_clk_freq(smu, clk_type, &cur_value); + if (ret) + return size; + + /* 10KHz -> MHz */ + cur_value = cur_value / 100; + + ret = smu_get_dpm_level_count(smu, clk_type, &count); + if (ret) + return size; + + if (!navi10_is_support_fine_grained_dpm(smu, clk_type)) { + for (i = 0; i < count; i++) { + ret = smu_get_dpm_freq_by_index(smu, clk_type, i, &value); + if (ret) + return size; + + size += sprintf(buf + size, "%d: %uMhz %s\n", i, value, + cur_value == value ? "*" : ""); + } + } else { + ret = smu_get_dpm_freq_by_index(smu, clk_type, 0, &freq_values[0]); + if (ret) + return size; + ret = smu_get_dpm_freq_by_index(smu, clk_type, count - 1, &freq_values[2]); + if (ret) + return size; + + freq_values[1] = cur_value; + mark_index = cur_value == freq_values[0] ? 0 : + cur_value == freq_values[2] ? 2 : 1; + if (mark_index != 1) + freq_values[1] = (freq_values[0] + freq_values[2]) / 2; + + for (i = 0; i < 3; i++) { + size += sprintf(buf + size, "%d: %uMhz %s\n", i, freq_values[i], + i == mark_index ? "*" : ""); + } + + } + break; + case SMU_PCIE: + gen_speed = (RREG32_PCIE(smnPCIE_LC_SPEED_CNTL) & + PSWUSP0_PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE_MASK) + >> PSWUSP0_PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE__SHIFT; + lane_width = (RREG32_PCIE(smnPCIE_LC_LINK_WIDTH_CNTL) & + PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_RD_MASK) + >> PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_RD__SHIFT; + for (i = 0; i < NUM_LINK_LEVELS; i++) + size += sprintf(buf + size, "%d: %s %s %dMhz %s\n", i, + (dpm_context->dpm_tables.pcie_table.pcie_gen[i] == 0) ? "2.5GT/s," : + (dpm_context->dpm_tables.pcie_table.pcie_gen[i] == 1) ? "5.0GT/s," : + (dpm_context->dpm_tables.pcie_table.pcie_gen[i] == 2) ? "8.0GT/s," : + (dpm_context->dpm_tables.pcie_table.pcie_gen[i] == 3) ? "16.0GT/s," : "", + (dpm_context->dpm_tables.pcie_table.pcie_lane[i] == 1) ? "x1" : + (dpm_context->dpm_tables.pcie_table.pcie_lane[i] == 2) ? "x2" : + (dpm_context->dpm_tables.pcie_table.pcie_lane[i] == 3) ? "x4" : + (dpm_context->dpm_tables.pcie_table.pcie_lane[i] == 4) ? "x8" : + (dpm_context->dpm_tables.pcie_table.pcie_lane[i] == 5) ? "x12" : + (dpm_context->dpm_tables.pcie_table.pcie_lane[i] == 6) ? "x16" : "", + pptable->LclkFreq[i], + (gen_speed == dpm_context->dpm_tables.pcie_table.pcie_gen[i]) && + (lane_width == dpm_context->dpm_tables.pcie_table.pcie_lane[i]) ? + "*" : ""); + break; + case SMU_OD_SCLK: + if (!smu->od_enabled || !od_table || !od_settings) + break; + if (!navi10_od_feature_is_supported(od_settings, SMU_11_0_ODFEATURE_GFXCLK_LIMITS)) + break; + size += sprintf(buf + size, "OD_SCLK:\n"); + size += sprintf(buf + size, "0: %uMhz\n1: %uMhz\n", od_table->GfxclkFmin, od_table->GfxclkFmax); + break; + case SMU_OD_MCLK: + if (!smu->od_enabled || !od_table || !od_settings) + break; + if (!navi10_od_feature_is_supported(od_settings, SMU_11_0_ODFEATURE_UCLK_MAX)) + break; + size += sprintf(buf + size, "OD_MCLK:\n"); + size += sprintf(buf + size, "0: %uMHz\n", od_table->UclkFmax); + break; + case SMU_OD_VDDC_CURVE: + if (!smu->od_enabled || !od_table || !od_settings) + break; + if (!navi10_od_feature_is_supported(od_settings, SMU_11_0_ODFEATURE_GFXCLK_CURVE)) + break; + size += sprintf(buf + size, "OD_VDDC_CURVE:\n"); + for (i = 0; i < 3; i++) { + switch (i) { + case 0: + curve_settings = &od_table->GfxclkFreq1; + break; + case 1: + curve_settings = &od_table->GfxclkFreq2; + break; + case 2: + curve_settings = &od_table->GfxclkFreq3; + break; + default: + break; + } + size += sprintf(buf + size, "%d: %uMHz @ %umV\n", i, curve_settings[0], curve_settings[1] / NAVI10_VOLTAGE_SCALE); + } + break; + default: + break; + } + + return size; +} + +static int navi10_force_clk_levels(struct smu_context *smu, + enum smu_clk_type clk_type, uint32_t mask) +{ + + int ret = 0, size = 0; + uint32_t soft_min_level = 0, soft_max_level = 0, min_freq = 0, max_freq = 0; + + soft_min_level = mask ? (ffs(mask) - 1) : 0; + soft_max_level = mask ? (fls(mask) - 1) : 0; + + switch (clk_type) { + case SMU_GFXCLK: + case SMU_SCLK: + case SMU_SOCCLK: + case SMU_MCLK: + case SMU_UCLK: + case SMU_DCEFCLK: + case SMU_FCLK: + /* There is only 2 levels for fine grained DPM */ + if (navi10_is_support_fine_grained_dpm(smu, clk_type)) { + soft_max_level = (soft_max_level >= 1 ? 1 : 0); + soft_min_level = (soft_min_level >= 1 ? 1 : 0); + } + + ret = smu_get_dpm_freq_by_index(smu, clk_type, soft_min_level, &min_freq); + if (ret) + return size; + + ret = smu_get_dpm_freq_by_index(smu, clk_type, soft_max_level, &max_freq); + if (ret) + return size; + + ret = smu_set_soft_freq_range(smu, clk_type, min_freq, max_freq); + if (ret) + return size; + break; + default: + break; + } + + return size; +} + +static int navi10_populate_umd_state_clk(struct smu_context *smu) +{ + int ret = 0; + uint32_t min_sclk_freq = 0, min_mclk_freq = 0; + + ret = smu_get_dpm_freq_range(smu, SMU_SCLK, &min_sclk_freq, NULL, false); + if (ret) + return ret; + + smu->pstate_sclk = min_sclk_freq * 100; + + ret = smu_get_dpm_freq_range(smu, SMU_MCLK, &min_mclk_freq, NULL, false); + if (ret) + return ret; + + smu->pstate_mclk = min_mclk_freq * 100; + + return ret; +} + +static int navi10_get_clock_by_type_with_latency(struct smu_context *smu, + enum smu_clk_type clk_type, + struct pp_clock_levels_with_latency *clocks) +{ + int ret = 0, i = 0; + uint32_t level_count = 0, freq = 0; + + switch (clk_type) { + case SMU_GFXCLK: + case SMU_DCEFCLK: + case SMU_SOCCLK: + ret = smu_get_dpm_level_count(smu, clk_type, &level_count); + if (ret) + return ret; + + level_count = min(level_count, (uint32_t)MAX_NUM_CLOCKS); + clocks->num_levels = level_count; + + for (i = 0; i < level_count; i++) { + ret = smu_get_dpm_freq_by_index(smu, clk_type, i, &freq); + if (ret) + return ret; + + clocks->data[i].clocks_in_khz = freq * 1000; + clocks->data[i].latency_in_us = 0; + } + break; + default: + break; + } + + return ret; +} + +static int navi10_pre_display_config_changed(struct smu_context *smu) +{ + int ret = 0; + uint32_t max_freq = 0; + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_NumOfDisplays, 0); + if (ret) + return ret; + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT)) { + ret = smu_get_dpm_freq_range(smu, SMU_UCLK, NULL, &max_freq, false); + if (ret) + return ret; + ret = smu_set_hard_freq_range(smu, SMU_UCLK, 0, max_freq); + if (ret) + return ret; + } + + return ret; +} + +static int navi10_display_config_changed(struct smu_context *smu) +{ + int ret = 0; + + if ((smu->watermarks_bitmap & WATERMARKS_EXIST) && + !(smu->watermarks_bitmap & WATERMARKS_LOADED)) { + ret = smu_write_watermarks_table(smu); + if (ret) + return ret; + + smu->watermarks_bitmap |= WATERMARKS_LOADED; + } + + if ((smu->watermarks_bitmap & WATERMARKS_EXIST) && + smu_feature_is_supported(smu, SMU_FEATURE_DPM_DCEFCLK_BIT) && + smu_feature_is_supported(smu, SMU_FEATURE_DPM_SOCCLK_BIT)) { + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_NumOfDisplays, + smu->display_config->num_display); + if (ret) + return ret; + } + + return ret; +} + +static int navi10_force_dpm_limit_value(struct smu_context *smu, bool highest) +{ + int ret = 0, i = 0; + uint32_t min_freq, max_freq, force_freq; + enum smu_clk_type clk_type; + + enum smu_clk_type clks[] = { + SMU_GFXCLK, + SMU_MCLK, + SMU_SOCCLK, + }; + + for (i = 0; i < ARRAY_SIZE(clks); i++) { + clk_type = clks[i]; + ret = smu_get_dpm_freq_range(smu, clk_type, &min_freq, &max_freq, false); + if (ret) + return ret; + + force_freq = highest ? max_freq : min_freq; + ret = smu_set_soft_freq_range(smu, clk_type, force_freq, force_freq); + if (ret) + return ret; + } + + return ret; +} + +static int navi10_unforce_dpm_levels(struct smu_context *smu) +{ + int ret = 0, i = 0; + uint32_t min_freq, max_freq; + enum smu_clk_type clk_type; + + enum smu_clk_type clks[] = { + SMU_GFXCLK, + SMU_MCLK, + SMU_SOCCLK, + }; + + for (i = 0; i < ARRAY_SIZE(clks); i++) { + clk_type = clks[i]; + ret = smu_get_dpm_freq_range(smu, clk_type, &min_freq, &max_freq, false); + if (ret) + return ret; + + ret = smu_set_soft_freq_range(smu, clk_type, min_freq, max_freq); + if (ret) + return ret; + } + + return ret; +} + +static int navi10_get_gpu_power(struct smu_context *smu, uint32_t *value) +{ + int ret = 0; + SmuMetrics_t metrics; + + if (!value) + return -EINVAL; + + ret = navi10_get_metrics_table(smu, &metrics); + if (ret) + return ret; + + *value = metrics.AverageSocketPower << 8; + + return 0; +} + +static int navi10_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 = navi10_get_metrics_table(smu, &metrics); + if (ret) + return ret; + + switch (sensor) { + case AMDGPU_PP_SENSOR_GPU_LOAD: + *value = metrics.AverageGfxActivity; + break; + case AMDGPU_PP_SENSOR_MEM_LOAD: + *value = metrics.AverageUclkActivity; + break; + default: + pr_err("Invalid sensor for retrieving clock activity\n"); + return -EINVAL; + } + + return 0; +} + +static bool navi10_is_dpm_running(struct smu_context *smu) +{ + int ret = 0; + uint32_t feature_mask[2]; + unsigned long feature_enabled; + ret = smu_feature_get_enabled_mask(smu, feature_mask, 2); + feature_enabled = (unsigned long)((uint64_t)feature_mask[0] | + ((uint64_t)feature_mask[1] << 32)); + return !!(feature_enabled & SMC_DPM_FEATURE); +} + +static int navi10_get_fan_speed_rpm(struct smu_context *smu, + uint32_t *speed) +{ + SmuMetrics_t metrics; + int ret = 0; + + if (!speed) + return -EINVAL; + + ret = navi10_get_metrics_table(smu, &metrics); + if (ret) + return ret; + + *speed = metrics.CurrFanSpeed; + + return ret; +} + +static int navi10_get_fan_speed_percent(struct smu_context *smu, + uint32_t *speed) +{ + int ret = 0; + uint32_t percent = 0; + uint32_t current_rpm; + PPTable_t *pptable = smu->smu_table.driver_pptable; + + ret = navi10_get_fan_speed_rpm(smu, ¤t_rpm); + if (ret) + return ret; + + percent = current_rpm * 100 / pptable->FanMaximumRpm; + *speed = percent > 100 ? 100 : percent; + + return ret; +} + +static int navi10_get_power_profile_mode(struct smu_context *smu, char *buf) +{ + DpmActivityMonitorCoeffInt_t activity_monitor; + uint32_t i, size = 0; + int16_t workload_type = 0; + static const char *profile_name[] = { + "BOOTUP_DEFAULT", + "3D_FULL_SCREEN", + "POWER_SAVING", + "VIDEO", + "VR", + "COMPUTE", + "CUSTOM"}; + static const char *title[] = { + "PROFILE_INDEX(NAME)", + "CLOCK_TYPE(NAME)", + "FPS", + "MinFreqType", + "MinActiveFreqType", + "MinActiveFreq", + "BoosterFreqType", + "BoosterFreq", + "PD_Data_limit_c", + "PD_Data_error_coeff", + "PD_Data_error_rate_coeff"}; + int result = 0; + + if (!buf) + return -EINVAL; + + 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]); + + for (i = 0; i <= PP_SMC_POWER_PROFILE_CUSTOM; i++) { + /* conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT */ + workload_type = smu_workload_get_type(smu, i); + if (workload_type < 0) + return -EINVAL; + + 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) ? "*" : " "); + + 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_MinFreqStep, + 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, + "SOCCLK", + activity_monitor.Soc_FPS, + activity_monitor.Soc_MinFreqStep, + activity_monitor.Soc_MinActiveFreqType, + activity_monitor.Soc_MinActiveFreq, + activity_monitor.Soc_BoosterFreqType, + activity_monitor.Soc_BoosterFreq, + activity_monitor.Soc_PD_Data_limit_c, + activity_monitor.Soc_PD_Data_error_coeff, + activity_monitor.Soc_PD_Data_error_rate_coeff); + + size += sprintf(buf + size, "%19s %d(%13s) %7d %7d %7d %7d %7d %7d %7d %7d %7d\n", + " ", + 2, + "MEMLK", + activity_monitor.Mem_FPS, + activity_monitor.Mem_MinFreqStep, + 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; +} + +static int navi10_set_power_profile_mode(struct smu_context *smu, long *input, uint32_t size) +{ + DpmActivityMonitorCoeffInt_t activity_monitor; + int workload_type, ret = 0; + + smu->power_profile_mode = input[size]; + + if (smu->power_profile_mode > PP_SMC_POWER_PROFILE_CUSTOM) { + pr_err("Invalid power profile mode %d\n", smu->power_profile_mode); + return -EINVAL; + } + + if (smu->power_profile_mode == PP_SMC_POWER_PROFILE_CUSTOM) { + if (size < 0) + return -EINVAL; + + 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_MinFreqStep = 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: /* Socclk */ + activity_monitor.Soc_FPS = input[1]; + activity_monitor.Soc_MinFreqStep = input[2]; + activity_monitor.Soc_MinActiveFreqType = input[3]; + activity_monitor.Soc_MinActiveFreq = input[4]; + activity_monitor.Soc_BoosterFreqType = input[5]; + activity_monitor.Soc_BoosterFreq = input[6]; + activity_monitor.Soc_PD_Data_limit_c = input[7]; + activity_monitor.Soc_PD_Data_error_coeff = input[8]; + activity_monitor.Soc_PD_Data_error_rate_coeff = input[9]; + break; + case 2: /* Memlk */ + activity_monitor.Mem_FPS = input[1]; + activity_monitor.Mem_MinFreqStep = 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 */ + workload_type = smu_workload_get_type(smu, smu->power_profile_mode); + if (workload_type < 0) + return -EINVAL; + smu_send_smc_msg_with_param(smu, SMU_MSG_SetWorkloadMask, + 1 << workload_type); + + return ret; +} + +static int navi10_get_profiling_clk_mask(struct smu_context *smu, + enum amd_dpm_forced_level level, + uint32_t *sclk_mask, + uint32_t *mclk_mask, + uint32_t *soc_mask) +{ + int ret = 0; + uint32_t level_count = 0; + + if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK) { + if (sclk_mask) + *sclk_mask = 0; + } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK) { + if (mclk_mask) + *mclk_mask = 0; + } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) { + if(sclk_mask) { + ret = smu_get_dpm_level_count(smu, SMU_SCLK, &level_count); + if (ret) + return ret; + *sclk_mask = level_count - 1; + } + + if(mclk_mask) { + ret = smu_get_dpm_level_count(smu, SMU_MCLK, &level_count); + if (ret) + return ret; + *mclk_mask = level_count - 1; + } + + if(soc_mask) { + ret = smu_get_dpm_level_count(smu, SMU_SOCCLK, &level_count); + if (ret) + return ret; + *soc_mask = level_count - 1; + } + } + + return ret; +} + +static int navi10_notify_smc_dispaly_config(struct smu_context *smu) +{ + struct smu_clocks min_clocks = {0}; + struct pp_display_clock_request clock_req; + int ret = 0; + + min_clocks.dcef_clock = smu->display_config->min_dcef_set_clk; + min_clocks.dcef_clock_in_sr = smu->display_config->min_dcef_deep_sleep_set_clk; + min_clocks.memory_clock = smu->display_config->min_mem_set_clock; + + if (smu_feature_is_supported(smu, SMU_FEATURE_DPM_DCEFCLK_BIT)) { + clock_req.clock_type = amd_pp_dcef_clock; + clock_req.clock_freq_in_khz = min_clocks.dcef_clock * 10; + + ret = smu_v11_0_display_clock_voltage_request(smu, &clock_req); + if (!ret) { + if (smu_feature_is_supported(smu, SMU_FEATURE_DS_DCEFCLK_BIT)) { + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetMinDeepSleepDcefclk, + min_clocks.dcef_clock_in_sr/100); + if (ret) { + pr_err("Attempt to set divider for DCEFCLK Failed!"); + return ret; + } + } + } else { + pr_info("Attempt to set Hard Min for DCEFCLK Failed!"); + } + } + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT)) { + ret = smu_set_hard_freq_range(smu, SMU_UCLK, min_clocks.memory_clock/100, 0); + if (ret) { + pr_err("[%s] Set hard min uclk failed!", __func__); + return ret; + } + } + + return 0; +} + +static int navi10_set_watermarks_table(struct smu_context *smu, + void *watermarks, struct + dm_pp_wm_sets_with_clock_ranges_soc15 + *clock_ranges) +{ + int i; + Watermarks_t *table = watermarks; + + if (!table || !clock_ranges) + return -EINVAL; + + if (clock_ranges->num_wm_dmif_sets > 4 || + clock_ranges->num_wm_mcif_sets > 4) + return -EINVAL; + + for (i = 0; i < clock_ranges->num_wm_dmif_sets; i++) { + table->WatermarkRow[1][i].MinClock = + cpu_to_le16((uint16_t) + (clock_ranges->wm_dmif_clocks_ranges[i].wm_min_dcfclk_clk_in_khz / + 1000)); + table->WatermarkRow[1][i].MaxClock = + cpu_to_le16((uint16_t) + (clock_ranges->wm_dmif_clocks_ranges[i].wm_max_dcfclk_clk_in_khz / + 1000)); + table->WatermarkRow[1][i].MinUclk = + cpu_to_le16((uint16_t) + (clock_ranges->wm_dmif_clocks_ranges[i].wm_min_mem_clk_in_khz / + 1000)); + table->WatermarkRow[1][i].MaxUclk = + cpu_to_le16((uint16_t) + (clock_ranges->wm_dmif_clocks_ranges[i].wm_max_mem_clk_in_khz / + 1000)); + table->WatermarkRow[1][i].WmSetting = (uint8_t) + clock_ranges->wm_dmif_clocks_ranges[i].wm_set_id; + } + + for (i = 0; i < clock_ranges->num_wm_mcif_sets; i++) { + table->WatermarkRow[0][i].MinClock = + cpu_to_le16((uint16_t) + (clock_ranges->wm_mcif_clocks_ranges[i].wm_min_socclk_clk_in_khz / + 1000)); + table->WatermarkRow[0][i].MaxClock = + cpu_to_le16((uint16_t) + (clock_ranges->wm_mcif_clocks_ranges[i].wm_max_socclk_clk_in_khz / + 1000)); + table->WatermarkRow[0][i].MinUclk = + cpu_to_le16((uint16_t) + (clock_ranges->wm_mcif_clocks_ranges[i].wm_min_mem_clk_in_khz / + 1000)); + table->WatermarkRow[0][i].MaxUclk = + cpu_to_le16((uint16_t) + (clock_ranges->wm_mcif_clocks_ranges[i].wm_max_mem_clk_in_khz / + 1000)); + table->WatermarkRow[0][i].WmSetting = (uint8_t) + clock_ranges->wm_mcif_clocks_ranges[i].wm_set_id; + } + + return 0; +} + +static int navi10_thermal_get_temperature(struct smu_context *smu, + enum amd_pp_sensors sensor, + uint32_t *value) +{ + SmuMetrics_t metrics; + int ret = 0; + + if (!value) + return -EINVAL; + + ret = navi10_get_metrics_table(smu, &metrics); + if (ret) + return ret; + + switch (sensor) { + case AMDGPU_PP_SENSOR_HOTSPOT_TEMP: + *value = metrics.TemperatureHotspot * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + break; + case AMDGPU_PP_SENSOR_EDGE_TEMP: + *value = metrics.TemperatureEdge * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + break; + case AMDGPU_PP_SENSOR_MEM_TEMP: + *value = metrics.TemperatureMem * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + break; + default: + pr_err("Invalid sensor for retrieving temp\n"); + return -EINVAL; + } + + return 0; +} + +static int navi10_read_sensor(struct smu_context *smu, + enum amd_pp_sensors sensor, + void *data, uint32_t *size) +{ + int ret = 0; + struct smu_table_context *table_context = &smu->smu_table; + PPTable_t *pptable = table_context->driver_pptable; + + if(!data || !size) + return -EINVAL; + + mutex_lock(&smu->sensor_lock); + switch (sensor) { + case AMDGPU_PP_SENSOR_MAX_FAN_RPM: + *(uint32_t *)data = pptable->FanMaximumRpm; + *size = 4; + break; + case AMDGPU_PP_SENSOR_MEM_LOAD: + case AMDGPU_PP_SENSOR_GPU_LOAD: + ret = navi10_get_current_activity_percent(smu, sensor, (uint32_t *)data); + *size = 4; + break; + case AMDGPU_PP_SENSOR_GPU_POWER: + ret = navi10_get_gpu_power(smu, (uint32_t *)data); + *size = 4; + break; + case AMDGPU_PP_SENSOR_HOTSPOT_TEMP: + case AMDGPU_PP_SENSOR_EDGE_TEMP: + case AMDGPU_PP_SENSOR_MEM_TEMP: + ret = navi10_thermal_get_temperature(smu, sensor, (uint32_t *)data); + *size = 4; + break; + default: + ret = smu_v11_0_read_sensor(smu, sensor, data, size); + } + mutex_unlock(&smu->sensor_lock); + + return ret; +} + +static int navi10_get_uclk_dpm_states(struct smu_context *smu, uint32_t *clocks_in_khz, uint32_t *num_states) +{ + uint32_t num_discrete_levels = 0; + uint16_t *dpm_levels = NULL; + uint16_t i = 0; + struct smu_table_context *table_context = &smu->smu_table; + PPTable_t *driver_ppt = NULL; + + if (!clocks_in_khz || !num_states || !table_context->driver_pptable) + return -EINVAL; + + driver_ppt = table_context->driver_pptable; + num_discrete_levels = driver_ppt->DpmDescriptor[PPCLK_UCLK].NumDiscreteLevels; + dpm_levels = driver_ppt->FreqTableUclk; + + if (num_discrete_levels == 0 || dpm_levels == NULL) + return -EINVAL; + + *num_states = num_discrete_levels; + for (i = 0; i < num_discrete_levels; i++) { + /* convert to khz */ + *clocks_in_khz = (*dpm_levels) * 1000; + clocks_in_khz++; + dpm_levels++; + } + + return 0; +} + +static int navi10_set_peak_clock_by_device(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: + switch (adev->pdev->revision) { + case 0xf0: /* XTX */ + case 0xc0: + sclk_freq = NAVI10_PEAK_SCLK_XTX; + break; + case 0xf1: /* XT */ + case 0xc1: + sclk_freq = NAVI10_PEAK_SCLK_XT; + break; + default: /* XL */ + sclk_freq = NAVI10_PEAK_SCLK_XL; + break; + } + break; + case CHIP_NAVI14: + switch (adev->pdev->revision) { + case 0xc7: /* XT */ + case 0xf4: + sclk_freq = NAVI14_UMD_PSTATE_PEAK_XT_GFXCLK; + break; + case 0xc1: /* XTM */ + case 0xf2: + sclk_freq = NAVI14_UMD_PSTATE_PEAK_XTM_GFXCLK; + break; + case 0xc3: /* XLM */ + case 0xf3: + sclk_freq = NAVI14_UMD_PSTATE_PEAK_XLM_GFXCLK; + break; + case 0xc5: /* XTX */ + case 0xf6: + sclk_freq = NAVI14_UMD_PSTATE_PEAK_XLM_GFXCLK; + break; + default: /* XL */ + sclk_freq = NAVI14_UMD_PSTATE_PEAK_XL_GFXCLK; + break; + } + break; + default: + return -EINVAL; + } + + 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); + if (ret) + return ret; + + 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_performance_level(struct smu_context *smu, enum amd_dpm_forced_level level) +{ + int ret = 0; + + switch (level) { + case AMD_DPM_FORCED_LEVEL_PROFILE_PEAK: + ret = navi10_set_peak_clock_by_device(smu); + break; + default: + ret = -EINVAL; + break; + } + + return ret; +} + +static int navi10_get_thermal_temperature_range(struct smu_context *smu, + struct smu_temperature_range *range) +{ + struct smu_table_context *table_context = &smu->smu_table; + struct smu_11_0_powerplay_table *powerplay_table = table_context->power_play_table; + + if (!range || !powerplay_table) + return -EINVAL; + + range->max = powerplay_table->software_shutdown_temp * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + + return 0; +} + +static int navi10_display_disable_memory_clock_switch(struct smu_context *smu, + bool disable_memory_clock_switch) +{ + int ret = 0; + struct smu_11_0_max_sustainable_clocks *max_sustainable_clocks = + (struct smu_11_0_max_sustainable_clocks *) + smu->smu_table.max_sustainable_clocks; + uint32_t min_memory_clock = smu->hard_min_uclk_req_from_dal; + uint32_t max_memory_clock = max_sustainable_clocks->uclock; + + if(smu->disable_uclk_switch == disable_memory_clock_switch) + return 0; + + if(disable_memory_clock_switch) + ret = smu_set_hard_freq_range(smu, SMU_UCLK, max_memory_clock, 0); + else + ret = smu_set_hard_freq_range(smu, SMU_UCLK, min_memory_clock, 0); + + if(!ret) + smu->disable_uclk_switch = disable_memory_clock_switch; + + return ret; +} + +static uint32_t navi10_get_pptable_power_limit(struct smu_context *smu) +{ + PPTable_t *pptable = smu->smu_table.driver_pptable; + return pptable->SocketPowerLimitAc[PPT_THROTTLER_PPT0]; +} + +static int navi10_get_power_limit(struct smu_context *smu, + uint32_t *limit, + bool cap) +{ + PPTable_t *pptable = smu->smu_table.driver_pptable; + uint32_t asic_default_power_limit = 0; + int ret = 0; + int power_src; + + if (!smu->power_limit) { + if (smu_feature_is_enabled(smu, SMU_FEATURE_PPT_BIT)) { + power_src = smu_power_get_index(smu, SMU_POWER_SOURCE_AC); + if (power_src < 0) + return -EINVAL; + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_GetPptLimit, + power_src << 16); + if (ret) { + pr_err("[%s] get PPT limit failed!", __func__); + return ret; + } + smu_read_smc_arg(smu, &asic_default_power_limit); + } else { + /* the last hope to figure out the ppt limit */ + if (!pptable) { + pr_err("Cannot get PPT limit due to pptable missing!"); + return -EINVAL; + } + asic_default_power_limit = + pptable->SocketPowerLimitAc[PPT_THROTTLER_PPT0]; + } + + smu->power_limit = asic_default_power_limit; + } + + if (cap) + *limit = smu_v11_0_get_max_power_limit(smu); + else + *limit = smu->power_limit; + + return 0; +} + +static int navi10_update_pcie_parameters(struct smu_context *smu, + uint32_t pcie_gen_cap, + uint32_t pcie_width_cap) +{ + PPTable_t *pptable = smu->smu_table.driver_pptable; + int ret, i; + uint32_t smu_pcie_arg; + + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + struct smu_11_0_dpm_context *dpm_context = smu_dpm->dpm_context; + + for (i = 0; i < NUM_LINK_LEVELS; i++) { + smu_pcie_arg = (i << 16) | + ((pptable->PcieGenSpeed[i] <= pcie_gen_cap) ? (pptable->PcieGenSpeed[i] << 8) : + (pcie_gen_cap << 8)) | ((pptable->PcieLaneCount[i] <= pcie_width_cap) ? + pptable->PcieLaneCount[i] : pcie_width_cap); + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_OverridePcieParameters, + smu_pcie_arg); + + if (ret) + return ret; + + if (pptable->PcieGenSpeed[i] > pcie_gen_cap) + dpm_context->dpm_tables.pcie_table.pcie_gen[i] = pcie_gen_cap; + if (pptable->PcieLaneCount[i] > pcie_width_cap) + dpm_context->dpm_tables.pcie_table.pcie_lane[i] = pcie_width_cap; + } + + return 0; +} + +static inline void navi10_dump_od_table(OverDriveTable_t *od_table) { + pr_debug("OD: Gfxclk: (%d, %d)\n", od_table->GfxclkFmin, od_table->GfxclkFmax); + pr_debug("OD: Gfx1: (%d, %d)\n", od_table->GfxclkFreq1, od_table->GfxclkVolt1); + pr_debug("OD: Gfx2: (%d, %d)\n", od_table->GfxclkFreq2, od_table->GfxclkVolt2); + pr_debug("OD: Gfx3: (%d, %d)\n", od_table->GfxclkFreq3, od_table->GfxclkVolt3); + pr_debug("OD: UclkFmax: %d\n", od_table->UclkFmax); + pr_debug("OD: OverDrivePct: %d\n", od_table->OverDrivePct); +} + +static int navi10_od_setting_check_range(struct smu_11_0_overdrive_table *od_table, enum SMU_11_0_ODSETTING_ID setting, uint32_t value) +{ + if (value < od_table->min[setting]) { + pr_warn("OD setting (%d, %d) is less than the minimum allowed (%d)\n", setting, value, od_table->min[setting]); + return -EINVAL; + } + if (value > od_table->max[setting]) { + pr_warn("OD setting (%d, %d) is greater than the maximum allowed (%d)\n", setting, value, od_table->max[setting]); + return -EINVAL; + } + return 0; +} + +static int navi10_setup_od_limits(struct smu_context *smu) { + struct smu_11_0_overdrive_table *overdrive_table = NULL; + struct smu_11_0_powerplay_table *powerplay_table = NULL; + + if (!smu->smu_table.power_play_table) { + pr_err("powerplay table uninitialized!\n"); + return -ENOENT; + } + powerplay_table = (struct smu_11_0_powerplay_table *)smu->smu_table.power_play_table; + overdrive_table = &powerplay_table->overdrive_table; + if (!smu->od_settings) { + smu->od_settings = kmemdup(overdrive_table, sizeof(struct smu_11_0_overdrive_table), GFP_KERNEL); + } else { + memcpy(smu->od_settings, overdrive_table, sizeof(struct smu_11_0_overdrive_table)); + } + return 0; +} + +static int navi10_set_default_od_settings(struct smu_context *smu, bool initialize) { + OverDriveTable_t *od_table; + int ret = 0; + + ret = smu_v11_0_set_default_od_settings(smu, initialize, sizeof(OverDriveTable_t)); + if (ret) + return ret; + + if (initialize) { + ret = navi10_setup_od_limits(smu); + if (ret) { + pr_err("Failed to retrieve board OD limits\n"); + return ret; + } + + } + + od_table = (OverDriveTable_t *)smu->smu_table.overdrive_table; + if (od_table) { + navi10_dump_od_table(od_table); + } + + return ret; +} + +static int navi10_od_edit_dpm_table(struct smu_context *smu, enum PP_OD_DPM_TABLE_COMMAND type, long input[], uint32_t size) { + int i; + int ret = 0; + struct smu_table_context *table_context = &smu->smu_table; + OverDriveTable_t *od_table; + struct smu_11_0_overdrive_table *od_settings; + enum SMU_11_0_ODSETTING_ID freq_setting, voltage_setting; + uint16_t *freq_ptr, *voltage_ptr; + od_table = (OverDriveTable_t *)table_context->overdrive_table; + + if (!smu->od_enabled) { + pr_warn("OverDrive is not enabled!\n"); + return -EINVAL; + } + + if (!smu->od_settings) { + pr_err("OD board limits are not set!\n"); + return -ENOENT; + } + + od_settings = smu->od_settings; + + switch (type) { + case PP_OD_EDIT_SCLK_VDDC_TABLE: + if (!navi10_od_feature_is_supported(od_settings, SMU_11_0_ODFEATURE_GFXCLK_LIMITS)) { + pr_warn("GFXCLK_LIMITS not supported!\n"); + return -ENOTSUPP; + } + if (!table_context->overdrive_table) { + pr_err("Overdrive is not initialized\n"); + return -EINVAL; + } + for (i = 0; i < size; i += 2) { + if (i + 2 > size) { + pr_info("invalid number of input parameters %d\n", size); + return -EINVAL; + } + switch (input[i]) { + case 0: + freq_setting = SMU_11_0_ODSETTING_GFXCLKFMIN; + freq_ptr = &od_table->GfxclkFmin; + if (input[i + 1] > od_table->GfxclkFmax) { + pr_info("GfxclkFmin (%ld) must be <= GfxclkFmax (%u)!\n", + input[i + 1], + od_table->GfxclkFmin); + return -EINVAL; + } + break; + case 1: + freq_setting = SMU_11_0_ODSETTING_GFXCLKFMAX; + freq_ptr = &od_table->GfxclkFmax; + if (input[i + 1] < od_table->GfxclkFmin) { + pr_info("GfxclkFmax (%ld) must be >= GfxclkFmin (%u)!\n", + input[i + 1], + od_table->GfxclkFmax); + return -EINVAL; + } + break; + default: + pr_info("Invalid SCLK_VDDC_TABLE index: %ld\n", input[i]); + pr_info("Supported indices: [0:min,1:max]\n"); + return -EINVAL; + } + ret = navi10_od_setting_check_range(od_settings, freq_setting, input[i + 1]); + if (ret) + return ret; + *freq_ptr = input[i + 1]; + } + break; + case PP_OD_EDIT_MCLK_VDDC_TABLE: + if (!navi10_od_feature_is_supported(od_settings, SMU_11_0_ODFEATURE_UCLK_MAX)) { + pr_warn("UCLK_MAX not supported!\n"); + return -ENOTSUPP; + } + if (size < 2) { + pr_info("invalid number of parameters: %d\n", size); + return -EINVAL; + } + if (input[0] != 1) { + pr_info("Invalid MCLK_VDDC_TABLE index: %ld\n", input[0]); + pr_info("Supported indices: [1:max]\n"); + return -EINVAL; + } + ret = navi10_od_setting_check_range(od_settings, SMU_11_0_ODSETTING_UCLKFMAX, input[1]); + if (ret) + return ret; + od_table->UclkFmax = input[1]; + break; + case PP_OD_COMMIT_DPM_TABLE: + navi10_dump_od_table(od_table); + ret = smu_update_table(smu, SMU_TABLE_OVERDRIVE, 0, (void *)od_table, true); + if (ret) { + pr_err("Failed to import overdrive table!\n"); + return ret; + } + // no lock needed because smu_od_edit_dpm_table has it + ret = smu_handle_task(smu, smu->smu_dpm.dpm_level, + AMD_PP_TASK_READJUST_POWER_STATE, + false); + if (ret) { + return ret; + } + break; + case PP_OD_EDIT_VDDC_CURVE: + if (!navi10_od_feature_is_supported(od_settings, SMU_11_0_ODFEATURE_GFXCLK_CURVE)) { + pr_warn("GFXCLK_CURVE not supported!\n"); + return -ENOTSUPP; + } + if (size < 3) { + pr_info("invalid number of parameters: %d\n", size); + return -EINVAL; + } + if (!od_table) { + pr_info("Overdrive is not initialized\n"); + return -EINVAL; + } + + switch (input[0]) { + case 0: + freq_setting = SMU_11_0_ODSETTING_VDDGFXCURVEFREQ_P1; + voltage_setting = SMU_11_0_ODSETTING_VDDGFXCURVEVOLTAGE_P1; + freq_ptr = &od_table->GfxclkFreq1; + voltage_ptr = &od_table->GfxclkVolt1; + break; + case 1: + freq_setting = SMU_11_0_ODSETTING_VDDGFXCURVEFREQ_P2; + voltage_setting = SMU_11_0_ODSETTING_VDDGFXCURVEVOLTAGE_P2; + freq_ptr = &od_table->GfxclkFreq2; + voltage_ptr = &od_table->GfxclkVolt2; + break; + case 2: + freq_setting = SMU_11_0_ODSETTING_VDDGFXCURVEFREQ_P3; + voltage_setting = SMU_11_0_ODSETTING_VDDGFXCURVEVOLTAGE_P3; + freq_ptr = &od_table->GfxclkFreq3; + voltage_ptr = &od_table->GfxclkVolt3; + break; + default: + pr_info("Invalid VDDC_CURVE index: %ld\n", input[0]); + pr_info("Supported indices: [0, 1, 2]\n"); + return -EINVAL; + } + ret = navi10_od_setting_check_range(od_settings, freq_setting, input[1]); + if (ret) + return ret; + // Allow setting zero to disable the OverDrive VDDC curve + if (input[2] != 0) { + ret = navi10_od_setting_check_range(od_settings, voltage_setting, input[2]); + if (ret) + return ret; + *freq_ptr = input[1]; + *voltage_ptr = ((uint16_t)input[2]) * NAVI10_VOLTAGE_SCALE; + pr_debug("OD: set curve %ld: (%d, %d)\n", input[0], *freq_ptr, *voltage_ptr); + } else { + // If setting 0, disable all voltage curve settings + od_table->GfxclkVolt1 = 0; + od_table->GfxclkVolt2 = 0; + od_table->GfxclkVolt3 = 0; + } + navi10_dump_od_table(od_table); + break; + default: + return -ENOSYS; + } + return ret; +} + +static int navi10_run_btc(struct smu_context *smu) +{ + int ret = 0; + + ret = smu_send_smc_msg(smu, SMU_MSG_RunBtc); + if (ret) + pr_err("RunBtc failed!\n"); + + return ret; +} + +static const struct pptable_funcs navi10_ppt_funcs = { + .tables_init = navi10_tables_init, + .alloc_dpm_context = navi10_allocate_dpm_context, + .store_powerplay_table = navi10_store_powerplay_table, + .check_powerplay_table = navi10_check_powerplay_table, + .append_powerplay_table = navi10_append_powerplay_table, + .get_smu_msg_index = navi10_get_smu_msg_index, + .get_smu_clk_index = navi10_get_smu_clk_index, + .get_smu_feature_index = navi10_get_smu_feature_index, + .get_smu_table_index = navi10_get_smu_table_index, + .get_smu_power_index = navi10_get_pwr_src_index, + .get_workload_type = navi10_get_workload_type, + .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, + .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, + .populate_umd_state_clk = navi10_populate_umd_state_clk, + .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, + .force_dpm_limit_value = navi10_force_dpm_limit_value, + .unforce_dpm_levels = navi10_unforce_dpm_levels, + .is_dpm_running = navi10_is_dpm_running, + .get_fan_speed_percent = navi10_get_fan_speed_percent, + .get_fan_speed_rpm = navi10_get_fan_speed_rpm, + .get_power_profile_mode = navi10_get_power_profile_mode, + .set_power_profile_mode = navi10_set_power_profile_mode, + .get_profiling_clk_mask = navi10_get_profiling_clk_mask, + .set_watermarks_table = navi10_set_watermarks_table, + .read_sensor = navi10_read_sensor, + .get_uclk_dpm_states = navi10_get_uclk_dpm_states, + .set_performance_level = navi10_set_performance_level, + .get_thermal_temperature_range = navi10_get_thermal_temperature_range, + .display_disable_memory_clock_switch = navi10_display_disable_memory_clock_switch, + .get_power_limit = navi10_get_power_limit, + .update_pcie_parameters = navi10_update_pcie_parameters, + .init_microcode = smu_v11_0_init_microcode, + .load_microcode = smu_v11_0_load_microcode, + .init_smc_tables = smu_v11_0_init_smc_tables, + .fini_smc_tables = smu_v11_0_fini_smc_tables, + .init_power = smu_v11_0_init_power, + .fini_power = smu_v11_0_fini_power, + .check_fw_status = smu_v11_0_check_fw_status, + .setup_pptable = smu_v11_0_setup_pptable, + .get_vbios_bootup_values = smu_v11_0_get_vbios_bootup_values, + .get_clk_info_from_vbios = smu_v11_0_get_clk_info_from_vbios, + .check_pptable = smu_v11_0_check_pptable, + .parse_pptable = smu_v11_0_parse_pptable, + .populate_smc_tables = smu_v11_0_populate_smc_pptable, + .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_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, + .send_smc_msg = smu_v11_0_send_msg, + .send_smc_msg_with_param = smu_v11_0_send_msg_with_param, + .read_smc_arg = smu_v11_0_read_arg, + .init_display_count = smu_v11_0_init_display_count, + .set_allowed_mask = smu_v11_0_set_allowed_mask, + .get_enabled_mask = smu_v11_0_get_enabled_mask, + .notify_display_change = smu_v11_0_notify_display_change, + .set_power_limit = smu_v11_0_set_power_limit, + .get_current_clk_freq = smu_v11_0_get_current_clk_freq, + .init_max_sustainable_clocks = smu_v11_0_init_max_sustainable_clocks, + .start_thermal_control = smu_v11_0_start_thermal_control, + .stop_thermal_control = smu_v11_0_stop_thermal_control, + .set_deep_sleep_dcefclk = smu_v11_0_set_deep_sleep_dcefclk, + .display_clock_voltage_request = smu_v11_0_display_clock_voltage_request, + .get_fan_control_mode = smu_v11_0_get_fan_control_mode, + .set_fan_control_mode = smu_v11_0_set_fan_control_mode, + .set_fan_speed_percent = smu_v11_0_set_fan_speed_percent, + .set_fan_speed_rpm = smu_v11_0_set_fan_speed_rpm, + .set_xgmi_pstate = smu_v11_0_set_xgmi_pstate, + .gfx_off_control = smu_v11_0_gfx_off_control, + .register_irq_handler = smu_v11_0_register_irq_handler, + .set_azalia_d3_pme = smu_v11_0_set_azalia_d3_pme, + .get_max_sustainable_clocks_by_dc = smu_v11_0_get_max_sustainable_clocks_by_dc, + .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, + .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, + .set_default_od_settings = navi10_set_default_od_settings, + .od_edit_dpm_table = navi10_od_edit_dpm_table, + .get_pptable_power_limit = navi10_get_pptable_power_limit, + .run_btc = navi10_run_btc, +}; + +void navi10_set_ppt_funcs(struct smu_context *smu) +{ + smu->ppt_funcs = &navi10_ppt_funcs; +} diff --git a/drivers/gpu/drm/amd/powerplay/navi10_ppt.h b/drivers/gpu/drm/amd/powerplay/navi10_ppt.h new file mode 100644 index 000000000000..ec03c7992f6d --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/navi10_ppt.h @@ -0,0 +1,43 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __NAVI10_PPT_H__ +#define __NAVI10_PPT_H__ + +#define NAVI10_PEAK_SCLK_XTX (1830) +#define NAVI10_PEAK_SCLK_XT (1755) +#define NAVI10_PEAK_SCLK_XL (1625) + +#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 NAVI10_VOLTAGE_SCALE (4) + +#define smnPCIE_LC_SPEED_CNTL 0x11140290 +#define smnPCIE_LC_LINK_WIDTH_CNTL 0x11140288 + +extern void navi10_set_ppt_funcs(struct smu_context *smu); + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/renoir_ppt.c b/drivers/gpu/drm/amd/powerplay/renoir_ppt.c new file mode 100644 index 000000000000..04daf7e9fe05 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/renoir_ppt.c @@ -0,0 +1,718 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "amdgpu.h" +#include "amdgpu_smu.h" +#include "smu_internal.h" +#include "soc15_common.h" +#include "smu_v12_0_ppsmc.h" +#include "smu12_driver_if.h" +#include "smu_v12_0.h" +#include "renoir_ppt.h" + + +#define MSG_MAP(msg, index) \ + [SMU_MSG_##msg] = {1, (index)} + +#define TAB_MAP_VALID(tab) \ + [SMU_TABLE_##tab] = {1, TABLE_##tab} + +#define TAB_MAP_INVALID(tab) \ + [SMU_TABLE_##tab] = {0, TABLE_##tab} + +static struct smu_12_0_cmn2aisc_mapping renoir_message_map[SMU_MSG_MAX_COUNT] = { + MSG_MAP(TestMessage, PPSMC_MSG_TestMessage), + MSG_MAP(GetSmuVersion, PPSMC_MSG_GetSmuVersion), + MSG_MAP(GetDriverIfVersion, PPSMC_MSG_GetDriverIfVersion), + MSG_MAP(PowerUpGfx, PPSMC_MSG_PowerUpGfx), + MSG_MAP(AllowGfxOff, PPSMC_MSG_EnableGfxOff), + MSG_MAP(DisallowGfxOff, PPSMC_MSG_DisableGfxOff), + MSG_MAP(PowerDownIspByTile, PPSMC_MSG_PowerDownIspByTile), + MSG_MAP(PowerUpIspByTile, PPSMC_MSG_PowerUpIspByTile), + MSG_MAP(PowerDownVcn, PPSMC_MSG_PowerDownVcn), + MSG_MAP(PowerUpVcn, PPSMC_MSG_PowerUpVcn), + MSG_MAP(PowerDownSdma, PPSMC_MSG_PowerDownSdma), + MSG_MAP(PowerUpSdma, PPSMC_MSG_PowerUpSdma), + MSG_MAP(SetHardMinIspclkByFreq, PPSMC_MSG_SetHardMinIspclkByFreq), + MSG_MAP(SetHardMinVcn, PPSMC_MSG_SetHardMinVcn), + MSG_MAP(Spare1, PPSMC_MSG_spare1), + MSG_MAP(Spare2, PPSMC_MSG_spare2), + MSG_MAP(SetAllowFclkSwitch, PPSMC_MSG_SetAllowFclkSwitch), + MSG_MAP(SetMinVideoGfxclkFreq, PPSMC_MSG_SetMinVideoGfxclkFreq), + MSG_MAP(ActiveProcessNotify, PPSMC_MSG_ActiveProcessNotify), + MSG_MAP(SetCustomPolicy, PPSMC_MSG_SetCustomPolicy), + MSG_MAP(SetVideoFps, PPSMC_MSG_SetVideoFps), + MSG_MAP(NumOfDisplays, PPSMC_MSG_SetDisplayCount), + MSG_MAP(QueryPowerLimit, PPSMC_MSG_QueryPowerLimit), + MSG_MAP(SetDriverDramAddrHigh, PPSMC_MSG_SetDriverDramAddrHigh), + MSG_MAP(SetDriverDramAddrLow, PPSMC_MSG_SetDriverDramAddrLow), + MSG_MAP(TransferTableSmu2Dram, PPSMC_MSG_TransferTableSmu2Dram), + MSG_MAP(TransferTableDram2Smu, PPSMC_MSG_TransferTableDram2Smu), + MSG_MAP(GfxDeviceDriverReset, PPSMC_MSG_GfxDeviceDriverReset), + MSG_MAP(SetGfxclkOverdriveByFreqVid, PPSMC_MSG_SetGfxclkOverdriveByFreqVid), + MSG_MAP(SetHardMinDcfclkByFreq, PPSMC_MSG_SetHardMinDcfclkByFreq), + MSG_MAP(SetHardMinSocclkByFreq, PPSMC_MSG_SetHardMinSocclkByFreq), + MSG_MAP(ControlIgpuATS, PPSMC_MSG_ControlIgpuATS), + MSG_MAP(SetMinVideoFclkFreq, PPSMC_MSG_SetMinVideoFclkFreq), + MSG_MAP(SetMinDeepSleepDcfclk, PPSMC_MSG_SetMinDeepSleepDcfclk), + MSG_MAP(ForcePowerDownGfx, PPSMC_MSG_ForcePowerDownGfx), + MSG_MAP(SetPhyclkVoltageByFreq, PPSMC_MSG_SetPhyclkVoltageByFreq), + MSG_MAP(SetDppclkVoltageByFreq, PPSMC_MSG_SetDppclkVoltageByFreq), + MSG_MAP(SetSoftMinVcn, PPSMC_MSG_SetSoftMinVcn), + MSG_MAP(EnablePostCode, PPSMC_MSG_EnablePostCode), + MSG_MAP(GetGfxclkFrequency, PPSMC_MSG_GetGfxclkFrequency), + MSG_MAP(GetFclkFrequency, PPSMC_MSG_GetFclkFrequency), + MSG_MAP(GetMinGfxclkFrequency, PPSMC_MSG_GetMinGfxclkFrequency), + MSG_MAP(GetMaxGfxclkFrequency, PPSMC_MSG_GetMaxGfxclkFrequency), + MSG_MAP(SoftReset, PPSMC_MSG_SoftReset), + MSG_MAP(SetGfxCGPG, PPSMC_MSG_SetGfxCGPG), + MSG_MAP(SetSoftMaxGfxClk, PPSMC_MSG_SetSoftMaxGfxClk), + MSG_MAP(SetHardMinGfxClk, PPSMC_MSG_SetHardMinGfxClk), + MSG_MAP(SetSoftMaxSocclkByFreq, PPSMC_MSG_SetSoftMaxSocclkByFreq), + MSG_MAP(SetSoftMaxFclkByFreq, PPSMC_MSG_SetSoftMaxFclkByFreq), + MSG_MAP(SetSoftMaxVcn, PPSMC_MSG_SetSoftMaxVcn), + MSG_MAP(PowerGateMmHub, PPSMC_MSG_PowerGateMmHub), + MSG_MAP(UpdatePmeRestore, PPSMC_MSG_UpdatePmeRestore), + MSG_MAP(GpuChangeState, PPSMC_MSG_GpuChangeState), + MSG_MAP(SetPowerLimitPercentage, PPSMC_MSG_SetPowerLimitPercentage), + MSG_MAP(ForceGfxContentSave, PPSMC_MSG_ForceGfxContentSave), + MSG_MAP(EnableTmdp48MHzRefclkPwrDown, PPSMC_MSG_EnableTmdp48MHzRefclkPwrDown), + MSG_MAP(PowerDownJpeg, PPSMC_MSG_PowerDownJpeg), + MSG_MAP(PowerUpJpeg, PPSMC_MSG_PowerUpJpeg), + MSG_MAP(PowerGateAtHub, PPSMC_MSG_PowerGateAtHub), + MSG_MAP(SetSoftMinJpeg, PPSMC_MSG_SetSoftMinJpeg), + MSG_MAP(SetHardMinFclkByFreq, PPSMC_MSG_SetHardMinFclkByFreq), +}; + +static struct smu_12_0_cmn2aisc_mapping renoir_table_map[SMU_TABLE_COUNT] = { + TAB_MAP_VALID(WATERMARKS), + TAB_MAP_INVALID(CUSTOM_DPM), + TAB_MAP_VALID(DPMCLOCKS), + TAB_MAP_VALID(SMU_METRICS), +}; + +static int renoir_get_smu_msg_index(struct smu_context *smc, uint32_t index) +{ + struct smu_12_0_cmn2aisc_mapping mapping; + + if (index >= SMU_MSG_MAX_COUNT) + return -EINVAL; + + mapping = renoir_message_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; + + if (index >= SMU_TABLE_COUNT) + return -EINVAL; + + mapping = renoir_table_map[index]; + if (!(mapping.valid_mapping)) + return -EINVAL; + + return mapping.map_to; +} + +static int renoir_tables_init(struct smu_context *smu, struct smu_table *tables) +{ + struct smu_table_context *smu_table = &smu->smu_table; + + SMU_TABLE_INIT(tables, SMU_TABLE_WATERMARKS, sizeof(Watermarks_t), + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + SMU_TABLE_INIT(tables, SMU_TABLE_DPMCLOCKS, sizeof(DpmClocks_t), + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + SMU_TABLE_INIT(tables, SMU_TABLE_SMU_METRICS, sizeof(SmuMetrics_t), + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + + smu_table->clocks_table = kzalloc(sizeof(DpmClocks_t), GFP_KERNEL); + if (!smu_table->clocks_table) + return -ENOMEM; + + return 0; +} + +/** + * This interface just for getting uclk ultimate freq and should't introduce + * other likewise function result in overmuch callback. + */ +static int renoir_get_dpm_clk_limited(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t dpm_level, uint32_t *freq) +{ + DpmClocks_t *clk_table = smu->smu_table.clocks_table; + + if (!clk_table || clk_type >= SMU_CLK_COUNT) + return -EINVAL; + + GET_DPM_CUR_FREQ(clk_table, clk_type, dpm_level, *freq); + + return 0; +} + +static int renoir_print_clk_levels(struct smu_context *smu, + enum smu_clk_type clk_type, char *buf) +{ + int i, size = 0, ret = 0; + uint32_t cur_value = 0, value = 0, count = 0, min = 0, max = 0; + DpmClocks_t *clk_table = smu->smu_table.clocks_table; + SmuMetrics_t metrics; + + if (!clk_table || clk_type >= SMU_CLK_COUNT) + return -EINVAL; + + memset(&metrics, 0, sizeof(metrics)); + + ret = smu_update_table(smu, SMU_TABLE_SMU_METRICS, 0, + (void *)&metrics, false); + if (ret) + return ret; + + switch (clk_type) { + case SMU_GFXCLK: + case SMU_SCLK: + /* retirve table returned paramters unit is MHz */ + cur_value = metrics.ClockFrequency[CLOCK_GFXCLK]; + ret = smu_get_dpm_freq_range(smu, SMU_GFXCLK, &min, &max, false); + if (!ret) { + /* driver only know min/max gfx_clk, Add level 1 for all other gfx clks */ + if (cur_value == max) + i = 2; + else if (cur_value == min) + i = 0; + else + i = 1; + + size += sprintf(buf + size, "0: %uMhz %s\n", min, + i == 0 ? "*" : ""); + size += sprintf(buf + size, "1: %uMhz %s\n", + i == 1 ? cur_value : RENOIR_UMD_PSTATE_GFXCLK, + i == 1 ? "*" : ""); + size += sprintf(buf + size, "2: %uMhz %s\n", max, + i == 2 ? "*" : ""); + } + return size; + case SMU_SOCCLK: + count = NUM_SOCCLK_DPM_LEVELS; + cur_value = metrics.ClockFrequency[CLOCK_SOCCLK]; + break; + case SMU_MCLK: + count = NUM_MEMCLK_DPM_LEVELS; + cur_value = metrics.ClockFrequency[CLOCK_UMCCLK]; + break; + case SMU_DCEFCLK: + count = NUM_DCFCLK_DPM_LEVELS; + cur_value = metrics.ClockFrequency[CLOCK_DCFCLK]; + break; + case SMU_FCLK: + count = NUM_FCLK_DPM_LEVELS; + cur_value = metrics.ClockFrequency[CLOCK_FCLK]; + break; + default: + return -EINVAL; + } + + for (i = 0; i < count; i++) { + GET_DPM_CUR_FREQ(clk_table, clk_type, i, value); + size += sprintf(buf + size, "%d: %uMhz %s\n", i, value, + cur_value == value ? "*" : ""); + } + + return size; +} + +static enum amd_pm_state_type renoir_get_current_power_state(struct smu_context *smu) +{ + enum amd_pm_state_type pm_type; + struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm); + + if (!smu_dpm_ctx->dpm_context || + !smu_dpm_ctx->dpm_current_power_state) + return -EINVAL; + + switch (smu_dpm_ctx->dpm_current_power_state->classification.ui_label) { + case SMU_STATE_UI_LABEL_BATTERY: + pm_type = POWER_STATE_TYPE_BATTERY; + break; + case SMU_STATE_UI_LABEL_BALLANCED: + pm_type = POWER_STATE_TYPE_BALANCED; + break; + case SMU_STATE_UI_LABEL_PERFORMANCE: + pm_type = POWER_STATE_TYPE_PERFORMANCE; + break; + default: + if (smu_dpm_ctx->dpm_current_power_state->classification.flags & SMU_STATE_CLASSIFICATION_FLAG_BOOT) + pm_type = POWER_STATE_TYPE_INTERNAL_BOOT; + else + pm_type = POWER_STATE_TYPE_DEFAULT; + break; + } + + return pm_type; +} + +static int renoir_dpm_set_uvd_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) { + /* vcn dpm on is a prerequisite for vcn power gate messages */ + if (smu_feature_is_enabled(smu, SMU_FEATURE_VCN_PG_BIT)) { + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_PowerUpVcn, 0); + if (ret) + return ret; + } + power_gate->vcn_gated = false; + } else { + if (smu_feature_is_enabled(smu, SMU_FEATURE_VCN_PG_BIT)) { + ret = smu_send_smc_msg(smu, SMU_MSG_PowerDownVcn); + if (ret) + return ret; + } + power_gate->vcn_gated = true; + } + + return ret; +} + +static int renoir_force_dpm_limit_value(struct smu_context *smu, bool highest) +{ + int ret = 0, i = 0; + uint32_t min_freq, max_freq, force_freq; + enum smu_clk_type clk_type; + + enum smu_clk_type clks[] = { + SMU_GFXCLK, + SMU_MCLK, + SMU_SOCCLK, + }; + + for (i = 0; i < ARRAY_SIZE(clks); i++) { + clk_type = clks[i]; + ret = smu_get_dpm_freq_range(smu, clk_type, &min_freq, &max_freq, false); + if (ret) + return ret; + + force_freq = highest ? max_freq : min_freq; + ret = smu_set_soft_freq_range(smu, clk_type, force_freq, force_freq); + if (ret) + return ret; + } + + return ret; +} + +static int renoir_unforce_dpm_levels(struct smu_context *smu) { + + int ret = 0, i = 0; + uint32_t min_freq, max_freq; + enum smu_clk_type clk_type; + + struct clk_feature_map { + enum smu_clk_type clk_type; + uint32_t feature; + } clk_feature_map[] = { + {SMU_GFXCLK, SMU_FEATURE_DPM_GFXCLK_BIT}, + {SMU_MCLK, SMU_FEATURE_DPM_UCLK_BIT}, + {SMU_SOCCLK, SMU_FEATURE_DPM_SOCCLK_BIT}, + }; + + for (i = 0; i < ARRAY_SIZE(clk_feature_map); i++) { + if (!smu_feature_is_enabled(smu, clk_feature_map[i].feature)) + continue; + + clk_type = clk_feature_map[i].clk_type; + + ret = smu_get_dpm_freq_range(smu, clk_type, &min_freq, &max_freq, false); + if (ret) + return ret; + + ret = smu_set_soft_freq_range(smu, clk_type, min_freq, max_freq); + if (ret) + return ret; + } + + return ret; +} + +static int renoir_get_workload_type(struct smu_context *smu, uint32_t profile) +{ + + uint32_t pplib_workload = 0; + + switch (profile) { + case PP_SMC_POWER_PROFILE_FULLSCREEN3D: + pplib_workload = WORKLOAD_PPLIB_FULL_SCREEN_3D_BIT; + break; + case PP_SMC_POWER_PROFILE_CUSTOM: + pplib_workload = WORKLOAD_PPLIB_COUNT; + break; + case PP_SMC_POWER_PROFILE_VIDEO: + pplib_workload = WORKLOAD_PPLIB_VIDEO_BIT; + break; + case PP_SMC_POWER_PROFILE_VR: + pplib_workload = WORKLOAD_PPLIB_VR_BIT; + break; + case PP_SMC_POWER_PROFILE_COMPUTE: + pplib_workload = WORKLOAD_PPLIB_COMPUTE_BIT; + break; + default: + return -EINVAL; + } + + return pplib_workload; +} + +static int renoir_get_profiling_clk_mask(struct smu_context *smu, + enum amd_dpm_forced_level level, + uint32_t *sclk_mask, + uint32_t *mclk_mask, + uint32_t *soc_mask) +{ + + if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK) { + if (sclk_mask) + *sclk_mask = 0; + } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK) { + if (mclk_mask) + *mclk_mask = 0; + } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) { + if(sclk_mask) + /* The sclk as gfxclk and has three level about max/min/current */ + *sclk_mask = 3 - 1; + + if(mclk_mask) + *mclk_mask = NUM_MEMCLK_DPM_LEVELS - 1; + + if(soc_mask) + *soc_mask = NUM_SOCCLK_DPM_LEVELS - 1; + } + + return 0; +} + +/** + * This interface get dpm clock table for dc + */ +static int renoir_get_dpm_clock_table(struct smu_context *smu, struct dpm_clocks *clock_table) +{ + DpmClocks_t *table = smu->smu_table.clocks_table; + int i; + + if (!clock_table || !table) + return -EINVAL; + + for (i = 0; i < NUM_DCFCLK_DPM_LEVELS; i++) { + clock_table->DcfClocks[i].Freq = table->DcfClocks[i].Freq; + clock_table->DcfClocks[i].Vol = table->DcfClocks[i].Vol; + } + + for (i = 0; i < NUM_SOCCLK_DPM_LEVELS; i++) { + clock_table->SocClocks[i].Freq = table->SocClocks[i].Freq; + clock_table->SocClocks[i].Vol = table->SocClocks[i].Vol; + } + + for (i = 0; i < NUM_FCLK_DPM_LEVELS; i++) { + clock_table->FClocks[i].Freq = table->FClocks[i].Freq; + clock_table->FClocks[i].Vol = table->FClocks[i].Vol; + } + + for (i = 0; i< NUM_MEMCLK_DPM_LEVELS; i++) { + clock_table->MemClocks[i].Freq = table->MemClocks[i].Freq; + clock_table->MemClocks[i].Vol = table->MemClocks[i].Vol; + } + + return 0; +} + +static int renoir_force_clk_levels(struct smu_context *smu, + enum smu_clk_type clk_type, uint32_t mask) +{ + + int ret = 0 ; + uint32_t soft_min_level = 0, soft_max_level = 0, min_freq = 0, max_freq = 0; + DpmClocks_t *clk_table = smu->smu_table.clocks_table; + + soft_min_level = mask ? (ffs(mask) - 1) : 0; + soft_max_level = mask ? (fls(mask) - 1) : 0; + + switch (clk_type) { + case SMU_GFXCLK: + case SMU_SCLK: + if (soft_min_level > 2 || soft_max_level > 2) { + pr_info("Currently sclk only support 3 levels on APU\n"); + return -EINVAL; + } + + ret = smu_get_dpm_freq_range(smu, SMU_GFXCLK, &min_freq, &max_freq, false); + if (ret) + return ret; + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetSoftMaxGfxClk, + soft_max_level == 0 ? min_freq : + soft_max_level == 1 ? RENOIR_UMD_PSTATE_GFXCLK : max_freq); + if (ret) + return ret; + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetHardMinGfxClk, + soft_min_level == 2 ? max_freq : + soft_min_level == 1 ? RENOIR_UMD_PSTATE_GFXCLK : min_freq); + if (ret) + return ret; + break; + case SMU_SOCCLK: + GET_DPM_CUR_FREQ(clk_table, clk_type, soft_min_level, min_freq); + GET_DPM_CUR_FREQ(clk_table, clk_type, soft_max_level, max_freq); + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetSoftMaxSocclkByFreq, max_freq); + if (ret) + return ret; + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetHardMinSocclkByFreq, min_freq); + if (ret) + return ret; + break; + case SMU_MCLK: + case SMU_FCLK: + GET_DPM_CUR_FREQ(clk_table, clk_type, soft_min_level, min_freq); + GET_DPM_CUR_FREQ(clk_table, clk_type, soft_max_level, max_freq); + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetSoftMaxFclkByFreq, max_freq); + if (ret) + return ret; + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetHardMinFclkByFreq, min_freq); + if (ret) + return ret; + break; + default: + break; + } + + return ret; +} + +static int renoir_set_power_profile_mode(struct smu_context *smu, long *input, uint32_t size) +{ + int workload_type, ret; + uint32_t profile_mode = input[size]; + + if (profile_mode > PP_SMC_POWER_PROFILE_CUSTOM) { + pr_err("Invalid power profile mode %d\n", smu->power_profile_mode); + return -EINVAL; + } + + /* conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT */ + workload_type = smu_workload_get_type(smu, smu->power_profile_mode); + if (workload_type < 0) { + pr_err("Unsupported power profile mode %d on RENOIR\n",smu->power_profile_mode); + return -EINVAL; + } + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetWorkloadMask, + 1 << workload_type); + if (ret) { + pr_err("Fail to set workload type %d\n", workload_type); + return ret; + } + + smu->power_profile_mode = profile_mode; + + return 0; +} + +static int renoir_set_peak_clock_by_device(struct smu_context *smu) +{ + int ret = 0; + uint32_t sclk_freq = 0, uclk_freq = 0; + + ret = smu_get_dpm_freq_range(smu, SMU_SCLK, NULL, &sclk_freq, false); + if (ret) + return ret; + + ret = smu_set_soft_freq_range(smu, SMU_SCLK, sclk_freq, sclk_freq); + if (ret) + return ret; + + ret = smu_get_dpm_freq_range(smu, SMU_UCLK, NULL, &uclk_freq, false); + 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 renoir_set_performance_level(struct smu_context *smu, enum amd_dpm_forced_level level) +{ + int ret = 0; + + switch (level) { + case AMD_DPM_FORCED_LEVEL_PROFILE_PEAK: + ret = renoir_set_peak_clock_by_device(smu); + break; + default: + ret = -EINVAL; + break; + } + + return ret; +} + +/* save watermark settings into pplib smu structure, + * also pass data to smu controller + */ +static int renoir_set_watermarks_table( + struct smu_context *smu, + void *watermarks, + struct dm_pp_wm_sets_with_clock_ranges_soc15 *clock_ranges) +{ + int i; + int ret = 0; + Watermarks_t *table = watermarks; + + if (!table || !clock_ranges) + return -EINVAL; + + if (clock_ranges->num_wm_dmif_sets > 4 || + clock_ranges->num_wm_mcif_sets > 4) + return -EINVAL; + + /* save into smu->smu_table.tables[SMU_TABLE_WATERMARKS]->cpu_addr*/ + for (i = 0; i < clock_ranges->num_wm_dmif_sets; i++) { + table->WatermarkRow[WM_DCFCLK][i].MinClock = + cpu_to_le16((uint16_t) + (clock_ranges->wm_dmif_clocks_ranges[i].wm_min_dcfclk_clk_in_khz)); + table->WatermarkRow[WM_DCFCLK][i].MaxClock = + cpu_to_le16((uint16_t) + (clock_ranges->wm_dmif_clocks_ranges[i].wm_max_dcfclk_clk_in_khz)); + table->WatermarkRow[WM_DCFCLK][i].MinMclk = + cpu_to_le16((uint16_t) + (clock_ranges->wm_dmif_clocks_ranges[i].wm_min_mem_clk_in_khz)); + table->WatermarkRow[WM_DCFCLK][i].MaxMclk = + cpu_to_le16((uint16_t) + (clock_ranges->wm_dmif_clocks_ranges[i].wm_max_mem_clk_in_khz)); + table->WatermarkRow[WM_DCFCLK][i].WmSetting = (uint8_t) + clock_ranges->wm_dmif_clocks_ranges[i].wm_set_id; + } + + for (i = 0; i < clock_ranges->num_wm_mcif_sets; i++) { + table->WatermarkRow[WM_SOCCLK][i].MinClock = + cpu_to_le16((uint16_t) + (clock_ranges->wm_mcif_clocks_ranges[i].wm_min_socclk_clk_in_khz)); + table->WatermarkRow[WM_SOCCLK][i].MaxClock = + cpu_to_le16((uint16_t) + (clock_ranges->wm_mcif_clocks_ranges[i].wm_max_socclk_clk_in_khz)); + table->WatermarkRow[WM_SOCCLK][i].MinMclk = + cpu_to_le16((uint16_t) + (clock_ranges->wm_mcif_clocks_ranges[i].wm_min_mem_clk_in_khz)); + table->WatermarkRow[WM_SOCCLK][i].MaxMclk = + cpu_to_le16((uint16_t) + (clock_ranges->wm_mcif_clocks_ranges[i].wm_max_mem_clk_in_khz)); + table->WatermarkRow[WM_SOCCLK][i].WmSetting = (uint8_t) + clock_ranges->wm_mcif_clocks_ranges[i].wm_set_id; + } + + /* pass data to smu controller */ + ret = smu_write_watermarks_table(smu); + + return ret; +} + +static int renoir_get_power_profile_mode(struct smu_context *smu, + char *buf) +{ + static const char *profile_name[] = { + "BOOTUP_DEFAULT", + "3D_FULL_SCREEN", + "POWER_SAVING", + "VIDEO", + "VR", + "COMPUTE", + "CUSTOM"}; + uint32_t i, size = 0; + int16_t workload_type = 0; + + if (!smu->pm_enabled || !buf) + return -EINVAL; + + for (i = 0; i <= PP_SMC_POWER_PROFILE_CUSTOM; i++) { + /* + * Conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT + * Not all profile modes are supported on arcturus. + */ + workload_type = smu_workload_get_type(smu, i); + if (workload_type < 0) + continue; + + size += sprintf(buf + size, "%2d %14s%s\n", + i, profile_name[i], (i == smu->power_profile_mode) ? "*" : " "); + } + + return size; +} + +static const struct pptable_funcs renoir_ppt_funcs = { + .get_smu_msg_index = renoir_get_smu_msg_index, + .get_smu_table_index = renoir_get_smu_table_index, + .tables_init = renoir_tables_init, + .set_power_state = NULL, + .get_dpm_clk_limited = renoir_get_dpm_clk_limited, + .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, + .force_dpm_limit_value = renoir_force_dpm_limit_value, + .unforce_dpm_levels = renoir_unforce_dpm_levels, + .get_workload_type = renoir_get_workload_type, + .get_profiling_clk_mask = renoir_get_profiling_clk_mask, + .force_clk_levels = renoir_force_clk_levels, + .set_power_profile_mode = renoir_set_power_profile_mode, + .set_performance_level = renoir_set_performance_level, + .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, + .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, + .send_smc_msg = smu_v12_0_send_msg, + .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, + .gfx_off_control = smu_v12_0_gfx_off_control, + .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_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, +}; + +void renoir_set_ppt_funcs(struct smu_context *smu) +{ + smu->ppt_funcs = &renoir_ppt_funcs; + smu->smc_if_version = SMU12_DRIVER_IF_VERSION; + smu->is_apu = true; +} diff --git a/drivers/gpu/drm/amd/powerplay/renoir_ppt.h b/drivers/gpu/drm/amd/powerplay/renoir_ppt.h new file mode 100644 index 000000000000..2a390ddd37dd --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/renoir_ppt.h @@ -0,0 +1,53 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __RENOIR_PPT_H__ +#define __RENOIR_PPT_H__ + +extern void renoir_set_ppt_funcs(struct smu_context *smu); + +/* UMD PState Renoir Msg Parameters in MHz */ +#define RENOIR_UMD_PSTATE_GFXCLK 700 +#define RENOIR_UMD_PSTATE_SOCCLK 678 +#define RENOIR_UMD_PSTATE_FCLK 800 + +#define GET_DPM_CUR_FREQ(table, clk_type, dpm_level, freq) \ + do { \ + switch (clk_type) { \ + case SMU_SOCCLK: \ + freq = table->SocClocks[dpm_level].Freq; \ + break; \ + case SMU_MCLK: \ + freq = table->MemClocks[dpm_level].Freq; \ + break; \ + case SMU_DCEFCLK: \ + freq = table->DcfClocks[dpm_level].Freq; \ + break; \ + case SMU_FCLK: \ + freq = table->FClocks[dpm_level].Freq; \ + break; \ + default: \ + break; \ + } \ + } while (0) + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/smu_internal.h b/drivers/gpu/drm/amd/powerplay/smu_internal.h new file mode 100644 index 000000000000..8bcda7871309 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/smu_internal.h @@ -0,0 +1,204 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#ifndef __SMU_INTERNAL_H__ +#define __SMU_INTERNAL_H__ + +#include "amdgpu_smu.h" + +#define smu_init_microcode(smu) \ + ((smu)->ppt_funcs->init_microcode ? (smu)->ppt_funcs->init_microcode((smu)) : 0) +#define smu_init_smc_tables(smu) \ + ((smu)->ppt_funcs->init_smc_tables ? (smu)->ppt_funcs->init_smc_tables((smu)) : 0) +#define smu_fini_smc_tables(smu) \ + ((smu)->ppt_funcs->fini_smc_tables ? (smu)->ppt_funcs->fini_smc_tables((smu)) : 0) +#define smu_init_power(smu) \ + ((smu)->ppt_funcs->init_power ? (smu)->ppt_funcs->init_power((smu)) : 0) +#define smu_fini_power(smu) \ + ((smu)->ppt_funcs->fini_power ? (smu)->ppt_funcs->fini_power((smu)) : 0) + +#define smu_setup_pptable(smu) \ + ((smu)->ppt_funcs->setup_pptable ? (smu)->ppt_funcs->setup_pptable((smu)) : 0) +#define smu_powergate_sdma(smu, gate) \ + ((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_get_vbios_bootup_values(smu) \ + ((smu)->ppt_funcs->get_vbios_bootup_values ? (smu)->ppt_funcs->get_vbios_bootup_values((smu)) : 0) +#define smu_get_clk_info_from_vbios(smu) \ + ((smu)->ppt_funcs->get_clk_info_from_vbios ? (smu)->ppt_funcs->get_clk_info_from_vbios((smu)) : 0) +#define smu_check_pptable(smu) \ + ((smu)->ppt_funcs->check_pptable ? (smu)->ppt_funcs->check_pptable((smu)) : 0) +#define smu_parse_pptable(smu) \ + ((smu)->ppt_funcs->parse_pptable ? (smu)->ppt_funcs->parse_pptable((smu)) : 0) +#define smu_populate_smc_tables(smu) \ + ((smu)->ppt_funcs->populate_smc_tables ? (smu)->ppt_funcs->populate_smc_tables((smu)) : 0) +#define smu_check_fw_version(smu) \ + ((smu)->ppt_funcs->check_fw_version ? (smu)->ppt_funcs->check_fw_version((smu)) : 0) +#define smu_write_pptable(smu) \ + ((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_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) \ + ((smu)->ppt_funcs->notify_memory_pool_location ? (smu)->ppt_funcs->notify_memory_pool_location((smu)) : 0) +#define smu_gfx_off_control(smu, enable) \ + ((smu)->ppt_funcs->gfx_off_control ? (smu)->ppt_funcs->gfx_off_control((smu), (enable)) : 0) + +#define smu_set_last_dcef_min_deep_sleep_clk(smu) \ + ((smu)->ppt_funcs->set_last_dcef_min_deep_sleep_clk ? (smu)->ppt_funcs->set_last_dcef_min_deep_sleep_clk((smu)) : 0) +#define smu_system_features_control(smu, en) \ + ((smu)->ppt_funcs->system_features_control ? (smu)->ppt_funcs->system_features_control((smu), (en)) : 0) +#define smu_init_max_sustainable_clocks(smu) \ + ((smu)->ppt_funcs->init_max_sustainable_clocks ? (smu)->ppt_funcs->init_max_sustainable_clocks((smu)) : 0) +#define smu_set_default_od_settings(smu, initialize) \ + ((smu)->ppt_funcs->set_default_od_settings ? (smu)->ppt_funcs->set_default_od_settings((smu), (initialize)) : 0) + +#define smu_send_smc_msg(smu, msg) \ + ((smu)->ppt_funcs->send_smc_msg? (smu)->ppt_funcs->send_smc_msg((smu), (msg)) : 0) +#define smu_send_smc_msg_with_param(smu, msg, param) \ + ((smu)->ppt_funcs->send_smc_msg_with_param? (smu)->ppt_funcs->send_smc_msg_with_param((smu), (msg), (param)) : 0) +#define smu_read_smc_arg(smu, arg) \ + ((smu)->ppt_funcs->read_smc_arg? (smu)->ppt_funcs->read_smc_arg((smu), (arg)) : 0) +#define smu_alloc_dpm_context(smu) \ + ((smu)->ppt_funcs->alloc_dpm_context ? (smu)->ppt_funcs->alloc_dpm_context((smu)) : 0) +#define smu_init_display_count(smu, count) \ + ((smu)->ppt_funcs->init_display_count ? (smu)->ppt_funcs->init_display_count((smu), (count)) : 0) +#define smu_feature_set_allowed_mask(smu) \ + ((smu)->ppt_funcs->set_allowed_mask? (smu)->ppt_funcs->set_allowed_mask((smu)) : 0) +#define smu_feature_get_enabled_mask(smu, mask, num) \ + ((smu)->ppt_funcs->get_enabled_mask? (smu)->ppt_funcs->get_enabled_mask((smu), (mask), (num)) : 0) +#define smu_is_dpm_running(smu) \ + ((smu)->ppt_funcs->is_dpm_running ? (smu)->ppt_funcs->is_dpm_running((smu)) : 0) +#define smu_notify_display_change(smu) \ + ((smu)->ppt_funcs->notify_display_change? (smu)->ppt_funcs->notify_display_change((smu)) : 0) +#define smu_store_powerplay_table(smu) \ + ((smu)->ppt_funcs->store_powerplay_table ? (smu)->ppt_funcs->store_powerplay_table((smu)) : 0) +#define smu_check_powerplay_table(smu) \ + ((smu)->ppt_funcs->check_powerplay_table ? (smu)->ppt_funcs->check_powerplay_table((smu)) : 0) +#define smu_append_powerplay_table(smu) \ + ((smu)->ppt_funcs->append_powerplay_table ? (smu)->ppt_funcs->append_powerplay_table((smu)) : 0) +#define smu_set_default_dpm_table(smu) \ + ((smu)->ppt_funcs->set_default_dpm_table ? (smu)->ppt_funcs->set_default_dpm_table((smu)) : 0) +#define smu_populate_umd_state_clk(smu) \ + ((smu)->ppt_funcs->populate_umd_state_clk ? (smu)->ppt_funcs->populate_umd_state_clk((smu)) : 0) +#define smu_set_default_od8_settings(smu) \ + ((smu)->ppt_funcs->set_default_od8_settings ? (smu)->ppt_funcs->set_default_od8_settings((smu)) : 0) + +#define smu_get_current_clk_freq(smu, clk_id, value) \ + ((smu)->ppt_funcs->get_current_clk_freq? (smu)->ppt_funcs->get_current_clk_freq((smu), (clk_id), (value)) : 0) + +#define smu_tables_init(smu, tab) \ + ((smu)->ppt_funcs->tables_init ? (smu)->ppt_funcs->tables_init((smu), (tab)) : 0) +#define smu_set_thermal_fan_table(smu) \ + ((smu)->ppt_funcs->set_thermal_fan_table ? (smu)->ppt_funcs->set_thermal_fan_table((smu)) : 0) +#define smu_start_thermal_control(smu) \ + ((smu)->ppt_funcs->start_thermal_control? (smu)->ppt_funcs->start_thermal_control((smu)) : 0) +#define smu_stop_thermal_control(smu) \ + ((smu)->ppt_funcs->stop_thermal_control? (smu)->ppt_funcs->stop_thermal_control((smu)) : 0) + +#define smu_smc_read_sensor(smu, sensor, data, size) \ + ((smu)->ppt_funcs->read_sensor? (smu)->ppt_funcs->read_sensor((smu), (sensor), (data), (size)) : -EINVAL) + +#define smu_pre_display_config_changed(smu) \ + ((smu)->ppt_funcs->pre_display_config_changed ? (smu)->ppt_funcs->pre_display_config_changed((smu)) : 0) +#define smu_display_config_changed(smu) \ + ((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_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) \ + ((smu)->ppt_funcs->unforce_dpm_levels ? (smu)->ppt_funcs->unforce_dpm_levels((smu)) : 0) +#define smu_get_profiling_clk_mask(smu, level, sclk_mask, mclk_mask, soc_mask) \ + ((smu)->ppt_funcs->get_profiling_clk_mask ? (smu)->ppt_funcs->get_profiling_clk_mask((smu), (level), (sclk_mask), (mclk_mask), (soc_mask)) : 0) +#define smu_set_cpu_power_state(smu) \ + ((smu)->ppt_funcs->set_cpu_power_state ? (smu)->ppt_funcs->set_cpu_power_state((smu)) : 0) + +#define smu_msg_get_index(smu, msg) \ + ((smu)->ppt_funcs? ((smu)->ppt_funcs->get_smu_msg_index? (smu)->ppt_funcs->get_smu_msg_index((smu), (msg)) : -EINVAL) : -EINVAL) +#define smu_clk_get_index(smu, msg) \ + ((smu)->ppt_funcs? ((smu)->ppt_funcs->get_smu_clk_index? (smu)->ppt_funcs->get_smu_clk_index((smu), (msg)) : -EINVAL) : -EINVAL) +#define smu_feature_get_index(smu, msg) \ + ((smu)->ppt_funcs? ((smu)->ppt_funcs->get_smu_feature_index? (smu)->ppt_funcs->get_smu_feature_index((smu), (msg)) : -EINVAL) : -EINVAL) +#define smu_table_get_index(smu, tab) \ + ((smu)->ppt_funcs? ((smu)->ppt_funcs->get_smu_table_index? (smu)->ppt_funcs->get_smu_table_index((smu), (tab)) : -EINVAL) : -EINVAL) +#define smu_power_get_index(smu, src) \ + ((smu)->ppt_funcs? ((smu)->ppt_funcs->get_smu_power_index? (smu)->ppt_funcs->get_smu_power_index((smu), (src)) : -EINVAL) : -EINVAL) +#define smu_workload_get_type(smu, profile) \ + ((smu)->ppt_funcs? ((smu)->ppt_funcs->get_workload_type? (smu)->ppt_funcs->get_workload_type((smu), (profile)) : -EINVAL) : -EINVAL) +#define smu_run_btc(smu) \ + ((smu)->ppt_funcs? ((smu)->ppt_funcs->run_btc? (smu)->ppt_funcs->run_btc((smu)) : 0) : 0) +#define smu_get_allowed_feature_mask(smu, feature_mask, num) \ + ((smu)->ppt_funcs? ((smu)->ppt_funcs->get_allowed_feature_mask? (smu)->ppt_funcs->get_allowed_feature_mask((smu), (feature_mask), (num)) : 0) : 0) + + +#define smu_store_cc6_data(smu, st, cc6_dis, pst_dis, pst_sw_dis) \ + ((smu)->ppt_funcs->store_cc6_data ? (smu)->ppt_funcs->store_cc6_data((smu), (st), (cc6_dis), (pst_dis), (pst_sw_dis)) : 0) + +#define smu_get_dal_power_level(smu, clocks) \ + ((smu)->ppt_funcs->get_dal_power_level ? (smu)->ppt_funcs->get_dal_power_level((smu), (clocks)) : 0) +#define smu_get_perf_level(smu, designation, level) \ + ((smu)->ppt_funcs->get_perf_level ? (smu)->ppt_funcs->get_perf_level((smu), (designation), (level)) : 0) +#define smu_get_current_shallow_sleep_clocks(smu, clocks) \ + ((smu)->ppt_funcs->get_current_shallow_sleep_clocks ? (smu)->ppt_funcs->get_current_shallow_sleep_clocks((smu), (clocks)) : 0) + +#define smu_dpm_set_uvd_enable(smu, enable) \ + ((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_set_watermarks_table(smu, tab, clock_ranges) \ + ((smu)->ppt_funcs->set_watermarks_table ? (smu)->ppt_funcs->set_watermarks_table((smu), (tab), (clock_ranges)) : 0) +#define smu_get_current_clk_freq_by_table(smu, clk_type, value) \ + ((smu)->ppt_funcs->get_current_clk_freq_by_table ? (smu)->ppt_funcs->get_current_clk_freq_by_table((smu), (clk_type), (value)) : 0) +#define smu_thermal_temperature_range_update(smu, range, rw) \ + ((smu)->ppt_funcs->thermal_temperature_range_update? (smu)->ppt_funcs->thermal_temperature_range_update((smu), (range), (rw)) : 0) +#define smu_get_thermal_temperature_range(smu, range) \ + ((smu)->ppt_funcs->get_thermal_temperature_range? (smu)->ppt_funcs->get_thermal_temperature_range((smu), (range)) : 0) +#define smu_register_irq_handler(smu) \ + ((smu)->ppt_funcs->register_irq_handler ? (smu)->ppt_funcs->register_irq_handler(smu) : 0) + +#define smu_get_dpm_ultimate_freq(smu, param, min, max) \ + ((smu)->ppt_funcs->get_dpm_ultimate_freq ? (smu)->ppt_funcs->get_dpm_ultimate_freq((smu), (param), (min), (max)) : 0) + +#define smu_asic_set_performance_level(smu, level) \ + ((smu)->ppt_funcs->set_performance_level? (smu)->ppt_funcs->set_performance_level((smu), (level)) : -EINVAL); +#define smu_dump_pptable(smu) \ + ((smu)->ppt_funcs->dump_pptable ? (smu)->ppt_funcs->dump_pptable((smu)) : 0) +#define smu_get_dpm_clk_limited(smu, clk_type, dpm_level, freq) \ + ((smu)->ppt_funcs->get_dpm_clk_limited ? (smu)->ppt_funcs->get_dpm_clk_limited((smu), (clk_type), (dpm_level), (freq)) : -EINVAL) + +#define smu_set_soft_freq_limited_range(smu, clk_type, min, max) \ + ((smu)->ppt_funcs->set_soft_freq_limited_range ? (smu)->ppt_funcs->set_soft_freq_limited_range((smu), (clk_type), (min), (max)) : -EINVAL) + +#define smu_override_pcie_parameters(smu) \ + ((smu)->ppt_funcs->override_pcie_parameters ? (smu)->ppt_funcs->override_pcie_parameters((smu)) : 0) + +#define smu_update_pcie_parameters(smu, pcie_gen_cap, pcie_width_cap) \ + ((smu)->ppt_funcs->update_pcie_parameters ? (smu)->ppt_funcs->update_pcie_parameters((smu), (pcie_gen_cap), (pcie_width_cap)) : 0) + +#endif diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c new file mode 100644 index 000000000000..fc9679ea2368 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c @@ -0,0 +1,1842 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include <linux/firmware.h> +#include <linux/module.h> +#include <linux/pci.h> + +#define SMU_11_0_PARTIAL_PPTABLE + +#include "pp_debug.h" +#include "amdgpu.h" +#include "amdgpu_smu.h" +#include "smu_internal.h" +#include "atomfirmware.h" +#include "amdgpu_atomfirmware.h" +#include "smu_v11_0.h" +#include "smu_v11_0_pptable.h" +#include "soc15_common.h" +#include "atom.h" +#include "amd_pcie.h" + +#include "asic_reg/thm/thm_11_0_2_offset.h" +#include "asic_reg/thm/thm_11_0_2_sh_mask.h" +#include "asic_reg/mp/mp_11_0_offset.h" +#include "asic_reg/mp/mp_11_0_sh_mask.h" +#include "asic_reg/nbio/nbio_7_4_offset.h" +#include "asic_reg/nbio/nbio_7_4_sh_mask.h" +#include "asic_reg/smuio/smuio_11_0_0_offset.h" +#include "asic_reg/smuio/smuio_11_0_0_sh_mask.h" + +MODULE_FIRMWARE("amdgpu/vega20_smc.bin"); +MODULE_FIRMWARE("amdgpu/arcturus_smc.bin"); +MODULE_FIRMWARE("amdgpu/navi10_smc.bin"); +MODULE_FIRMWARE("amdgpu/navi14_smc.bin"); +MODULE_FIRMWARE("amdgpu/navi12_smc.bin"); + +#define SMU11_VOLTAGE_SCALE 4 + +static int smu_v11_0_send_msg_without_waiting(struct smu_context *smu, + uint16_t msg) +{ + struct amdgpu_device *adev = smu->adev; + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_66, msg); + return 0; +} + +int smu_v11_0_read_arg(struct smu_context *smu, uint32_t *arg) +{ + struct amdgpu_device *adev = smu->adev; + + *arg = RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_82); + return 0; +} + +static int smu_v11_0_wait_for_response(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + uint32_t cur_value, i, timeout = adev->usec_timeout * 10; + + 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; + udelay(1); + } + + /* timeout means wrong logic */ + if (i == timeout) + return -ETIME; + + return RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90) == 0x1 ? 0 : -EIO; +} + +int smu_v11_0_send_msg(struct smu_context *smu, uint16_t msg) +{ + struct amdgpu_device *adev = smu->adev; + int ret = 0, index = 0; + + index = smu_msg_get_index(smu, msg); + if (index < 0) + return index; + + smu_v11_0_wait_for_response(smu); + + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90, 0); + + smu_v11_0_send_msg_without_waiting(smu, (uint16_t)index); + + ret = smu_v11_0_wait_for_response(smu); + + if (ret) + pr_err("failed send message: %10s (%d) response %#x\n", + smu_get_message_name(smu, msg), index, ret); + + return ret; + +} + +int +smu_v11_0_send_msg_with_param(struct smu_context *smu, uint16_t msg, + uint32_t param) +{ + + struct amdgpu_device *adev = smu->adev; + int ret = 0, index = 0; + + index = smu_msg_get_index(smu, msg); + if (index < 0) + 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); + + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90, 0); + + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_82, param); + + smu_v11_0_send_msg_without_waiting(smu, (uint16_t)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); + + return ret; +} + +int smu_v11_0_init_microcode(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + const char *chip_name; + char fw_name[30]; + int err = 0; + const struct smc_firmware_header_v1_0 *hdr; + const struct common_firmware_header *header; + struct amdgpu_firmware_info *ucode = NULL; + + switch (adev->asic_type) { + case CHIP_VEGA20: + chip_name = "vega20"; + break; + case CHIP_ARCTURUS: + chip_name = "arcturus"; + break; + case CHIP_NAVI10: + chip_name = "navi10"; + break; + case CHIP_NAVI14: + chip_name = "navi14"; + break; + case CHIP_NAVI12: + chip_name = "navi12"; + break; + default: + BUG(); + } + + snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_smc.bin", chip_name); + + err = request_firmware(&adev->pm.fw, fw_name, adev->dev); + if (err) + goto out; + err = amdgpu_ucode_validate(adev->pm.fw); + if (err) + goto out; + + hdr = (const struct smc_firmware_header_v1_0 *) adev->pm.fw->data; + amdgpu_ucode_print_smc_hdr(&hdr->header); + adev->pm.fw_version = le32_to_cpu(hdr->header.ucode_version); + + if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) { + ucode = &adev->firmware.ucode[AMDGPU_UCODE_ID_SMC]; + ucode->ucode_id = AMDGPU_UCODE_ID_SMC; + ucode->fw = adev->pm.fw; + header = (const struct common_firmware_header *)ucode->fw->data; + adev->firmware.fw_size += + ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE); + } + +out: + if (err) { + DRM_ERROR("smu_v11_0: Failed to load firmware \"%s\"\n", + fw_name); + release_firmware(adev->pm.fw); + adev->pm.fw = NULL; + } + return err; +} + +int smu_v11_0_load_microcode(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + const uint32_t *src; + const struct smc_firmware_header_v1_0 *hdr; + uint32_t addr_start = MP1_SRAM; + uint32_t i; + uint32_t mp1_fw_flags; + + hdr = (const struct smc_firmware_header_v1_0 *) adev->pm.fw->data; + src = (const uint32_t *)(adev->pm.fw->data + + le32_to_cpu(hdr->header.ucode_array_offset_bytes)); + + for (i = 1; i < MP1_SMC_SIZE/4 - 1; i++) { + WREG32_PCIE(addr_start, src[i]); + addr_start += 4; + } + + WREG32_PCIE(MP1_Public | (smnMP1_PUB_CTRL & 0xffffffff), + 1 & MP1_SMN_PUB_CTRL__RESET_MASK); + WREG32_PCIE(MP1_Public | (smnMP1_PUB_CTRL & 0xffffffff), + 1 & ~MP1_SMN_PUB_CTRL__RESET_MASK); + + for (i = 0; i < adev->usec_timeout; i++) { + mp1_fw_flags = RREG32_PCIE(MP1_Public | + (smnMP1_FIRMWARE_FLAGS & 0xffffffff)); + if ((mp1_fw_flags & MP1_FIRMWARE_FLAGS__INTERRUPTS_ENABLED_MASK) >> + MP1_FIRMWARE_FLAGS__INTERRUPTS_ENABLED__SHIFT) + break; + udelay(1); + } + + if (i == adev->usec_timeout) + return -ETIME; + + return 0; +} + +int smu_v11_0_check_fw_status(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + uint32_t mp1_fw_flags; + + mp1_fw_flags = RREG32_PCIE(MP1_Public | + (smnMP1_FIRMWARE_FLAGS & 0xffffffff)); + + if ((mp1_fw_flags & MP1_FIRMWARE_FLAGS__INTERRUPTS_ENABLED_MASK) >> + MP1_FIRMWARE_FLAGS__INTERRUPTS_ENABLED__SHIFT) + return 0; + + return -EIO; +} + +int smu_v11_0_check_fw_version(struct smu_context *smu) +{ + uint32_t if_version = 0xff, smu_version = 0xff; + uint16_t smu_major; + uint8_t smu_minor, smu_debug; + int ret = 0; + + ret = smu_get_smc_version(smu, &if_version, &smu_version); + if (ret) + return ret; + + smu_major = (smu_version >> 16) & 0xffff; + smu_minor = (smu_version >> 8) & 0xff; + smu_debug = (smu_version >> 0) & 0xff; + + switch (smu->adev->asic_type) { + case CHIP_VEGA20: + smu->smc_if_version = SMU11_DRIVER_IF_VERSION_VG20; + break; + case CHIP_ARCTURUS: + smu->smc_if_version = SMU11_DRIVER_IF_VERSION_ARCT; + break; + case CHIP_NAVI10: + smu->smc_if_version = SMU11_DRIVER_IF_VERSION_NV10; + break; + case CHIP_NAVI14: + smu->smc_if_version = SMU11_DRIVER_IF_VERSION_NV14; + break; + default: + pr_err("smu unsupported asic type:%d.\n", smu->adev->asic_type); + smu->smc_if_version = SMU11_DRIVER_IF_VERSION_INV; + break; + } + + /* + * 1. if_version mismatch is not critical as our fw is designed + * to be backward compatible. + * 2. New fw usually brings some optimizations. But that's visible + * only on the paired driver. + * Considering above, we just leave user a warning message instead + * of halt driver loading. + */ + if (if_version != smu->smc_if_version) { + pr_info("smu driver if version = 0x%08x, smu fw if version = 0x%08x, " + "smu fw version = 0x%08x (%d.%d.%d)\n", + smu->smc_if_version, if_version, + smu_version, smu_major, smu_minor, smu_debug); + pr_warn("SMU driver if version not matched\n"); + } + + return ret; +} + +static int smu_v11_0_set_pptable_v2_0(struct smu_context *smu, void **table, uint32_t *size) +{ + struct amdgpu_device *adev = smu->adev; + uint32_t ppt_offset_bytes; + const struct smc_firmware_header_v2_0 *v2; + + v2 = (const struct smc_firmware_header_v2_0 *) adev->pm.fw->data; + + ppt_offset_bytes = le32_to_cpu(v2->ppt_offset_bytes); + *size = le32_to_cpu(v2->ppt_size_bytes); + *table = (uint8_t *)v2 + ppt_offset_bytes; + + return 0; +} + +static int smu_v11_0_set_pptable_v2_1(struct smu_context *smu, void **table, + uint32_t *size, uint32_t pptable_id) +{ + struct amdgpu_device *adev = smu->adev; + const struct smc_firmware_header_v2_1 *v2_1; + struct smc_soft_pptable_entry *entries; + uint32_t pptable_count = 0; + int i = 0; + + v2_1 = (const struct smc_firmware_header_v2_1 *) adev->pm.fw->data; + entries = (struct smc_soft_pptable_entry *) + ((uint8_t *)v2_1 + le32_to_cpu(v2_1->pptable_entry_offset)); + pptable_count = le32_to_cpu(v2_1->pptable_count); + for (i = 0; i < pptable_count; i++) { + if (le32_to_cpu(entries[i].id) == pptable_id) { + *table = ((uint8_t *)v2_1 + le32_to_cpu(entries[i].ppt_offset_bytes)); + *size = le32_to_cpu(entries[i].ppt_size_bytes); + break; + } + } + + if (i == pptable_count) + return -EINVAL; + + return 0; +} + +int smu_v11_0_setup_pptable(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + const struct smc_firmware_header_v1_0 *hdr; + int ret, index; + uint32_t size = 0; + uint16_t atom_table_size; + uint8_t frev, crev; + void *table; + uint16_t version_major, version_minor; + + hdr = (const struct smc_firmware_header_v1_0 *) adev->pm.fw->data; + version_major = le16_to_cpu(hdr->header.header_version_major); + version_minor = le16_to_cpu(hdr->header.header_version_minor); + if (version_major == 2 && smu->smu_table.boot_values.pp_table_id > 0) { + pr_info("use driver provided pptable %d\n", smu->smu_table.boot_values.pp_table_id); + switch (version_minor) { + case 0: + ret = smu_v11_0_set_pptable_v2_0(smu, &table, &size); + break; + case 1: + ret = smu_v11_0_set_pptable_v2_1(smu, &table, &size, + smu->smu_table.boot_values.pp_table_id); + break; + default: + ret = -EINVAL; + break; + } + if (ret) + return ret; + + } else { + pr_info("use vbios provided pptable\n"); + index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1, + powerplayinfo); + + ret = smu_get_atom_data_table(smu, index, &atom_table_size, &frev, &crev, + (uint8_t **)&table); + if (ret) + return ret; + size = atom_table_size; + } + + if (!smu->smu_table.power_play_table) + smu->smu_table.power_play_table = table; + if (!smu->smu_table.power_play_table_size) + smu->smu_table.power_play_table_size = size; + + return 0; +} + +static int smu_v11_0_init_dpm_context(struct smu_context *smu) +{ + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + + if (smu_dpm->dpm_context || smu_dpm->dpm_context_size != 0) + return -EINVAL; + + return smu_alloc_dpm_context(smu); +} + +static int smu_v11_0_fini_dpm_context(struct smu_context *smu) +{ + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + + if (!smu_dpm->dpm_context || smu_dpm->dpm_context_size == 0) + return -EINVAL; + + kfree(smu_dpm->dpm_context); + kfree(smu_dpm->golden_dpm_context); + kfree(smu_dpm->dpm_current_power_state); + kfree(smu_dpm->dpm_request_power_state); + smu_dpm->dpm_context = NULL; + smu_dpm->golden_dpm_context = NULL; + smu_dpm->dpm_context_size = 0; + smu_dpm->dpm_current_power_state = NULL; + smu_dpm->dpm_request_power_state = NULL; + + return 0; +} + +int smu_v11_0_init_smc_tables(struct smu_context *smu) +{ + struct smu_table_context *smu_table = &smu->smu_table; + struct smu_table *tables = NULL; + int ret = 0; + + if (smu_table->tables) + return -EINVAL; + + tables = kcalloc(SMU_TABLE_COUNT, sizeof(struct smu_table), + GFP_KERNEL); + if (!tables) + return -ENOMEM; + + smu_table->tables = tables; + + ret = smu_tables_init(smu, tables); + if (ret) + return ret; + + ret = smu_v11_0_init_dpm_context(smu); + if (ret) + return ret; + + return 0; +} + +int smu_v11_0_fini_smc_tables(struct smu_context *smu) +{ + struct smu_table_context *smu_table = &smu->smu_table; + int ret = 0; + + if (!smu_table->tables) + return -EINVAL; + + kfree(smu_table->tables); + kfree(smu_table->metrics_table); + smu_table->tables = NULL; + smu_table->metrics_table = NULL; + smu_table->metrics_time = 0; + + ret = smu_v11_0_fini_dpm_context(smu); + if (ret) + return ret; + return 0; +} + +int smu_v11_0_init_power(struct smu_context *smu) +{ + struct smu_power_context *smu_power = &smu->smu_power; + + if (!smu->pm_enabled) + return 0; + if (smu_power->power_context || smu_power->power_context_size != 0) + return -EINVAL; + + smu_power->power_context = kzalloc(sizeof(struct smu_11_0_dpm_context), + GFP_KERNEL); + if (!smu_power->power_context) + return -ENOMEM; + smu_power->power_context_size = sizeof(struct smu_11_0_dpm_context); + + return 0; +} + +int smu_v11_0_fini_power(struct smu_context *smu) +{ + struct smu_power_context *smu_power = &smu->smu_power; + + if (!smu->pm_enabled) + return 0; + if (!smu_power->power_context || smu_power->power_context_size == 0) + return -EINVAL; + + kfree(smu_power->power_context); + smu_power->power_context = NULL; + smu_power->power_context_size = 0; + + return 0; +} + +int smu_v11_0_get_vbios_bootup_values(struct smu_context *smu) +{ + int ret, index; + uint16_t size; + uint8_t frev, crev; + struct atom_common_table_header *header; + struct atom_firmware_info_v3_3 *v_3_3; + struct atom_firmware_info_v3_1 *v_3_1; + + index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1, + firmwareinfo); + + ret = smu_get_atom_data_table(smu, index, &size, &frev, &crev, + (uint8_t **)&header); + if (ret) + return ret; + + if (header->format_revision != 3) { + pr_err("unknown atom_firmware_info version! for smu11\n"); + return -EINVAL; + } + + switch (header->content_revision) { + case 0: + case 1: + case 2: + v_3_1 = (struct atom_firmware_info_v3_1 *)header; + smu->smu_table.boot_values.revision = v_3_1->firmware_revision; + smu->smu_table.boot_values.gfxclk = v_3_1->bootup_sclk_in10khz; + smu->smu_table.boot_values.uclk = v_3_1->bootup_mclk_in10khz; + smu->smu_table.boot_values.socclk = 0; + smu->smu_table.boot_values.dcefclk = 0; + smu->smu_table.boot_values.vddc = v_3_1->bootup_vddc_mv; + smu->smu_table.boot_values.vddci = v_3_1->bootup_vddci_mv; + smu->smu_table.boot_values.mvddc = v_3_1->bootup_mvddc_mv; + smu->smu_table.boot_values.vdd_gfx = v_3_1->bootup_vddgfx_mv; + smu->smu_table.boot_values.cooling_id = v_3_1->coolingsolution_id; + smu->smu_table.boot_values.pp_table_id = 0; + break; + case 3: + default: + v_3_3 = (struct atom_firmware_info_v3_3 *)header; + smu->smu_table.boot_values.revision = v_3_3->firmware_revision; + smu->smu_table.boot_values.gfxclk = v_3_3->bootup_sclk_in10khz; + smu->smu_table.boot_values.uclk = v_3_3->bootup_mclk_in10khz; + smu->smu_table.boot_values.socclk = 0; + smu->smu_table.boot_values.dcefclk = 0; + smu->smu_table.boot_values.vddc = v_3_3->bootup_vddc_mv; + smu->smu_table.boot_values.vddci = v_3_3->bootup_vddci_mv; + smu->smu_table.boot_values.mvddc = v_3_3->bootup_mvddc_mv; + smu->smu_table.boot_values.vdd_gfx = v_3_3->bootup_vddgfx_mv; + smu->smu_table.boot_values.cooling_id = v_3_3->coolingsolution_id; + smu->smu_table.boot_values.pp_table_id = v_3_3->pplib_pptable_id; + } + + smu->smu_table.boot_values.format_revision = header->format_revision; + smu->smu_table.boot_values.content_revision = header->content_revision; + + return 0; +} + +int smu_v11_0_get_clk_info_from_vbios(struct smu_context *smu) +{ + int ret, index; + struct amdgpu_device *adev = smu->adev; + struct atom_get_smu_clock_info_parameters_v3_1 input = {0}; + struct atom_get_smu_clock_info_output_parameters_v3_1 *output; + + input.clk_id = SMU11_SYSPLL0_SOCCLK_ID; + input.command = GET_SMU_CLOCK_INFO_V3_1_GET_CLOCK_FREQ; + index = get_index_into_master_table(atom_master_list_of_command_functions_v2_1, + getsmuclockinfo); + + ret = amdgpu_atom_execute_table(adev->mode_info.atom_context, index, + (uint32_t *)&input); + if (ret) + return -EINVAL; + + output = (struct atom_get_smu_clock_info_output_parameters_v3_1 *)&input; + smu->smu_table.boot_values.socclk = le32_to_cpu(output->atom_smu_outputclkfreq.smu_clock_freq_hz) / 10000; + + memset(&input, 0, sizeof(input)); + input.clk_id = SMU11_SYSPLL0_DCEFCLK_ID; + input.command = GET_SMU_CLOCK_INFO_V3_1_GET_CLOCK_FREQ; + index = get_index_into_master_table(atom_master_list_of_command_functions_v2_1, + getsmuclockinfo); + + ret = amdgpu_atom_execute_table(adev->mode_info.atom_context, index, + (uint32_t *)&input); + if (ret) + return -EINVAL; + + output = (struct atom_get_smu_clock_info_output_parameters_v3_1 *)&input; + smu->smu_table.boot_values.dcefclk = le32_to_cpu(output->atom_smu_outputclkfreq.smu_clock_freq_hz) / 10000; + + memset(&input, 0, sizeof(input)); + input.clk_id = SMU11_SYSPLL0_ECLK_ID; + input.command = GET_SMU_CLOCK_INFO_V3_1_GET_CLOCK_FREQ; + index = get_index_into_master_table(atom_master_list_of_command_functions_v2_1, + getsmuclockinfo); + + ret = amdgpu_atom_execute_table(adev->mode_info.atom_context, index, + (uint32_t *)&input); + if (ret) + return -EINVAL; + + output = (struct atom_get_smu_clock_info_output_parameters_v3_1 *)&input; + smu->smu_table.boot_values.eclk = le32_to_cpu(output->atom_smu_outputclkfreq.smu_clock_freq_hz) / 10000; + + memset(&input, 0, sizeof(input)); + input.clk_id = SMU11_SYSPLL0_VCLK_ID; + input.command = GET_SMU_CLOCK_INFO_V3_1_GET_CLOCK_FREQ; + index = get_index_into_master_table(atom_master_list_of_command_functions_v2_1, + getsmuclockinfo); + + ret = amdgpu_atom_execute_table(adev->mode_info.atom_context, index, + (uint32_t *)&input); + if (ret) + return -EINVAL; + + output = (struct atom_get_smu_clock_info_output_parameters_v3_1 *)&input; + smu->smu_table.boot_values.vclk = le32_to_cpu(output->atom_smu_outputclkfreq.smu_clock_freq_hz) / 10000; + + memset(&input, 0, sizeof(input)); + input.clk_id = SMU11_SYSPLL0_DCLK_ID; + input.command = GET_SMU_CLOCK_INFO_V3_1_GET_CLOCK_FREQ; + index = get_index_into_master_table(atom_master_list_of_command_functions_v2_1, + getsmuclockinfo); + + ret = amdgpu_atom_execute_table(adev->mode_info.atom_context, index, + (uint32_t *)&input); + if (ret) + return -EINVAL; + + output = (struct atom_get_smu_clock_info_output_parameters_v3_1 *)&input; + smu->smu_table.boot_values.dclk = le32_to_cpu(output->atom_smu_outputclkfreq.smu_clock_freq_hz) / 10000; + + if ((smu->smu_table.boot_values.format_revision == 3) && + (smu->smu_table.boot_values.content_revision >= 2)) { + memset(&input, 0, sizeof(input)); + input.clk_id = SMU11_SYSPLL1_0_FCLK_ID; + input.syspll_id = SMU11_SYSPLL1_2_ID; + input.command = GET_SMU_CLOCK_INFO_V3_1_GET_CLOCK_FREQ; + index = get_index_into_master_table(atom_master_list_of_command_functions_v2_1, + getsmuclockinfo); + + ret = amdgpu_atom_execute_table(adev->mode_info.atom_context, index, + (uint32_t *)&input); + if (ret) + return -EINVAL; + + output = (struct atom_get_smu_clock_info_output_parameters_v3_1 *)&input; + smu->smu_table.boot_values.fclk = le32_to_cpu(output->atom_smu_outputclkfreq.smu_clock_freq_hz) / 10000; + } + + return 0; +} + +int smu_v11_0_notify_memory_pool_location(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; + uint64_t address; + uint32_t address_low, address_high; + + if (memory_pool->size == 0 || memory_pool->cpu_addr == NULL) + return ret; + + address = (uintptr_t)memory_pool->cpu_addr; + address_high = (uint32_t)upper_32_bits(address); + address_low = (uint32_t)lower_32_bits(address); + + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetSystemVirtualDramAddrHigh, + address_high); + if (ret) + return ret; + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetSystemVirtualDramAddrLow, + address_low); + if (ret) + return ret; + + address = memory_pool->mc_address; + address_high = (uint32_t)upper_32_bits(address); + address_low = (uint32_t)lower_32_bits(address); + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_DramLogSetDramAddrHigh, + address_high); + if (ret) + return ret; + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_DramLogSetDramAddrLow, + address_low); + if (ret) + return ret; + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_DramLogSetDramSize, + (uint32_t)memory_pool->size); + if (ret) + return ret; + + return ret; +} + +int smu_v11_0_check_pptable(struct smu_context *smu) +{ + int ret; + + ret = smu_check_powerplay_table(smu); + return ret; +} + +int smu_v11_0_parse_pptable(struct smu_context *smu) +{ + int ret; + + struct smu_table_context *table_context = &smu->smu_table; + struct smu_table *table = &table_context->tables[SMU_TABLE_PPTABLE]; + + if (table_context->driver_pptable) + return -EINVAL; + + table_context->driver_pptable = kzalloc(table->size, GFP_KERNEL); + + if (!table_context->driver_pptable) + return -ENOMEM; + + ret = smu_store_powerplay_table(smu); + if (ret) + return -EINVAL; + + ret = smu_append_powerplay_table(smu); + + return ret; +} + +int smu_v11_0_populate_smc_pptable(struct smu_context *smu) +{ + int ret; + + ret = smu_set_default_dpm_table(smu); + + return ret; +} + +int smu_v11_0_write_pptable(struct smu_context *smu) +{ + struct smu_table_context *table_context = &smu->smu_table; + int ret = 0; + + ret = smu_update_table(smu, SMU_TABLE_PPTABLE, 0, + table_context->driver_pptable, true); + + return ret; +} + +int smu_v11_0_set_deep_sleep_dcefclk(struct smu_context *smu, uint32_t clk) +{ + int ret; + + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetMinDeepSleepDcefclk, clk); + if (ret) + pr_err("SMU11 attempt to set divider for DCEFCLK Failed!"); + + return ret; +} + +int smu_v11_0_set_min_dcef_deep_sleep(struct smu_context *smu) +{ + struct smu_table_context *table_context = &smu->smu_table; + + if (!smu->pm_enabled) + return 0; + if (!table_context) + return -EINVAL; + + return smu_v11_0_set_deep_sleep_dcefclk(smu, table_context->boot_values.dcefclk / 100); +} + +int smu_v11_0_set_tool_table_location(struct smu_context *smu) +{ + int ret = 0; + struct smu_table *tool_table = &smu->smu_table.tables[SMU_TABLE_PMSTATUSLOG]; + + if (tool_table->mc_address) { + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetToolsDramAddrHigh, + upper_32_bits(tool_table->mc_address)); + if (!ret) + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetToolsDramAddrLow, + lower_32_bits(tool_table->mc_address)); + } + + return ret; +} + +int smu_v11_0_init_display_count(struct smu_context *smu, uint32_t count) +{ + int ret = 0; + + if (!smu->pm_enabled) + return ret; + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_NumOfDisplays, count); + return ret; +} + + +int smu_v11_0_set_allowed_mask(struct smu_context *smu) +{ + struct smu_feature *feature = &smu->smu_feature; + int ret = 0; + uint32_t feature_mask[2]; + + mutex_lock(&feature->mutex); + if (bitmap_empty(feature->allowed, SMU_FEATURE_MAX) || feature->feature_num < 64) + goto failed; + + bitmap_copy((unsigned long *)feature_mask, feature->allowed, 64); + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetAllowedFeaturesMaskHigh, + feature_mask[1]); + if (ret) + goto failed; + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetAllowedFeaturesMaskLow, + feature_mask[0]); + if (ret) + goto failed; + +failed: + mutex_unlock(&feature->mutex); + return ret; +} + +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; + 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; + + 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_v11_0_system_features_control(struct smu_context *smu, + bool en) +{ + struct smu_feature *feature = &smu->smu_feature; + 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); + 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); + + return ret; +} + +int smu_v11_0_notify_display_change(struct smu_context *smu) +{ + int ret = 0; + + if (!smu->pm_enabled) + return ret; + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT) && + smu->adev->gmc.vram_type == AMDGPU_VRAM_TYPE_HBM) + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetUclkFastSwitch, 1); + + return ret; +} + +static int +smu_v11_0_get_max_sustainable_clock(struct smu_context *smu, uint32_t *clock, + enum smu_clk_type clock_select) +{ + int ret = 0; + int clk_id; + + if (!smu->pm_enabled) + return ret; + + if ((smu_msg_get_index(smu, SMU_MSG_GetDcModeMaxDpmFreq) < 0) || + (smu_msg_get_index(smu, SMU_MSG_GetMaxDpmFreq) < 0)) + return 0; + + clk_id = smu_clk_get_index(smu, clock_select); + if (clk_id < 0) + return -EINVAL; + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_GetDcModeMaxDpmFreq, + clk_id << 16); + if (ret) { + pr_err("[GetMaxSustainableClock] Failed to get max DC clock from SMC!"); + return ret; + } + + ret = smu_read_smc_arg(smu, clock); + if (ret) + return ret; + + if (*clock != 0) + return 0; + + /* if DC limit is zero, return AC limit */ + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_GetMaxDpmFreq, + clk_id << 16); + if (ret) { + pr_err("[GetMaxSustainableClock] failed to get max AC clock from SMC!"); + return ret; + } + + ret = smu_read_smc_arg(smu, clock); + + return ret; +} + +int smu_v11_0_init_max_sustainable_clocks(struct smu_context *smu) +{ + struct smu_11_0_max_sustainable_clocks *max_sustainable_clocks; + int ret = 0; + + max_sustainable_clocks = kzalloc(sizeof(struct smu_11_0_max_sustainable_clocks), + GFP_KERNEL); + smu->smu_table.max_sustainable_clocks = (void *)max_sustainable_clocks; + + max_sustainable_clocks->uclock = smu->smu_table.boot_values.uclk / 100; + max_sustainable_clocks->soc_clock = smu->smu_table.boot_values.socclk / 100; + max_sustainable_clocks->dcef_clock = smu->smu_table.boot_values.dcefclk / 100; + max_sustainable_clocks->display_clock = 0xFFFFFFFF; + max_sustainable_clocks->phy_clock = 0xFFFFFFFF; + max_sustainable_clocks->pixel_clock = 0xFFFFFFFF; + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT)) { + ret = smu_v11_0_get_max_sustainable_clock(smu, + &(max_sustainable_clocks->uclock), + SMU_UCLK); + if (ret) { + pr_err("[%s] failed to get max UCLK from SMC!", + __func__); + return ret; + } + } + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_SOCCLK_BIT)) { + ret = smu_v11_0_get_max_sustainable_clock(smu, + &(max_sustainable_clocks->soc_clock), + SMU_SOCCLK); + if (ret) { + pr_err("[%s] failed to get max SOCCLK from SMC!", + __func__); + return ret; + } + } + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT)) { + ret = smu_v11_0_get_max_sustainable_clock(smu, + &(max_sustainable_clocks->dcef_clock), + SMU_DCEFCLK); + if (ret) { + pr_err("[%s] failed to get max DCEFCLK from SMC!", + __func__); + return ret; + } + + ret = smu_v11_0_get_max_sustainable_clock(smu, + &(max_sustainable_clocks->display_clock), + SMU_DISPCLK); + if (ret) { + pr_err("[%s] failed to get max DISPCLK from SMC!", + __func__); + return ret; + } + ret = smu_v11_0_get_max_sustainable_clock(smu, + &(max_sustainable_clocks->phy_clock), + SMU_PHYCLK); + if (ret) { + pr_err("[%s] failed to get max PHYCLK from SMC!", + __func__); + return ret; + } + ret = smu_v11_0_get_max_sustainable_clock(smu, + &(max_sustainable_clocks->pixel_clock), + SMU_PIXCLK); + if (ret) { + pr_err("[%s] failed to get max PIXCLK from SMC!", + __func__); + return ret; + } + } + + if (max_sustainable_clocks->soc_clock < max_sustainable_clocks->uclock) + max_sustainable_clocks->uclock = max_sustainable_clocks->soc_clock; + + return 0; +} + +uint32_t smu_v11_0_get_max_power_limit(struct smu_context *smu) { + uint32_t od_limit, max_power_limit; + struct smu_11_0_powerplay_table *powerplay_table = NULL; + struct smu_table_context *table_context = &smu->smu_table; + powerplay_table = table_context->power_play_table; + + max_power_limit = smu_get_pptable_power_limit(smu); + + if (!max_power_limit) { + // If we couldn't get the table limit, fall back on first-read value + if (!smu->default_power_limit) + smu->default_power_limit = smu->power_limit; + max_power_limit = smu->default_power_limit; + } + + if (smu->od_enabled) { + od_limit = le32_to_cpu(powerplay_table->overdrive_table.max[SMU_11_0_ODSETTING_POWERPERCENTAGE]); + + pr_debug("ODSETTING_POWERPERCENTAGE: %d (default: %d)\n", od_limit, smu->default_power_limit); + + max_power_limit *= (100 + od_limit); + max_power_limit /= 100; + } + + return max_power_limit; +} + +int smu_v11_0_set_power_limit(struct smu_context *smu, uint32_t n) +{ + int ret = 0; + uint32_t max_power_limit; + + max_power_limit = smu_v11_0_get_max_power_limit(smu); + + if (n > max_power_limit) { + pr_err("New power limit (%d) is over the max allowed %d\n", + n, + max_power_limit); + return -EINVAL; + } + + if (n == 0) + n = smu->default_power_limit; + + if (!smu_feature_is_enabled(smu, SMU_FEATURE_PPT_BIT)) { + pr_err("Setting new power limit is not supported!\n"); + return -EOPNOTSUPP; + } + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetPptLimit, n); + if (ret) { + pr_err("[%s] Set power limit Failed!\n", __func__); + return ret; + } + smu->power_limit = n; + + return 0; +} + +int smu_v11_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; + int asic_clk_id; + + if (clk_id >= SMU_CLK_COUNT || !value) + return -EINVAL; + + asic_clk_id = smu_clk_get_index(smu, clk_id); + if (asic_clk_id < 0) + return -EINVAL; + + /* if don't has GetDpmClockFreq Message, try get current clock by SmuMetrics_t */ + if (smu_msg_get_index(smu, SMU_MSG_GetDpmClockFreq) < 0) + ret = smu_get_current_clk_freq_by_table(smu, clk_id, &freq); + else { + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_GetDpmClockFreq, + (asic_clk_id << 16)); + if (ret) + return ret; + + ret = smu_read_smc_arg(smu, &freq); + if (ret) + return ret; + } + + freq *= 100; + *value = freq; + + return ret; +} + +static int smu_v11_0_set_thermal_range(struct smu_context *smu, + struct smu_temperature_range range) +{ + struct amdgpu_device *adev = smu->adev; + int low = SMU_THERMAL_MINIMUM_ALERT_TEMP; + int high = SMU_THERMAL_MAXIMUM_ALERT_TEMP; + uint32_t val; + + 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); + + if (low > high) + return -EINVAL; + + val = RREG32_SOC15(THM, 0, mmTHM_THERMAL_INT_CTRL); + val = REG_SET_FIELD(val, THM_THERMAL_INT_CTRL, MAX_IH_CREDIT, 5); + val = REG_SET_FIELD(val, THM_THERMAL_INT_CTRL, THERM_IH_HW_ENA, 1); + val = REG_SET_FIELD(val, THM_THERMAL_INT_CTRL, THERM_INTH_MASK, 0); + val = REG_SET_FIELD(val, THM_THERMAL_INT_CTRL, THERM_INTL_MASK, 0); + val = REG_SET_FIELD(val, THM_THERMAL_INT_CTRL, DIG_THERM_INTH, (high & 0xff)); + val = REG_SET_FIELD(val, THM_THERMAL_INT_CTRL, DIG_THERM_INTL, (low & 0xff)); + val = val & (~THM_THERMAL_INT_CTRL__THERM_TRIGGER_MASK_MASK); + + WREG32_SOC15(THM, 0, mmTHM_THERMAL_INT_CTRL, val); + + return 0; +} + +static int smu_v11_0_enable_thermal_alert(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + uint32_t val = 0; + + val |= (1 << THM_THERMAL_INT_ENA__THERM_INTH_CLR__SHIFT); + val |= (1 << THM_THERMAL_INT_ENA__THERM_INTL_CLR__SHIFT); + val |= (1 << THM_THERMAL_INT_ENA__THERM_TRIGGER_CLR__SHIFT); + + WREG32_SOC15(THM, 0, mmTHM_THERMAL_INT_ENA, val); + + return 0; +} + +int smu_v11_0_start_thermal_control(struct smu_context *smu) +{ + int ret = 0; + struct smu_temperature_range range; + struct amdgpu_device *adev = smu->adev; + + if (!smu->pm_enabled) + return ret; + + memcpy(&range, &smu11_thermal_policy[0], sizeof(struct smu_temperature_range)); + + ret = smu_get_thermal_temperature_range(smu, &range); + if (ret) + return ret; + + if (smu->smu_table.thermal_controller_type) { + ret = smu_v11_0_set_thermal_range(smu, range); + if (ret) + return ret; + + ret = smu_v11_0_enable_thermal_alert(smu); + if (ret) + return ret; + + ret = smu_set_thermal_fan_table(smu); + if (ret) + return ret; + } + + adev->pm.dpm.thermal.min_temp = range.min; + adev->pm.dpm.thermal.max_temp = range.max; + adev->pm.dpm.thermal.max_edge_emergency_temp = range.edge_emergency_max; + adev->pm.dpm.thermal.min_hotspot_temp = range.hotspot_min; + adev->pm.dpm.thermal.max_hotspot_crit_temp = range.hotspot_crit_max; + adev->pm.dpm.thermal.max_hotspot_emergency_temp = range.hotspot_emergency_max; + adev->pm.dpm.thermal.min_mem_temp = range.mem_min; + adev->pm.dpm.thermal.max_mem_crit_temp = range.mem_crit_max; + adev->pm.dpm.thermal.max_mem_emergency_temp = range.mem_emergency_max; + + return ret; +} + +int smu_v11_0_stop_thermal_control(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + + WREG32_SOC15(THM, 0, mmTHM_THERMAL_INT_ENA, 0); + + return 0; +} + +static uint16_t convert_to_vddc(uint8_t vid) +{ + return (uint16_t) ((6200 - (vid * 25)) / SMU11_VOLTAGE_SCALE); +} + +static int smu_v11_0_get_gfx_vdd(struct smu_context *smu, uint32_t *value) +{ + struct amdgpu_device *adev = smu->adev; + uint32_t vdd = 0, val_vid = 0; + + if (!value) + return -EINVAL; + val_vid = (RREG32_SOC15(SMUIO, 0, mmSMUSVI0_TEL_PLANE0) & + SMUSVI0_TEL_PLANE0__SVI0_PLANE0_VDDCOR_MASK) >> + SMUSVI0_TEL_PLANE0__SVI0_PLANE0_VDDCOR__SHIFT; + + vdd = (uint32_t)convert_to_vddc((uint8_t)val_vid); + + *value = vdd; + + return 0; + +} + +int smu_v11_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_VDDGFX: + ret = smu_v11_0_get_gfx_vdd(smu, (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; +} + +int +smu_v11_0_display_clock_voltage_request(struct smu_context *smu, + struct pp_display_clock_request + *clock_req) +{ + enum amd_pp_clock_type clk_type = clock_req->clock_type; + int ret = 0; + enum smu_clk_type clk_select = 0; + uint32_t clk_freq = clock_req->clock_freq_in_khz / 1000; + + if (!smu->pm_enabled) + return -EINVAL; + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT) || + smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT)) { + switch (clk_type) { + case amd_pp_dcef_clock: + clk_select = SMU_DCEFCLK; + break; + case amd_pp_disp_clock: + clk_select = SMU_DISPCLK; + break; + case amd_pp_pixel_clock: + clk_select = SMU_PIXCLK; + break; + case amd_pp_phy_clock: + clk_select = SMU_PHYCLK; + break; + case amd_pp_mem_clock: + clk_select = SMU_UCLK; + break; + default: + pr_info("[%s] Invalid Clock Type!", __func__); + ret = -EINVAL; + break; + } + + if (ret) + goto failed; + + if (clk_select == SMU_UCLK && smu->disable_uclk_switch) + return 0; + + ret = smu_set_hard_freq_range(smu, clk_select, clk_freq, 0); + + if(clk_select == SMU_UCLK) + smu->hard_min_uclk_req_from_dal = clk_freq; + } + +failed: + return ret; +} + +int smu_v11_0_gfx_off_control(struct smu_context *smu, bool enable) +{ + int ret = 0; + struct amdgpu_device *adev = smu->adev; + + switch (adev->asic_type) { + case CHIP_VEGA20: + break; + case CHIP_NAVI10: + case CHIP_NAVI14: + case CHIP_NAVI12: + if (!(adev->pm.pp_feature & PP_GFXOFF_MASK)) + return 0; + if (enable) + ret = smu_send_smc_msg(smu, SMU_MSG_AllowGfxOff); + else + ret = smu_send_smc_msg(smu, SMU_MSG_DisallowGfxOff); + break; + default: + break; + } + + return ret; +} + +uint32_t +smu_v11_0_get_fan_control_mode(struct smu_context *smu) +{ + if (!smu_feature_is_enabled(smu, SMU_FEATURE_FAN_CONTROL_BIT)) + return AMD_FAN_CTRL_MANUAL; + else + return AMD_FAN_CTRL_AUTO; +} + +static int +smu_v11_0_auto_fan_control(struct smu_context *smu, bool auto_fan_control) +{ + int ret = 0; + + if (!smu_feature_is_supported(smu, SMU_FEATURE_FAN_CONTROL_BIT)) + return 0; + + ret = smu_feature_set_enabled(smu, SMU_FEATURE_FAN_CONTROL_BIT, auto_fan_control); + if (ret) + pr_err("[%s]%s smc FAN CONTROL feature failed!", + __func__, (auto_fan_control ? "Start" : "Stop")); + + return ret; +} + +static int +smu_v11_0_set_fan_static_mode(struct smu_context *smu, uint32_t mode) +{ + struct amdgpu_device *adev = smu->adev; + + WREG32_SOC15(THM, 0, mmCG_FDO_CTRL2, + REG_SET_FIELD(RREG32_SOC15(THM, 0, mmCG_FDO_CTRL2), + CG_FDO_CTRL2, TMIN, 0)); + WREG32_SOC15(THM, 0, mmCG_FDO_CTRL2, + REG_SET_FIELD(RREG32_SOC15(THM, 0, mmCG_FDO_CTRL2), + CG_FDO_CTRL2, FDO_PWM_MODE, mode)); + + return 0; +} + +int +smu_v11_0_set_fan_speed_percent(struct smu_context *smu, uint32_t speed) +{ + struct amdgpu_device *adev = smu->adev; + uint32_t duty100, duty; + uint64_t tmp64; + + if (speed > 100) + speed = 100; + + if (smu_v11_0_auto_fan_control(smu, 0)) + return -EINVAL; + + duty100 = REG_GET_FIELD(RREG32_SOC15(THM, 0, mmCG_FDO_CTRL1), + CG_FDO_CTRL1, FMAX_DUTY100); + if (!duty100) + return -EINVAL; + + tmp64 = (uint64_t)speed * duty100; + do_div(tmp64, 100); + duty = (uint32_t)tmp64; + + WREG32_SOC15(THM, 0, mmCG_FDO_CTRL0, + REG_SET_FIELD(RREG32_SOC15(THM, 0, mmCG_FDO_CTRL0), + CG_FDO_CTRL0, FDO_STATIC_DUTY, duty)); + + return smu_v11_0_set_fan_static_mode(smu, FDO_PWM_MODE_STATIC); +} + +int +smu_v11_0_set_fan_control_mode(struct smu_context *smu, + uint32_t mode) +{ + int ret = 0; + + switch (mode) { + case AMD_FAN_CTRL_NONE: + ret = smu_v11_0_set_fan_speed_percent(smu, 100); + break; + case AMD_FAN_CTRL_MANUAL: + ret = smu_v11_0_auto_fan_control(smu, 0); + break; + case AMD_FAN_CTRL_AUTO: + ret = smu_v11_0_auto_fan_control(smu, 1); + break; + default: + break; + } + + if (ret) { + pr_err("[%s]Set fan control mode failed!", __func__); + return -EINVAL; + } + + return ret; +} + +int smu_v11_0_set_fan_speed_rpm(struct smu_context *smu, + uint32_t speed) +{ + struct amdgpu_device *adev = smu->adev; + int ret; + uint32_t tach_period, crystal_clock_freq; + + if (!speed) + return -EINVAL; + + ret = smu_v11_0_auto_fan_control(smu, 0); + if (ret) + return ret; + + crystal_clock_freq = amdgpu_asic_get_xclk(adev); + tach_period = 60 * crystal_clock_freq * 10000 / (8 * speed); + WREG32_SOC15(THM, 0, mmCG_TACH_CTRL, + REG_SET_FIELD(RREG32_SOC15(THM, 0, mmCG_TACH_CTRL), + CG_TACH_CTRL, TARGET_PERIOD, + tach_period)); + + ret = smu_v11_0_set_fan_static_mode(smu, FDO_PWM_MODE_STATIC_RPM); + + return ret; +} + +int smu_v11_0_set_xgmi_pstate(struct smu_context *smu, + uint32_t pstate) +{ + int ret = 0; + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetXgmiMode, + pstate ? XGMI_MODE_PSTATE_D0 : XGMI_MODE_PSTATE_D3); + return ret; +} + +#define THM_11_0__SRCID__THM_DIG_THERM_L2H 0 /* ASIC_TEMP > CG_THERMAL_INT.DIG_THERM_INTH */ +#define THM_11_0__SRCID__THM_DIG_THERM_H2L 1 /* ASIC_TEMP < CG_THERMAL_INT.DIG_THERM_INTL */ + +static int smu_v11_0_irq_process(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + uint32_t client_id = entry->client_id; + uint32_t src_id = entry->src_id; + + if (client_id == SOC15_IH_CLIENTID_THM) { + switch (src_id) { + case THM_11_0__SRCID__THM_DIG_THERM_L2H: + pr_warn("GPU over temperature range detected on PCIe %d:%d.%d!\n", + PCI_BUS_NUM(adev->pdev->devfn), + PCI_SLOT(adev->pdev->devfn), + PCI_FUNC(adev->pdev->devfn)); + break; + case THM_11_0__SRCID__THM_DIG_THERM_H2L: + pr_warn("GPU under temperature range detected on PCIe %d:%d.%d!\n", + PCI_BUS_NUM(adev->pdev->devfn), + PCI_SLOT(adev->pdev->devfn), + PCI_FUNC(adev->pdev->devfn)); + break; + default: + pr_warn("GPU under temperature range unknown src id (%d), detected on PCIe %d:%d.%d!\n", + src_id, + PCI_BUS_NUM(adev->pdev->devfn), + PCI_SLOT(adev->pdev->devfn), + PCI_FUNC(adev->pdev->devfn)); + break; + + } + } + + return 0; +} + +static const struct amdgpu_irq_src_funcs smu_v11_0_irq_funcs = +{ + .process = smu_v11_0_irq_process, +}; + +int smu_v11_0_register_irq_handler(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + struct amdgpu_irq_src *irq_src = smu->irq_source; + int ret = 0; + + /* already register */ + if (irq_src) + return 0; + + irq_src = kzalloc(sizeof(struct amdgpu_irq_src), GFP_KERNEL); + if (!irq_src) + return -ENOMEM; + smu->irq_source = irq_src; + + irq_src->funcs = &smu_v11_0_irq_funcs; + + ret = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_THM, + THM_11_0__SRCID__THM_DIG_THERM_L2H, + irq_src); + if (ret) + return ret; + + ret = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_THM, + THM_11_0__SRCID__THM_DIG_THERM_H2L, + irq_src); + if (ret) + return ret; + + return ret; +} + +int smu_v11_0_get_max_sustainable_clocks_by_dc(struct smu_context *smu, + struct pp_smu_nv_clock_table *max_clocks) +{ + struct smu_table_context *table_context = &smu->smu_table; + struct smu_11_0_max_sustainable_clocks *sustainable_clocks = NULL; + + if (!max_clocks || !table_context->max_sustainable_clocks) + return -EINVAL; + + sustainable_clocks = table_context->max_sustainable_clocks; + + max_clocks->dcfClockInKhz = + (unsigned int) sustainable_clocks->dcef_clock * 1000; + max_clocks->displayClockInKhz = + (unsigned int) sustainable_clocks->display_clock * 1000; + max_clocks->phyClockInKhz = + (unsigned int) sustainable_clocks->phy_clock * 1000; + max_clocks->pixelClockInKhz = + (unsigned int) sustainable_clocks->pixel_clock * 1000; + max_clocks->uClockInKhz = + (unsigned int) sustainable_clocks->uclock * 1000; + max_clocks->socClockInKhz = + (unsigned int) sustainable_clocks->soc_clock * 1000; + max_clocks->dscClockInKhz = 0; + max_clocks->dppClockInKhz = 0; + max_clocks->fabricClockInKhz = 0; + + return 0; +} + +int smu_v11_0_set_azalia_d3_pme(struct smu_context *smu) +{ + int ret = 0; + + ret = smu_send_smc_msg(smu, SMU_MSG_BacoAudioD3PME); + + return ret; +} + +static int smu_v11_0_baco_set_armd3_sequence(struct smu_context *smu, enum smu_v11_0_baco_seq baco_seq) +{ + return smu_send_smc_msg_with_param(smu, SMU_MSG_ArmD3, baco_seq); +} + +bool smu_v11_0_baco_is_support(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + struct smu_baco_context *smu_baco = &smu->smu_baco; + uint32_t val; + bool baco_support; + + mutex_lock(&smu_baco->mutex); + baco_support = smu_baco->platform_support; + mutex_unlock(&smu_baco->mutex); + + if (!baco_support) + return false; + + if (!smu_feature_is_enabled(smu, SMU_FEATURE_BACO_BIT)) + return false; + + val = RREG32_SOC15(NBIO, 0, mmRCC_BIF_STRAP0); + if (val & RCC_BIF_STRAP0__STRAP_PX_CAPABLE_MASK) + return true; + + return false; +} + +enum smu_baco_state smu_v11_0_baco_get_state(struct smu_context *smu) +{ + struct smu_baco_context *smu_baco = &smu->smu_baco; + enum smu_baco_state baco_state; + + mutex_lock(&smu_baco->mutex); + baco_state = smu_baco->state; + mutex_unlock(&smu_baco->mutex); + + return baco_state; +} + +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; + int ret = 0; + + if (smu_v11_0_baco_get_state(smu) == state) + return 0; + + 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 + ret = smu_send_smc_msg(smu, SMU_MSG_ExitBaco); + if (ret) + goto out; + + smu_baco->state = state; +out: + mutex_unlock(&smu_baco->mutex); + return ret; +} + +int smu_v11_0_baco_reset(struct smu_context *smu) +{ + int ret = 0; + + 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) + return ret; + + msleep(10); + + ret = smu_v11_0_baco_set_state(smu, SMU_BACO_STATE_EXIT); + if (ret) + return ret; + + return ret; +} + +int smu_v11_0_get_dpm_ultimate_freq(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t *min, uint32_t *max) +{ + int ret = 0, clk_id = 0; + uint32_t param = 0; + + clk_id = smu_clk_get_index(smu, clk_type); + if (clk_id < 0) { + ret = -EINVAL; + goto failed; + } + param = (clk_id & 0xffff) << 16; + + if (max) { + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_GetMaxDpmFreq, param); + if (ret) + goto failed; + ret = smu_read_smc_arg(smu, max); + if (ret) + goto failed; + } + + if (min) { + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_GetMinDpmFreq, param); + if (ret) + goto failed; + ret = smu_read_smc_arg(smu, min); + if (ret) + goto failed; + } + +failed: + return ret; +} + +int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t min, uint32_t max) +{ + int ret = 0, clk_id = 0; + uint32_t param; + + clk_id = smu_clk_get_index(smu, clk_type); + if (clk_id < 0) + return clk_id; + + if (max > 0) { + param = (uint32_t)((clk_id << 16) | (max & 0xffff)); + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetSoftMaxByFreq, + param); + if (ret) + return ret; + } + + if (min > 0) { + param = (uint32_t)((clk_id << 16) | (min & 0xffff)); + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetSoftMinByFreq, + param); + if (ret) + return ret; + } + + return ret; +} + +int smu_v11_0_override_pcie_parameters(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + uint32_t pcie_gen = 0, pcie_width = 0; + int ret; + + if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN4) + pcie_gen = 3; + else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3) + pcie_gen = 2; + else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2) + pcie_gen = 1; + else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1) + pcie_gen = 0; + + /* Bit 31:16: LCLK DPM level. 0 is DPM0, and 1 is DPM1 + * Bit 15:8: PCIE GEN, 0 to 3 corresponds to GEN1 to GEN4 + * Bit 7:0: PCIE lane width, 1 to 7 corresponds is x1 to x32 + */ + if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X16) + pcie_width = 6; + else if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X12) + pcie_width = 5; + else if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X8) + pcie_width = 4; + else if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X4) + pcie_width = 3; + else if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X2) + pcie_width = 2; + else if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X1) + pcie_width = 1; + + ret = smu_update_pcie_parameters(smu, pcie_gen, pcie_width); + + if (ret) + pr_err("[%s] Attempt to override pcie params failed!\n", __func__); + + return ret; + +} + +int smu_v11_0_set_default_od_settings(struct smu_context *smu, bool initialize, size_t overdrive_table_size) +{ + struct smu_table_context *table_context = &smu->smu_table; + int ret = 0; + + if (initialize) { + if (table_context->overdrive_table) { + return -EINVAL; + } + table_context->overdrive_table = kzalloc(overdrive_table_size, GFP_KERNEL); + if (!table_context->overdrive_table) { + return -ENOMEM; + } + ret = smu_update_table(smu, SMU_TABLE_OVERDRIVE, 0, table_context->overdrive_table, false); + if (ret) { + pr_err("Failed to export overdrive table!\n"); + return ret; + } + } + ret = smu_update_table(smu, SMU_TABLE_OVERDRIVE, 0, table_context->overdrive_table, true); + if (ret) { + pr_err("Failed to import overdrive table!\n"); + return ret; + } + return ret; +} diff --git a/drivers/gpu/drm/amd/powerplay/smu_v12_0.c b/drivers/gpu/drm/amd/powerplay/smu_v12_0.c new file mode 100644 index 000000000000..139dd737eaa5 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/smu_v12_0.c @@ -0,0 +1,447 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include "pp_debug.h" +#include <linux/firmware.h> +#include "amdgpu.h" +#include "amdgpu_smu.h" +#include "smu_internal.h" +#include "atomfirmware.h" +#include "amdgpu_atomfirmware.h" +#include "smu_v12_0.h" +#include "soc15_common.h" +#include "atom.h" + +#include "asic_reg/mp/mp_12_0_0_offset.h" +#include "asic_reg/mp/mp_12_0_0_sh_mask.h" + +#define smnMP1_FIRMWARE_FLAGS 0x3010024 + +#define mmSMUIO_GFX_MISC_CNTL 0x00c8 +#define mmSMUIO_GFX_MISC_CNTL_BASE_IDX 0 +#define SMUIO_GFX_MISC_CNTL__PWR_GFXOFF_STATUS_MASK 0x00000006L +#define SMUIO_GFX_MISC_CNTL__PWR_GFXOFF_STATUS__SHIFT 0x1 + +int smu_v12_0_send_msg_without_waiting(struct smu_context *smu, + uint16_t msg) +{ + struct amdgpu_device *adev = smu->adev; + + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_66, msg); + return 0; +} + +int smu_v12_0_read_arg(struct smu_context *smu, uint32_t *arg) +{ + struct amdgpu_device *adev = smu->adev; + + *arg = RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_82); + return 0; +} + +int smu_v12_0_wait_for_response(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + uint32_t cur_value, i; + + 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; + 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; +} + +int smu_v12_0_send_msg(struct smu_context *smu, uint16_t msg) +{ + struct amdgpu_device *adev = smu->adev; + int ret = 0, index = 0; + + index = smu_msg_get_index(smu, msg); + if (index < 0) + return index; + + smu_v12_0_wait_for_response(smu); + + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90, 0); + + smu_v12_0_send_msg_without_waiting(smu, (uint16_t)index); + + ret = smu_v12_0_wait_for_response(smu); + + if (ret) + pr_err("Failed to send message 0x%x, response 0x%x\n", index, + ret); + + return ret; + +} + +int +smu_v12_0_send_msg_with_param(struct smu_context *smu, uint16_t msg, + uint32_t param) +{ + struct amdgpu_device *adev = smu->adev; + int ret = 0, index = 0; + + index = smu_msg_get_index(smu, msg); + if (index < 0) + 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); + + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90, 0); + + WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_82, param); + + smu_v12_0_send_msg_without_waiting(smu, (uint16_t)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); + + return ret; +} + +int smu_v12_0_check_fw_status(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + uint32_t mp1_fw_flags; + + mp1_fw_flags = RREG32_PCIE(MP1_Public | + (smnMP1_FIRMWARE_FLAGS & 0xffffffff)); + + if ((mp1_fw_flags & MP1_FIRMWARE_FLAGS__INTERRUPTS_ENABLED_MASK) >> + MP1_FIRMWARE_FLAGS__INTERRUPTS_ENABLED__SHIFT) + return 0; + + return -EIO; +} + +int smu_v12_0_check_fw_version(struct smu_context *smu) +{ + uint32_t if_version = 0xff, smu_version = 0xff; + uint16_t smu_major; + uint8_t smu_minor, smu_debug; + int ret = 0; + + ret = smu_get_smc_version(smu, &if_version, &smu_version); + if (ret) + return ret; + + smu_major = (smu_version >> 16) & 0xffff; + smu_minor = (smu_version >> 8) & 0xff; + smu_debug = (smu_version >> 0) & 0xff; + + /* + * 1. if_version mismatch is not critical as our fw is designed + * to be backward compatible. + * 2. New fw usually brings some optimizations. But that's visible + * only on the paired driver. + * Considering above, we just leave user a warning message instead + * of halt driver loading. + */ + if (if_version != smu->smc_if_version) { + pr_info("smu driver if version = 0x%08x, smu fw if version = 0x%08x, " + "smu fw version = 0x%08x (%d.%d.%d)\n", + smu->smc_if_version, if_version, + smu_version, smu_major, smu_minor, smu_debug); + pr_warn("SMU driver if version not matched\n"); + } + + return ret; +} + +int smu_v12_0_powergate_sdma(struct smu_context *smu, bool gate) +{ + if (!(smu->adev->flags & AMD_IS_APU)) + return 0; + + if (gate) + return smu_send_smc_msg(smu, SMU_MSG_PowerDownSdma); + else + return smu_send_smc_msg(smu, SMU_MSG_PowerUpSdma); +} + +int smu_v12_0_powergate_vcn(struct smu_context *smu, bool gate) +{ + if (!(smu->adev->flags & AMD_IS_APU)) + return 0; + + if (gate) + return smu_send_smc_msg(smu, SMU_MSG_PowerDownVcn); + else + return smu_send_smc_msg(smu, SMU_MSG_PowerUpVcn); +} + +int smu_v12_0_set_gfx_cgpg(struct smu_context *smu, bool enable) +{ + if (!(smu->adev->pg_flags & AMD_PG_SUPPORT_GFX_PG)) + return 0; + + return smu_v12_0_send_msg_with_param(smu, + SMU_MSG_SetGfxCGPG, enable ? 1 : 0); +} + +/** + * smu_v12_0_get_gfxoff_status - get gfxoff status + * + * @smu: amdgpu_device pointer + * + * This function will be used to get gfxoff status + * + * Returns 0=GFXOFF(default). + * Returns 1=Transition out of GFX State. + * Returns 2=Not in GFXOFF. + * Returns 3=Transition into GFXOFF. + */ +uint32_t smu_v12_0_get_gfxoff_status(struct smu_context *smu) +{ + uint32_t reg; + uint32_t gfxOff_Status = 0; + struct amdgpu_device *adev = smu->adev; + + reg = RREG32_SOC15(SMUIO, 0, mmSMUIO_GFX_MISC_CNTL); + gfxOff_Status = (reg & SMUIO_GFX_MISC_CNTL__PWR_GFXOFF_STATUS_MASK) + >> SMUIO_GFX_MISC_CNTL__PWR_GFXOFF_STATUS__SHIFT; + + return gfxOff_Status; +} + +int smu_v12_0_gfx_off_control(struct smu_context *smu, bool enable) +{ + int ret = 0, timeout = 500; + + if (enable) { + ret = smu_send_smc_msg(smu, SMU_MSG_AllowGfxOff); + + } else { + ret = smu_send_smc_msg(smu, SMU_MSG_DisallowGfxOff); + + /* confirm gfx is back to "on" state, timeout is 0.5 second */ + while (!(smu_v12_0_get_gfxoff_status(smu) == 2)) { + msleep(1); + timeout--; + if (timeout == 0) { + DRM_ERROR("disable gfxoff timeout and failed!\n"); + break; + } + } + } + + return ret; +} + +int smu_v12_0_init_smc_tables(struct smu_context *smu) +{ + struct smu_table_context *smu_table = &smu->smu_table; + struct smu_table *tables = NULL; + + if (smu_table->tables) + return -EINVAL; + + tables = kcalloc(SMU_TABLE_COUNT, sizeof(struct smu_table), + GFP_KERNEL); + if (!tables) + return -ENOMEM; + + smu_table->tables = tables; + + return smu_tables_init(smu, tables); +} + +int smu_v12_0_fini_smc_tables(struct smu_context *smu) +{ + struct smu_table_context *smu_table = &smu->smu_table; + + if (!smu_table->tables) + return -EINVAL; + + kfree(smu_table->clocks_table); + kfree(smu_table->tables); + + smu_table->clocks_table = NULL; + smu_table->tables = NULL; + + return 0; +} + +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 -EINVAL; + + if (!table->cpu_addr) + return -EINVAL; + + return smu_update_table(smu, SMU_TABLE_DPMCLOCKS, 0, smu_table->clocks_table, false); +} + +int smu_v12_0_get_dpm_ultimate_freq(struct smu_context *smu, enum smu_clk_type clk_type, + uint32_t *min, uint32_t *max) +{ + int ret = 0; + uint32_t mclk_mask, soc_mask; + + if (max) { + ret = smu_get_profiling_clk_mask(smu, AMD_DPM_FORCED_LEVEL_PROFILE_PEAK, + NULL, + &mclk_mask, + &soc_mask); + if (ret) + goto failed; + + switch (clk_type) { + case SMU_GFXCLK: + case SMU_SCLK: + ret = smu_send_smc_msg(smu, SMU_MSG_GetMaxGfxclkFrequency); + if (ret) { + pr_err("Attempt to get max GX frequency from SMC Failed !\n"); + goto failed; + } + ret = smu_read_smc_arg(smu, max); + if (ret) + goto failed; + break; + case SMU_UCLK: + case SMU_FCLK: + case SMU_MCLK: + ret = smu_get_dpm_clk_limited(smu, clk_type, mclk_mask, max); + if (ret) + goto failed; + break; + case SMU_SOCCLK: + ret = smu_get_dpm_clk_limited(smu, clk_type, soc_mask, max); + if (ret) + goto failed; + break; + default: + ret = -EINVAL; + goto failed; + } + } + + if (min) { + switch (clk_type) { + case SMU_GFXCLK: + case SMU_SCLK: + ret = smu_send_smc_msg(smu, SMU_MSG_GetMinGfxclkFrequency); + if (ret) { + pr_err("Attempt to get min GX frequency from SMC Failed !\n"); + goto failed; + } + ret = smu_read_smc_arg(smu, min); + if (ret) + goto failed; + break; + case SMU_UCLK: + case SMU_FCLK: + case SMU_MCLK: + ret = smu_get_dpm_clk_limited(smu, clk_type, 0, min); + if (ret) + goto failed; + break; + case SMU_SOCCLK: + ret = smu_get_dpm_clk_limited(smu, clk_type, 0, min); + if (ret) + goto failed; + break; + default: + ret = -EINVAL; + goto failed; + } + } +failed: + return ret; +} + +int smu_v12_0_mode2_reset(struct smu_context *smu){ + return smu_v12_0_send_msg_with_param(smu, SMU_MSG_GfxDeviceDriverReset, SMU_RESET_MODE_2); +} + +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 ret = 0; + + if (max < min) + return -EINVAL; + + switch (clk_type) { + case SMU_GFXCLK: + case SMU_SCLK: + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetHardMinGfxClk, min); + if (ret) + return ret; + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetSoftMaxGfxClk, max); + if (ret) + return ret; + break; + case SMU_FCLK: + case SMU_MCLK: + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetHardMinFclkByFreq, min); + if (ret) + return ret; + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetSoftMaxFclkByFreq, max); + if (ret) + return ret; + break; + case SMU_SOCCLK: + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetHardMinSocclkByFreq, min); + if (ret) + return ret; + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetSoftMaxSocclkByFreq, max); + if (ret) + return ret; + break; + case SMU_VCLK: + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetHardMinVcn, min); + if (ret) + return ret; + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetSoftMaxVcn, max); + if (ret) + return ret; + break; + default: + return -EINVAL; + } + + 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 669bd0c2a16c..15590fd86ef4 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/ci_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/ci_smumgr.c @@ -25,6 +25,7 @@ #include <linux/fb.h> #include "linux/delay.h" #include <linux/types.h> +#include <linux/pci.h> #include "smumgr.h" #include "pp_debug.h" @@ -2254,6 +2255,8 @@ static uint32_t ci_get_offsetof(uint32_t type, uint32_t member) return offsetof(SMU7_SoftRegisters, VoltageChangeTimeout); case AverageGraphicsActivity: return offsetof(SMU7_SoftRegisters, AverageGraphicsA); + case AverageMemoryActivity: + return offsetof(SMU7_SoftRegisters, AverageMemoryA); case PreVBlankGap: return offsetof(SMU7_SoftRegisters, PreVBlankGap); case VBlankTimeout: @@ -2702,8 +2705,6 @@ static int ci_initialize_mc_reg_table(struct pp_hwmgr *hwmgr) cgs_write_register(hwmgr->device, mmMC_SEQ_PMG_CMD_MRS2_LP, cgs_read_register(hwmgr->device, mmMC_PMG_CMD_MRS2)); cgs_write_register(hwmgr->device, mmMC_SEQ_WR_CTL_2_LP, cgs_read_register(hwmgr->device, mmMC_SEQ_WR_CTL_2)); - memset(table, 0x00, sizeof(pp_atomctrl_mc_reg_table)); - result = atomctrl_initialize_mc_reg_table(hwmgr, module_index, table); if (0 == result) @@ -2933,6 +2934,7 @@ static int ci_update_smc_table(struct pp_hwmgr *hwmgr, uint32_t type) } const struct pp_smumgr_func ci_smu_funcs = { + .name = "ci_smu", .smu_init = ci_smu_init, .smu_fini = ci_smu_fini, .start_smu = ci_start_smu, diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/fiji_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/fiji_smumgr.c index bc8375cbf297..da025b1d302d 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/fiji_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/fiji_smumgr.c @@ -2304,6 +2304,8 @@ static uint32_t fiji_get_offsetof(uint32_t type, uint32_t member) return offsetof(SMU73_SoftRegisters, VoltageChangeTimeout); case AverageGraphicsActivity: return offsetof(SMU73_SoftRegisters, AverageGraphicsActivity); + case AverageMemoryActivity: + return offsetof(SMU73_SoftRegisters, AverageMemoryActivity); case PreVBlankGap: return offsetof(SMU73_SoftRegisters, PreVBlankGap); case VBlankTimeout: @@ -2641,6 +2643,7 @@ static int fiji_update_dpm_settings(struct pp_hwmgr *hwmgr, } const struct pp_smumgr_func fiji_smu_funcs = { + .name = "fiji_smu", .smu_init = &fiji_smu_init, .smu_fini = &smu7_smu_fini, .start_smu = &fiji_start_smu, diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/iceland_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/iceland_smumgr.c index 375ccf6ff5f2..732005c03a82 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/iceland_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/iceland_smumgr.c @@ -25,6 +25,7 @@ #include "pp_debug.h" #include <linux/types.h> #include <linux/kernel.h> +#include <linux/pci.h> #include <linux/slab.h> #include <linux/gfp.h> @@ -2219,6 +2220,8 @@ static uint32_t iceland_get_offsetof(uint32_t type, uint32_t member) return offsetof(SMU71_SoftRegisters, VoltageChangeTimeout); case AverageGraphicsActivity: return offsetof(SMU71_SoftRegisters, AverageGraphicsActivity); + case AverageMemoryActivity: + return offsetof(SMU71_SoftRegisters, AverageMemoryActivity); case PreVBlankGap: return offsetof(SMU71_SoftRegisters, PreVBlankGap); case VBlankTimeout: @@ -2631,8 +2634,6 @@ static int iceland_initialize_mc_reg_table(struct pp_hwmgr *hwmgr) cgs_write_register(hwmgr->device, mmMC_SEQ_PMG_CMD_MRS2_LP, cgs_read_register(hwmgr->device, mmMC_PMG_CMD_MRS2)); cgs_write_register(hwmgr->device, mmMC_SEQ_WR_CTL_2_LP, cgs_read_register(hwmgr->device, mmMC_SEQ_WR_CTL_2)); - memset(table, 0x00, sizeof(pp_atomctrl_mc_reg_table)); - result = atomctrl_initialize_mc_reg_table(hwmgr, module_index, table); if (0 == result) @@ -2659,6 +2660,7 @@ static bool iceland_is_dpm_running(struct pp_hwmgr *hwmgr) } const struct pp_smumgr_func iceland_smu_funcs = { + .name = "iceland_smu", .smu_init = &iceland_smu_init, .smu_fini = &smu7_smu_fini, .start_smu = &iceland_start_smu, diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/polaris10_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/polaris10_smumgr.c index 2d4cfe14f72e..23c12018dbc1 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/polaris10_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/polaris10_smumgr.c @@ -21,6 +21,8 @@ * */ +#include <linux/pci.h> + #include "pp_debug.h" #include "smumgr.h" #include "smu74.h" @@ -653,7 +655,7 @@ static int polaris10_populate_smc_mvdd_table(struct pp_hwmgr *hwmgr, count = SMU_MAX_SMIO_LEVELS; for (level = 0; level < count; level++) { table->SmioTable2.Pattern[level].Voltage = - PP_HOST_TO_SMC_US(data->mvdd_voltage_table.entries[count].value * VOLTAGE_SCALE); + PP_HOST_TO_SMC_US(data->mvdd_voltage_table.entries[level].value * VOLTAGE_SCALE); /* Index into DpmTable.Smio. Drive bits from Smio entry to get this voltage level.*/ table->SmioTable2.Pattern[level].Smio = (uint8_t) level; @@ -2092,6 +2094,10 @@ static int polaris10_thermal_setup_fan_table(struct pp_hwmgr *hwmgr) return 0; } + /* use hardware fan control */ + if (hwmgr->thermal_controller.use_hw_fan_control) + return 0; + tmp64 = hwmgr->thermal_controller.advanceFanControlParameters. usPWMMin * duty100; do_div(tmp64, 10000); @@ -2313,6 +2319,8 @@ static uint32_t polaris10_get_offsetof(uint32_t type, uint32_t member) return offsetof(SMU74_SoftRegisters, VoltageChangeTimeout); case AverageGraphicsActivity: return offsetof(SMU74_SoftRegisters, AverageGraphicsActivity); + case AverageMemoryActivity: + return offsetof(SMU74_SoftRegisters, AverageMemoryActivity); case PreVBlankGap: return offsetof(SMU74_SoftRegisters, PreVBlankGap); case VBlankTimeout: @@ -2548,6 +2556,7 @@ static int polaris10_update_dpm_settings(struct pp_hwmgr *hwmgr, } const struct pp_smumgr_func polaris10_smu_funcs = { + .name = "polaris10_smu", .smu_init = polaris10_smu_init, .smu_fini = smu7_smu_fini, .start_smu = polaris10_start_smu, diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/smu10_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/smu10_smumgr.c index d111dd4e03d7..aa0ee2b46135 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/smu10_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/smu10_smumgr.c @@ -21,6 +21,8 @@ * */ +#include <linux/pci.h> + #include "smumgr.h" #include "smu10_inc.h" #include "soc15_common.h" @@ -116,6 +118,7 @@ static int smu10_copy_table_from_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;); @@ -133,6 +136,9 @@ static int smu10_copy_table_from_smc(struct pp_hwmgr *hwmgr, PPSMC_MSG_TransferTableSmu2Dram, priv->smu_tables.entry[table_id].table_id); + /* flush hdp cache */ + adev->nbio.funcs->hdp_flush(adev, NULL); + memcpy(table, (uint8_t *)priv->smu_tables.entry[table_id].table, priv->smu_tables.entry[table_id].size); @@ -212,6 +218,10 @@ static int smu10_start_smu(struct pp_hwmgr *hwmgr) hwmgr->smu_version = smu10_read_arg_from_smc(hwmgr); adev->pm.fw_version = hwmgr->smu_version >> 8; + if (adev->rev_id < 0x8 && adev->pdev->device != 0x15d8 && + adev->pm.fw_version < 0x1e45) + adev->pm.pp_feature &= ~PP_GFXOFF_MASK; + if (smu10_verify_smc_interface(hwmgr)) return -EINVAL; @@ -287,6 +297,7 @@ static int smu10_smc_table_manager(struct pp_hwmgr *hwmgr, uint8_t *table, uint1 const struct pp_smumgr_func smu10_smu_funcs = { + .name = "smu10_smu", .smu_init = &smu10_smu_init, .smu_fini = &smu10_smu_fini, .start_smu = &smu10_start_smu, diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/smu8_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/smu8_smumgr.c index e2787e14a500..7dca04a89217 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/smu8_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/smu8_smumgr.c @@ -177,12 +177,10 @@ static int smu8_load_mec_firmware(struct pp_hwmgr *hwmgr) uint32_t tmp; int ret = 0; struct cgs_firmware_info info = {0}; - struct smu8_smumgr *smu8_smu; if (hwmgr == NULL || hwmgr->device == NULL) return -EINVAL; - smu8_smu = hwmgr->smu_backend; ret = cgs_get_firmware_info(hwmgr->device, CGS_UCODE_ID_CP_MEC, &info); @@ -722,16 +720,17 @@ static int smu8_request_smu_load_fw(struct pp_hwmgr *hwmgr) static int smu8_start_smu(struct pp_hwmgr *hwmgr) { - struct amdgpu_device *adev = hwmgr->adev; + struct amdgpu_device *adev; uint32_t index = SMN_MP1_SRAM_START_ADDR + SMU8_FIRMWARE_HEADER_LOCATION + offsetof(struct SMU8_Firmware_Header, Version); - if (hwmgr == NULL || hwmgr->device == NULL) return -EINVAL; + adev = hwmgr->adev; + cgs_write_register(hwmgr->device, mmMP0PUB_IND_INDEX, index); hwmgr->smu_version = cgs_read_register(hwmgr->device, mmMP0PUB_IND_DATA); pr_info("smu version %02d.%02d.%02d\n", @@ -881,6 +880,7 @@ static bool smu8_is_dpm_running(struct pp_hwmgr *hwmgr) } const struct pp_smumgr_func smu8_smu_funcs = { + .name = "smu8_smu", .smu_init = smu8_smu_init, .smu_fini = smu8_smu_fini, .start_smu = smu8_start_smu, diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/tonga_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/tonga_smumgr.c index 3ed6c5f1e5cf..f19bac7ef7ba 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/tonga_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/tonga_smumgr.c @@ -23,6 +23,7 @@ #include "pp_debug.h" #include <linux/types.h> #include <linux/kernel.h> +#include <linux/pci.h> #include <linux/slab.h> #include <linux/gfp.h> @@ -2611,6 +2612,8 @@ static uint32_t tonga_get_offsetof(uint32_t type, uint32_t member) return offsetof(SMU72_SoftRegisters, VoltageChangeTimeout); case AverageGraphicsActivity: return offsetof(SMU72_SoftRegisters, AverageGraphicsActivity); + case AverageMemoryActivity: + return offsetof(SMU72_SoftRegisters, AverageMemoryActivity); case PreVBlankGap: return offsetof(SMU72_SoftRegisters, PreVBlankGap); case VBlankTimeout: @@ -3114,8 +3117,6 @@ static int tonga_initialize_mc_reg_table(struct pp_hwmgr *hwmgr) cgs_write_register(hwmgr->device, mmMC_SEQ_WR_CTL_2_LP, cgs_read_register(hwmgr->device, mmMC_SEQ_WR_CTL_2)); - memset(table, 0x00, sizeof(pp_atomctrl_mc_reg_table)); - result = atomctrl_initialize_mc_reg_table(hwmgr, module_index, table); if (!result) @@ -3238,6 +3239,7 @@ static int tonga_update_dpm_settings(struct pp_hwmgr *hwmgr, } const struct pp_smumgr_func tonga_smu_funcs = { + .name = "tonga_smu", .smu_init = &tonga_smu_init, .smu_fini = &smu7_smu_fini, .start_smu = &tonga_start_smu, diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.c index c81acc3192ad..0f3836fd9666 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.c @@ -21,6 +21,8 @@ * */ +#include <linux/pci.h> + #include "smumgr.h" #include "vega10_inc.h" #include "soc15_common.h" @@ -37,6 +39,7 @@ static int vega10_copy_table_from_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; PP_ASSERT_WITH_CODE(table_id < MAX_SMU_TABLE, "Invalid SMU Table ID!", return -EINVAL); @@ -54,6 +57,9 @@ static int vega10_copy_table_from_smc(struct pp_hwmgr *hwmgr, PPSMC_MSG_TransferTableSmu2Dram, priv->smu_tables.entry[table_id].table_id); + /* flush hdp cache */ + adev->nbio.funcs->hdp_flush(adev, NULL); + memcpy(table, priv->smu_tables.entry[table_id].table, priv->smu_tables.entry[table_id].size); @@ -346,6 +352,7 @@ static int vega10_smc_table_manager(struct pp_hwmgr *hwmgr, uint8_t *table, } const struct pp_smumgr_func vega10_smu_funcs = { + .name = "vega10_smu", .smu_init = &vega10_smu_init, .smu_fini = &vega10_smu_fini, .start_smu = &vega10_start_smu, diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.c index ddb801517667..90c782c132d2 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.c @@ -42,6 +42,7 @@ static int vega12_copy_table_from_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); @@ -64,6 +65,9 @@ static int vega12_copy_table_from_smc(struct pp_hwmgr *hwmgr, "[CopyTableFromSMC] Attempt to Transfer Table From SMU Failed!", return -EINVAL); + /* flush hdp cache */ + adev->nbio.funcs->hdp_flush(adev, NULL); + memcpy(table, priv->smu_tables.entry[table_id].table, priv->smu_tables.entry[table_id].size); @@ -287,8 +291,26 @@ static int vega12_smu_init(struct pp_hwmgr *hwmgr) priv->smu_tables.entry[TABLE_OVERDRIVE].version = 0x01; priv->smu_tables.entry[TABLE_OVERDRIVE].size = sizeof(OverDriveTable_t); + /* allocate space for SMU_METRICS table */ + ret = amdgpu_bo_create_kernel((struct amdgpu_device *)hwmgr->adev, + sizeof(SmuMetrics_t), + PAGE_SIZE, + AMDGPU_GEM_DOMAIN_VRAM, + &priv->smu_tables.entry[TABLE_SMU_METRICS].handle, + &priv->smu_tables.entry[TABLE_SMU_METRICS].mc_addr, + &priv->smu_tables.entry[TABLE_SMU_METRICS].table); + if (ret) + goto err4; + + priv->smu_tables.entry[TABLE_SMU_METRICS].version = 0x01; + priv->smu_tables.entry[TABLE_SMU_METRICS].size = sizeof(SmuMetrics_t); + return 0; +err4: + amdgpu_bo_free_kernel(&priv->smu_tables.entry[TABLE_OVERDRIVE].handle, + &priv->smu_tables.entry[TABLE_OVERDRIVE].mc_addr, + &priv->smu_tables.entry[TABLE_OVERDRIVE].table); err3: amdgpu_bo_free_kernel(&priv->smu_tables.entry[TABLE_AVFS_FUSE_OVERRIDE].handle, &priv->smu_tables.entry[TABLE_AVFS_FUSE_OVERRIDE].mc_addr, @@ -334,6 +356,9 @@ static int vega12_smu_fini(struct pp_hwmgr *hwmgr) amdgpu_bo_free_kernel(&priv->smu_tables.entry[TABLE_OVERDRIVE].handle, &priv->smu_tables.entry[TABLE_OVERDRIVE].mc_addr, &priv->smu_tables.entry[TABLE_OVERDRIVE].table); + amdgpu_bo_free_kernel(&priv->smu_tables.entry[TABLE_SMU_METRICS].handle, + &priv->smu_tables.entry[TABLE_SMU_METRICS].mc_addr, + &priv->smu_tables.entry[TABLE_SMU_METRICS].table); kfree(hwmgr->smu_backend); hwmgr->smu_backend = NULL; } @@ -365,6 +390,7 @@ static int vega12_smc_table_manager(struct pp_hwmgr *hwmgr, uint8_t *table, } const struct pp_smumgr_func vega12_smu_funcs = { + .name = "vega12_smu", .smu_init = &vega12_smu_init, .smu_fini = &vega12_smu_fini, .start_smu = &vega12_start_smu, diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.c index ba00744c3413..f604612f411f 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.c @@ -44,7 +44,7 @@ #define smnMP0_FW_INTF 0x30101c0 #define smnMP1_PUB_CTRL 0x3010b14 -static bool vega20_is_smc_ram_running(struct pp_hwmgr *hwmgr) +bool vega20_is_smc_ram_running(struct pp_hwmgr *hwmgr) { struct amdgpu_device *adev = hwmgr->adev; uint32_t mp1_fw_flags; @@ -163,6 +163,7 @@ static int vega20_copy_table_from_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, @@ -187,6 +188,9 @@ static int vega20_copy_table_from_smc(struct pp_hwmgr *hwmgr, "[CopyTableFromSMC] Attempt to Transfer Table From SMU Failed!", return ret); + /* flush hdp cache */ + adev->nbio.funcs->hdp_flush(adev, NULL); + memcpy(table, priv->smu_tables.entry[table_id].table, priv->smu_tables.entry[table_id].size); @@ -266,6 +270,7 @@ int vega20_get_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; PP_ASSERT_WITH_CODE((ret = vega20_send_msg_to_smc_with_parameter(hwmgr, @@ -284,6 +289,9 @@ int vega20_get_activity_monitor_coeff(struct pp_hwmgr *hwmgr, "[GetActivityMonitor] Attempt to Transfer Table From SMU Failed!", return ret); + /* flush hdp cache */ + adev->nbio.funcs->hdp_flush(adev, NULL); + memcpy(table, priv->smu_tables.entry[TABLE_ACTIVITY_MONITOR_COEFF].table, priv->smu_tables.entry[TABLE_ACTIVITY_MONITOR_COEFF].size); @@ -367,6 +375,26 @@ static int vega20_set_tools_address(struct pp_hwmgr *hwmgr) return ret; } +int vega20_set_pptable_driver_address(struct pp_hwmgr *hwmgr) +{ + struct vega20_smumgr *priv = + (struct vega20_smumgr *)(hwmgr->smu_backend); + int ret = 0; + + PP_ASSERT_WITH_CODE((ret = vega20_send_msg_to_smc_with_parameter(hwmgr, + PPSMC_MSG_SetDriverDramAddrHigh, + upper_32_bits(priv->smu_tables.entry[TABLE_PPTABLE].mc_addr))) == 0, + "[SetPPtabeDriverAddress] Attempt to Set Dram Addr High Failed!", + return ret); + PP_ASSERT_WITH_CODE((ret = vega20_send_msg_to_smc_with_parameter(hwmgr, + PPSMC_MSG_SetDriverDramAddrLow, + lower_32_bits(priv->smu_tables.entry[TABLE_PPTABLE].mc_addr))) == 0, + "[SetPPtabeDriverAddress] Attempt to Set Dram Addr Low Failed!", + return ret); + + return ret; +} + static int vega20_smu_init(struct pp_hwmgr *hwmgr) { struct vega20_smumgr *priv; @@ -572,6 +600,7 @@ static int vega20_smc_table_manager(struct pp_hwmgr *hwmgr, uint8_t *table, } const struct pp_smumgr_func vega20_smu_funcs = { + .name = "vega20_smu", .smu_init = &vega20_smu_init, .smu_fini = &vega20_smu_fini, .start_smu = &vega20_start_smu, diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.h b/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.h index 77349c3f0162..62ebbfd6068f 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.h +++ b/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.h @@ -55,6 +55,9 @@ int vega20_set_activity_monitor_coeff(struct pp_hwmgr *hwmgr, uint8_t *table, uint16_t workload_type); int vega20_get_activity_monitor_coeff(struct pp_hwmgr *hwmgr, uint8_t *table, uint16_t workload_type); +int vega20_set_pptable_driver_address(struct pp_hwmgr *hwmgr); + +bool vega20_is_smc_ram_running(struct pp_hwmgr *hwmgr); #endif diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c index 1e69300f6175..ae18fbcb26fb 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c @@ -456,7 +456,7 @@ static int vegam_populate_smc_mvdd_table(struct pp_hwmgr *hwmgr, count = SMU_MAX_SMIO_LEVELS; for (level = 0; level < count; level++) { table->SmioTable2.Pattern[level].Voltage = PP_HOST_TO_SMC_US( - data->mvdd_voltage_table.entries[count].value * VOLTAGE_SCALE); + data->mvdd_voltage_table.entries[level].value * VOLTAGE_SCALE); /* Index into DpmTable.Smio. Drive bits from Smio entry to get this voltage level.*/ table->SmioTable2.Pattern[level].Smio = (uint8_t) level; @@ -2167,6 +2167,8 @@ static uint32_t vegam_get_offsetof(uint32_t type, uint32_t member) return offsetof(SMU75_SoftRegisters, VoltageChangeTimeout); case AverageGraphicsActivity: return offsetof(SMU75_SoftRegisters, AverageGraphicsActivity); + case AverageMemoryActivity: + return offsetof(SMU75_SoftRegisters, AverageMemoryActivity); case PreVBlankGap: return offsetof(SMU75_SoftRegisters, PreVBlankGap); case VBlankTimeout: @@ -2277,6 +2279,7 @@ static int vegam_thermal_setup_fan_table(struct pp_hwmgr *hwmgr) } const struct pp_smumgr_func vegam_smu_funcs = { + .name = "vegam_smu", .smu_init = vegam_smu_init, .smu_fini = smu7_smu_fini, .start_smu = vegam_start_smu, diff --git a/drivers/gpu/drm/amd/powerplay/vega20_ppt.c b/drivers/gpu/drm/amd/powerplay/vega20_ppt.c new file mode 100644 index 000000000000..0b4892833808 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/vega20_ppt.c @@ -0,0 +1,3269 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "pp_debug.h" +#include <linux/firmware.h> +#include "amdgpu.h" +#include "amdgpu_smu.h" +#include "smu_internal.h" +#include "atomfirmware.h" +#include "amdgpu_atomfirmware.h" +#include "smu_v11_0.h" +#include "smu11_driver_if.h" +#include "soc15_common.h" +#include "atom.h" +#include "power_state.h" +#include "vega20_ppt.h" +#include "vega20_pptable.h" +#include "vega20_ppsmc.h" +#include "nbio/nbio_7_4_sh_mask.h" +#include "asic_reg/thm/thm_11_0_2_offset.h" +#include "asic_reg/thm/thm_11_0_2_sh_mask.h" + +#define smnPCIE_LC_SPEED_CNTL 0x11140290 +#define smnPCIE_LC_LINK_WIDTH_CNTL 0x11140288 + +#define CTF_OFFSET_EDGE 5 +#define CTF_OFFSET_HOTSPOT 5 +#define CTF_OFFSET_HBM 5 + +#define MSG_MAP(msg) \ + [SMU_MSG_##msg] = {1, PPSMC_MSG_##msg} + +#define SMC_DPM_FEATURE (FEATURE_DPM_PREFETCHER_MASK | \ + FEATURE_DPM_GFXCLK_MASK | \ + FEATURE_DPM_UCLK_MASK | \ + FEATURE_DPM_SOCCLK_MASK | \ + FEATURE_DPM_UVD_MASK | \ + FEATURE_DPM_VCE_MASK | \ + FEATURE_DPM_MP0CLK_MASK | \ + FEATURE_DPM_LINK_MASK | \ + FEATURE_DPM_DCEFCLK_MASK) + +static struct smu_11_0_cmn2aisc_mapping vega20_message_map[SMU_MSG_MAX_COUNT] = { + MSG_MAP(TestMessage), + MSG_MAP(GetSmuVersion), + MSG_MAP(GetDriverIfVersion), + MSG_MAP(SetAllowedFeaturesMaskLow), + MSG_MAP(SetAllowedFeaturesMaskHigh), + MSG_MAP(EnableAllSmuFeatures), + MSG_MAP(DisableAllSmuFeatures), + MSG_MAP(EnableSmuFeaturesLow), + MSG_MAP(EnableSmuFeaturesHigh), + MSG_MAP(DisableSmuFeaturesLow), + MSG_MAP(DisableSmuFeaturesHigh), + MSG_MAP(GetEnabledSmuFeaturesLow), + MSG_MAP(GetEnabledSmuFeaturesHigh), + MSG_MAP(SetWorkloadMask), + MSG_MAP(SetPptLimit), + MSG_MAP(SetDriverDramAddrHigh), + MSG_MAP(SetDriverDramAddrLow), + MSG_MAP(SetToolsDramAddrHigh), + MSG_MAP(SetToolsDramAddrLow), + MSG_MAP(TransferTableSmu2Dram), + MSG_MAP(TransferTableDram2Smu), + MSG_MAP(UseDefaultPPTable), + MSG_MAP(UseBackupPPTable), + MSG_MAP(RunBtc), + MSG_MAP(RequestI2CBus), + MSG_MAP(ReleaseI2CBus), + MSG_MAP(SetFloorSocVoltage), + MSG_MAP(SoftReset), + MSG_MAP(StartBacoMonitor), + MSG_MAP(CancelBacoMonitor), + MSG_MAP(EnterBaco), + MSG_MAP(SetSoftMinByFreq), + MSG_MAP(SetSoftMaxByFreq), + MSG_MAP(SetHardMinByFreq), + MSG_MAP(SetHardMaxByFreq), + MSG_MAP(GetMinDpmFreq), + MSG_MAP(GetMaxDpmFreq), + MSG_MAP(GetDpmFreqByIndex), + MSG_MAP(GetDpmClockFreq), + MSG_MAP(GetSsVoltageByDpm), + MSG_MAP(SetMemoryChannelConfig), + MSG_MAP(SetGeminiMode), + MSG_MAP(SetGeminiApertureHigh), + MSG_MAP(SetGeminiApertureLow), + MSG_MAP(SetMinLinkDpmByIndex), + MSG_MAP(OverridePcieParameters), + MSG_MAP(OverDriveSetPercentage), + MSG_MAP(SetMinDeepSleepDcefclk), + MSG_MAP(ReenableAcDcInterrupt), + MSG_MAP(NotifyPowerSource), + MSG_MAP(SetUclkFastSwitch), + MSG_MAP(SetUclkDownHyst), + MSG_MAP(GetCurrentRpm), + MSG_MAP(SetVideoFps), + MSG_MAP(SetTjMax), + MSG_MAP(SetFanTemperatureTarget), + MSG_MAP(PrepareMp1ForUnload), + MSG_MAP(DramLogSetDramAddrHigh), + MSG_MAP(DramLogSetDramAddrLow), + MSG_MAP(DramLogSetDramSize), + MSG_MAP(SetFanMaxRpm), + MSG_MAP(SetFanMinPwm), + MSG_MAP(ConfigureGfxDidt), + MSG_MAP(NumOfDisplays), + MSG_MAP(RemoveMargins), + MSG_MAP(ReadSerialNumTop32), + MSG_MAP(ReadSerialNumBottom32), + MSG_MAP(SetSystemVirtualDramAddrHigh), + MSG_MAP(SetSystemVirtualDramAddrLow), + MSG_MAP(WaflTest), + MSG_MAP(SetFclkGfxClkRatio), + MSG_MAP(AllowGfxOff), + MSG_MAP(DisallowGfxOff), + MSG_MAP(GetPptLimit), + MSG_MAP(GetDcModeMaxDpmFreq), + MSG_MAP(GetDebugData), + MSG_MAP(SetXgmiMode), + MSG_MAP(RunAfllBtc), + MSG_MAP(ExitBaco), + MSG_MAP(PrepareMp1ForReset), + MSG_MAP(PrepareMp1ForShutdown), + MSG_MAP(SetMGpuFanBoostLimitRpm), + MSG_MAP(GetAVFSVoltageByDpm), + MSG_MAP(DFCstateControl), +}; + +static struct smu_11_0_cmn2aisc_mapping vega20_clk_map[SMU_CLK_COUNT] = { + CLK_MAP(GFXCLK, PPCLK_GFXCLK), + CLK_MAP(VCLK, PPCLK_VCLK), + CLK_MAP(DCLK, PPCLK_DCLK), + CLK_MAP(ECLK, PPCLK_ECLK), + CLK_MAP(SOCCLK, PPCLK_SOCCLK), + CLK_MAP(UCLK, PPCLK_UCLK), + CLK_MAP(DCEFCLK, PPCLK_DCEFCLK), + CLK_MAP(DISPCLK, PPCLK_DISPCLK), + CLK_MAP(PIXCLK, PPCLK_PIXCLK), + CLK_MAP(PHYCLK, PPCLK_PHYCLK), + CLK_MAP(FCLK, PPCLK_FCLK), +}; + +static struct smu_11_0_cmn2aisc_mapping vega20_feature_mask_map[SMU_FEATURE_COUNT] = { + FEA_MAP(DPM_PREFETCHER), + FEA_MAP(DPM_GFXCLK), + FEA_MAP(DPM_UCLK), + FEA_MAP(DPM_SOCCLK), + FEA_MAP(DPM_UVD), + FEA_MAP(DPM_VCE), + FEA_MAP(ULV), + FEA_MAP(DPM_MP0CLK), + FEA_MAP(DPM_LINK), + FEA_MAP(DPM_DCEFCLK), + FEA_MAP(DS_GFXCLK), + FEA_MAP(DS_SOCCLK), + FEA_MAP(DS_LCLK), + FEA_MAP(PPT), + FEA_MAP(TDC), + FEA_MAP(THERMAL), + FEA_MAP(GFX_PER_CU_CG), + FEA_MAP(RM), + FEA_MAP(DS_DCEFCLK), + FEA_MAP(ACDC), + FEA_MAP(VR0HOT), + FEA_MAP(VR1HOT), + FEA_MAP(FW_CTF), + FEA_MAP(LED_DISPLAY), + FEA_MAP(FAN_CONTROL), + FEA_MAP(GFX_EDC), + FEA_MAP(GFXOFF), + FEA_MAP(CG), + FEA_MAP(DPM_FCLK), + FEA_MAP(DS_FCLK), + FEA_MAP(DS_MP1CLK), + FEA_MAP(DS_MP0CLK), + FEA_MAP(XGMI), +}; + +static struct smu_11_0_cmn2aisc_mapping vega20_table_map[SMU_TABLE_COUNT] = { + TAB_MAP(PPTABLE), + TAB_MAP(WATERMARKS), + TAB_MAP(AVFS), + TAB_MAP(AVFS_PSM_DEBUG), + TAB_MAP(AVFS_FUSE_OVERRIDE), + TAB_MAP(PMSTATUSLOG), + TAB_MAP(SMU_METRICS), + TAB_MAP(DRIVER_SMU_CONFIG), + TAB_MAP(ACTIVITY_MONITOR_COEFF), + TAB_MAP(OVERDRIVE), +}; + +static struct smu_11_0_cmn2aisc_mapping vega20_pwr_src_map[SMU_POWER_SOURCE_COUNT] = { + PWR_MAP(AC), + PWR_MAP(DC), +}; + +static struct smu_11_0_cmn2aisc_mapping vega20_workload_map[PP_SMC_POWER_PROFILE_COUNT] = { + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT, WORKLOAD_DEFAULT_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_FULLSCREEN3D, WORKLOAD_PPLIB_FULL_SCREEN_3D_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_POWERSAVING, WORKLOAD_PPLIB_POWER_SAVING_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_VIDEO, WORKLOAD_PPLIB_VIDEO_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_VR, WORKLOAD_PPLIB_VR_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_COMPUTE, WORKLOAD_PPLIB_COMPUTE_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_CUSTOM, WORKLOAD_PPLIB_CUSTOM_BIT), +}; + +static int vega20_get_smu_table_index(struct smu_context *smc, uint32_t index) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (index >= SMU_TABLE_COUNT) + return -EINVAL; + + mapping = vega20_table_map[index]; + if (!(mapping.valid_mapping)) { + return -EINVAL; + } + + return mapping.map_to; +} + +static int vega20_get_pwr_src_index(struct smu_context *smc, uint32_t index) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (index >= SMU_POWER_SOURCE_COUNT) + return -EINVAL; + + mapping = vega20_pwr_src_map[index]; + if (!(mapping.valid_mapping)) { + return -EINVAL; + } + + return mapping.map_to; +} + +static int vega20_get_smu_feature_index(struct smu_context *smc, uint32_t index) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (index >= SMU_FEATURE_COUNT) + return -EINVAL; + + mapping = vega20_feature_mask_map[index]; + if (!(mapping.valid_mapping)) { + return -EINVAL; + } + + return mapping.map_to; +} + +static int vega20_get_smu_clk_index(struct smu_context *smc, uint32_t index) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (index >= SMU_CLK_COUNT) + return -EINVAL; + + mapping = vega20_clk_map[index]; + if (!(mapping.valid_mapping)) { + return -EINVAL; + } + + return mapping.map_to; +} + +static int vega20_get_smu_msg_index(struct smu_context *smc, uint32_t index) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (index >= SMU_MSG_MAX_COUNT) + return -EINVAL; + + mapping = vega20_message_map[index]; + if (!(mapping.valid_mapping)) { + return -EINVAL; + } + + return mapping.map_to; +} + +static int vega20_get_workload_type(struct smu_context *smu, enum PP_SMC_POWER_PROFILE profile) +{ + struct smu_11_0_cmn2aisc_mapping mapping; + + if (profile > PP_SMC_POWER_PROFILE_CUSTOM) + return -EINVAL; + + mapping = vega20_workload_map[profile]; + if (!(mapping.valid_mapping)) { + return -EINVAL; + } + + return mapping.map_to; +} + +static int vega20_tables_init(struct smu_context *smu, struct smu_table *tables) +{ + struct smu_table_context *smu_table = &smu->smu_table; + + SMU_TABLE_INIT(tables, SMU_TABLE_PPTABLE, sizeof(PPTable_t), + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + SMU_TABLE_INIT(tables, SMU_TABLE_WATERMARKS, sizeof(Watermarks_t), + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + SMU_TABLE_INIT(tables, SMU_TABLE_SMU_METRICS, sizeof(SmuMetrics_t), + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + SMU_TABLE_INIT(tables, SMU_TABLE_OVERDRIVE, sizeof(OverDriveTable_t), + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM); + SMU_TABLE_INIT(tables, SMU_TABLE_PMSTATUSLOG, SMU11_TOOL_SIZE, + 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; + smu_table->metrics_time = 0; + + return 0; +} + +static int vega20_allocate_dpm_context(struct smu_context *smu) +{ + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + + if (smu_dpm->dpm_context) + return -EINVAL; + + smu_dpm->dpm_context = kzalloc(sizeof(struct vega20_dpm_table), + GFP_KERNEL); + if (!smu_dpm->dpm_context) + return -ENOMEM; + + if (smu_dpm->golden_dpm_context) + return -EINVAL; + + smu_dpm->golden_dpm_context = kzalloc(sizeof(struct vega20_dpm_table), + GFP_KERNEL); + if (!smu_dpm->golden_dpm_context) + return -ENOMEM; + + smu_dpm->dpm_context_size = sizeof(struct vega20_dpm_table); + + smu_dpm->dpm_current_power_state = kzalloc(sizeof(struct smu_power_state), + GFP_KERNEL); + if (!smu_dpm->dpm_current_power_state) + return -ENOMEM; + + smu_dpm->dpm_request_power_state = kzalloc(sizeof(struct smu_power_state), + GFP_KERNEL); + if (!smu_dpm->dpm_request_power_state) + return -ENOMEM; + + return 0; +} + +static int vega20_setup_od8_information(struct smu_context *smu) +{ + ATOM_Vega20_POWERPLAYTABLE *powerplay_table = NULL; + struct smu_table_context *table_context = &smu->smu_table; + struct vega20_od8_settings *od8_settings = (struct vega20_od8_settings *)smu->od_settings; + + uint32_t od_feature_count, od_feature_array_size, + od_setting_count, od_setting_array_size; + + if (!table_context->power_play_table) + return -EINVAL; + + powerplay_table = table_context->power_play_table; + + if (powerplay_table->OverDrive8Table.ucODTableRevision == 1) { + /* Setup correct ODFeatureCount, and store ODFeatureArray from + * powerplay table to od_feature_capabilities */ + od_feature_count = + (le32_to_cpu(powerplay_table->OverDrive8Table.ODFeatureCount) > + ATOM_VEGA20_ODFEATURE_COUNT) ? + ATOM_VEGA20_ODFEATURE_COUNT : + le32_to_cpu(powerplay_table->OverDrive8Table.ODFeatureCount); + + od_feature_array_size = sizeof(uint8_t) * od_feature_count; + + if (od8_settings->od_feature_capabilities) + return -EINVAL; + + od8_settings->od_feature_capabilities = kmemdup(&powerplay_table->OverDrive8Table.ODFeatureCapabilities, + od_feature_array_size, + GFP_KERNEL); + if (!od8_settings->od_feature_capabilities) + return -ENOMEM; + + /* Setup correct ODSettingCount, and store ODSettingArray from + * powerplay table to od_settings_max and od_setting_min */ + od_setting_count = + (le32_to_cpu(powerplay_table->OverDrive8Table.ODSettingCount) > + ATOM_VEGA20_ODSETTING_COUNT) ? + ATOM_VEGA20_ODSETTING_COUNT : + le32_to_cpu(powerplay_table->OverDrive8Table.ODSettingCount); + + od_setting_array_size = sizeof(uint32_t) * od_setting_count; + + if (od8_settings->od_settings_max) + return -EINVAL; + + od8_settings->od_settings_max = kmemdup(&powerplay_table->OverDrive8Table.ODSettingsMax, + od_setting_array_size, + GFP_KERNEL); + + if (!od8_settings->od_settings_max) { + kfree(od8_settings->od_feature_capabilities); + od8_settings->od_feature_capabilities = NULL; + return -ENOMEM; + } + + if (od8_settings->od_settings_min) + return -EINVAL; + + od8_settings->od_settings_min = kmemdup(&powerplay_table->OverDrive8Table.ODSettingsMin, + od_setting_array_size, + GFP_KERNEL); + + if (!od8_settings->od_settings_min) { + kfree(od8_settings->od_feature_capabilities); + od8_settings->od_feature_capabilities = NULL; + kfree(od8_settings->od_settings_max); + od8_settings->od_settings_max = NULL; + return -ENOMEM; + } + } + + return 0; +} + +static int vega20_store_powerplay_table(struct smu_context *smu) +{ + ATOM_Vega20_POWERPLAYTABLE *powerplay_table = NULL; + struct smu_table_context *table_context = &smu->smu_table; + + if (!table_context->power_play_table) + return -EINVAL; + + powerplay_table = table_context->power_play_table; + + memcpy(table_context->driver_pptable, &powerplay_table->smcPPTable, + sizeof(PPTable_t)); + + table_context->thermal_controller_type = powerplay_table->ucThermalControllerType; + + return 0; +} + +static int vega20_append_powerplay_table(struct smu_context *smu) +{ + struct smu_table_context *table_context = &smu->smu_table; + PPTable_t *smc_pptable = table_context->driver_pptable; + struct atom_smc_dpm_info_v4_4 *smc_dpm_table; + int index, i, ret; + + index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1, + smc_dpm_info); + + ret = smu_get_atom_data_table(smu, index, NULL, NULL, NULL, + (uint8_t **)&smc_dpm_table); + if (ret) + return ret; + + smc_pptable->MaxVoltageStepGfx = smc_dpm_table->maxvoltagestepgfx; + smc_pptable->MaxVoltageStepSoc = smc_dpm_table->maxvoltagestepsoc; + + smc_pptable->VddGfxVrMapping = smc_dpm_table->vddgfxvrmapping; + smc_pptable->VddSocVrMapping = smc_dpm_table->vddsocvrmapping; + smc_pptable->VddMem0VrMapping = smc_dpm_table->vddmem0vrmapping; + smc_pptable->VddMem1VrMapping = smc_dpm_table->vddmem1vrmapping; + + smc_pptable->GfxUlvPhaseSheddingMask = smc_dpm_table->gfxulvphasesheddingmask; + smc_pptable->SocUlvPhaseSheddingMask = smc_dpm_table->soculvphasesheddingmask; + smc_pptable->ExternalSensorPresent = smc_dpm_table->externalsensorpresent; + + smc_pptable->GfxMaxCurrent = smc_dpm_table->gfxmaxcurrent; + smc_pptable->GfxOffset = smc_dpm_table->gfxoffset; + smc_pptable->Padding_TelemetryGfx = smc_dpm_table->padding_telemetrygfx; + + smc_pptable->SocMaxCurrent = smc_dpm_table->socmaxcurrent; + smc_pptable->SocOffset = smc_dpm_table->socoffset; + smc_pptable->Padding_TelemetrySoc = smc_dpm_table->padding_telemetrysoc; + + smc_pptable->Mem0MaxCurrent = smc_dpm_table->mem0maxcurrent; + smc_pptable->Mem0Offset = smc_dpm_table->mem0offset; + smc_pptable->Padding_TelemetryMem0 = smc_dpm_table->padding_telemetrymem0; + + smc_pptable->Mem1MaxCurrent = smc_dpm_table->mem1maxcurrent; + smc_pptable->Mem1Offset = smc_dpm_table->mem1offset; + smc_pptable->Padding_TelemetryMem1 = smc_dpm_table->padding_telemetrymem1; + + smc_pptable->AcDcGpio = smc_dpm_table->acdcgpio; + smc_pptable->AcDcPolarity = smc_dpm_table->acdcpolarity; + smc_pptable->VR0HotGpio = smc_dpm_table->vr0hotgpio; + smc_pptable->VR0HotPolarity = smc_dpm_table->vr0hotpolarity; + + smc_pptable->VR1HotGpio = smc_dpm_table->vr1hotgpio; + smc_pptable->VR1HotPolarity = smc_dpm_table->vr1hotpolarity; + smc_pptable->Padding1 = smc_dpm_table->padding1; + smc_pptable->Padding2 = smc_dpm_table->padding2; + + smc_pptable->LedPin0 = smc_dpm_table->ledpin0; + smc_pptable->LedPin1 = smc_dpm_table->ledpin1; + smc_pptable->LedPin2 = smc_dpm_table->ledpin2; + + smc_pptable->PllGfxclkSpreadEnabled = smc_dpm_table->pllgfxclkspreadenabled; + smc_pptable->PllGfxclkSpreadPercent = smc_dpm_table->pllgfxclkspreadpercent; + smc_pptable->PllGfxclkSpreadFreq = smc_dpm_table->pllgfxclkspreadfreq; + + smc_pptable->UclkSpreadEnabled = 0; + smc_pptable->UclkSpreadPercent = smc_dpm_table->uclkspreadpercent; + smc_pptable->UclkSpreadFreq = smc_dpm_table->uclkspreadfreq; + + smc_pptable->FclkSpreadEnabled = smc_dpm_table->fclkspreadenabled; + smc_pptable->FclkSpreadPercent = smc_dpm_table->fclkspreadpercent; + smc_pptable->FclkSpreadFreq = smc_dpm_table->fclkspreadfreq; + + smc_pptable->FllGfxclkSpreadEnabled = smc_dpm_table->fllgfxclkspreadenabled; + smc_pptable->FllGfxclkSpreadPercent = smc_dpm_table->fllgfxclkspreadpercent; + smc_pptable->FllGfxclkSpreadFreq = smc_dpm_table->fllgfxclkspreadfreq; + + for (i = 0; i < I2C_CONTROLLER_NAME_COUNT; i++) { + smc_pptable->I2cControllers[i].Enabled = + smc_dpm_table->i2ccontrollers[i].enabled; + smc_pptable->I2cControllers[i].SlaveAddress = + smc_dpm_table->i2ccontrollers[i].slaveaddress; + smc_pptable->I2cControllers[i].ControllerPort = + smc_dpm_table->i2ccontrollers[i].controllerport; + smc_pptable->I2cControllers[i].ThermalThrottler = + smc_dpm_table->i2ccontrollers[i].thermalthrottler; + smc_pptable->I2cControllers[i].I2cProtocol = + smc_dpm_table->i2ccontrollers[i].i2cprotocol; + smc_pptable->I2cControllers[i].I2cSpeed = + smc_dpm_table->i2ccontrollers[i].i2cspeed; + } + + return 0; +} + +static int vega20_check_powerplay_table(struct smu_context *smu) +{ + ATOM_Vega20_POWERPLAYTABLE *powerplay_table = NULL; + struct smu_table_context *table_context = &smu->smu_table; + + powerplay_table = table_context->power_play_table; + + if (powerplay_table->sHeader.format_revision < ATOM_VEGA20_TABLE_REVISION_VEGA20) { + pr_err("Unsupported PPTable format!"); + return -EINVAL; + } + + if (!powerplay_table->sHeader.structuresize) { + pr_err("Invalid PowerPlay Table!"); + return -EINVAL; + } + + return 0; +} + +static int vega20_run_btc_afll(struct smu_context *smu) +{ + return smu_send_smc_msg(smu, SMU_MSG_RunAfllBtc); +} + +#define FEATURE_MASK(feature) (1ULL << feature) +static int +vega20_get_allowed_feature_mask(struct smu_context *smu, + uint32_t *feature_mask, uint32_t num) +{ + if (num > 2) + return -EINVAL; + + memset(feature_mask, 0, sizeof(uint32_t) * num); + + *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_DPM_PREFETCHER_BIT) + | FEATURE_MASK(FEATURE_DPM_GFXCLK_BIT) + | FEATURE_MASK(FEATURE_DPM_UCLK_BIT) + | FEATURE_MASK(FEATURE_DPM_SOCCLK_BIT) + | FEATURE_MASK(FEATURE_DPM_UVD_BIT) + | FEATURE_MASK(FEATURE_DPM_VCE_BIT) + | FEATURE_MASK(FEATURE_ULV_BIT) + | FEATURE_MASK(FEATURE_DPM_MP0CLK_BIT) + | FEATURE_MASK(FEATURE_DPM_LINK_BIT) + | FEATURE_MASK(FEATURE_DPM_DCEFCLK_BIT) + | FEATURE_MASK(FEATURE_PPT_BIT) + | FEATURE_MASK(FEATURE_TDC_BIT) + | FEATURE_MASK(FEATURE_THERMAL_BIT) + | FEATURE_MASK(FEATURE_GFX_PER_CU_CG_BIT) + | FEATURE_MASK(FEATURE_RM_BIT) + | FEATURE_MASK(FEATURE_ACDC_BIT) + | FEATURE_MASK(FEATURE_VR0HOT_BIT) + | FEATURE_MASK(FEATURE_VR1HOT_BIT) + | FEATURE_MASK(FEATURE_FW_CTF_BIT) + | FEATURE_MASK(FEATURE_LED_DISPLAY_BIT) + | FEATURE_MASK(FEATURE_FAN_CONTROL_BIT) + | FEATURE_MASK(FEATURE_GFX_EDC_BIT) + | FEATURE_MASK(FEATURE_GFXOFF_BIT) + | FEATURE_MASK(FEATURE_CG_BIT) + | FEATURE_MASK(FEATURE_DPM_FCLK_BIT) + | FEATURE_MASK(FEATURE_XGMI_BIT); + return 0; +} + +static enum +amd_pm_state_type vega20_get_current_power_state(struct smu_context *smu) +{ + enum amd_pm_state_type pm_type; + struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm); + + if (!smu_dpm_ctx->dpm_context || + !smu_dpm_ctx->dpm_current_power_state) + return -EINVAL; + + switch (smu_dpm_ctx->dpm_current_power_state->classification.ui_label) { + case SMU_STATE_UI_LABEL_BATTERY: + pm_type = POWER_STATE_TYPE_BATTERY; + break; + case SMU_STATE_UI_LABEL_BALLANCED: + pm_type = POWER_STATE_TYPE_BALANCED; + break; + case SMU_STATE_UI_LABEL_PERFORMANCE: + pm_type = POWER_STATE_TYPE_PERFORMANCE; + break; + default: + if (smu_dpm_ctx->dpm_current_power_state->classification.flags & SMU_STATE_CLASSIFICATION_FLAG_BOOT) + pm_type = POWER_STATE_TYPE_INTERNAL_BOOT; + else + pm_type = POWER_STATE_TYPE_DEFAULT; + break; + } + + return pm_type; +} + +static int +vega20_set_single_dpm_table(struct smu_context *smu, + struct vega20_single_dpm_table *single_dpm_table, + PPCLK_e clk_id) +{ + int ret = 0; + uint32_t i, num_of_levels = 0, clk; + + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_GetDpmFreqByIndex, + (clk_id << 16 | 0xFF)); + if (ret) { + pr_err("[GetNumOfDpmLevel] failed to get dpm levels!"); + return ret; + } + + smu_read_smc_arg(smu, &num_of_levels); + if (!num_of_levels) { + pr_err("[GetNumOfDpmLevel] number of clk levels is invalid!"); + return -EINVAL; + } + + single_dpm_table->count = num_of_levels; + + for (i = 0; i < num_of_levels; i++) { + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_GetDpmFreqByIndex, + (clk_id << 16 | i)); + if (ret) { + pr_err("[GetDpmFreqByIndex] failed to get dpm freq by index!"); + return ret; + } + smu_read_smc_arg(smu, &clk); + if (!clk) { + pr_err("[GetDpmFreqByIndex] clk value is invalid!"); + return -EINVAL; + } + single_dpm_table->dpm_levels[i].value = clk; + single_dpm_table->dpm_levels[i].enabled = true; + } + return 0; +} + +static void vega20_init_single_dpm_state(struct vega20_dpm_state *dpm_state) +{ + dpm_state->soft_min_level = 0x0; + dpm_state->soft_max_level = 0xffff; + dpm_state->hard_min_level = 0x0; + dpm_state->hard_max_level = 0xffff; +} + +static int vega20_set_default_dpm_table(struct smu_context *smu) +{ + int ret; + + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + struct vega20_dpm_table *dpm_table = NULL; + struct vega20_single_dpm_table *single_dpm_table; + + dpm_table = smu_dpm->dpm_context; + + /* socclk */ + single_dpm_table = &(dpm_table->soc_table); + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_SOCCLK_BIT)) { + ret = vega20_set_single_dpm_table(smu, single_dpm_table, + PPCLK_SOCCLK); + if (ret) { + pr_err("[SetupDefaultDpmTable] failed to get socclk dpm levels!"); + return ret; + } + } else { + single_dpm_table->count = 1; + single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.socclk / 100; + } + vega20_init_single_dpm_state(&(single_dpm_table->dpm_state)); + + /* gfxclk */ + single_dpm_table = &(dpm_table->gfx_table); + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_GFXCLK_BIT)) { + ret = vega20_set_single_dpm_table(smu, single_dpm_table, + PPCLK_GFXCLK); + if (ret) { + pr_err("[SetupDefaultDpmTable] failed to get gfxclk dpm levels!"); + return ret; + } + } else { + single_dpm_table->count = 1; + single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.gfxclk / 100; + } + vega20_init_single_dpm_state(&(single_dpm_table->dpm_state)); + + /* memclk */ + single_dpm_table = &(dpm_table->mem_table); + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT)) { + ret = vega20_set_single_dpm_table(smu, single_dpm_table, + PPCLK_UCLK); + if (ret) { + pr_err("[SetupDefaultDpmTable] failed to get memclk dpm levels!"); + return ret; + } + } else { + single_dpm_table->count = 1; + single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.uclk / 100; + } + vega20_init_single_dpm_state(&(single_dpm_table->dpm_state)); + + /* eclk */ + single_dpm_table = &(dpm_table->eclk_table); + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_VCE_BIT)) { + ret = vega20_set_single_dpm_table(smu, single_dpm_table, PPCLK_ECLK); + if (ret) { + pr_err("[SetupDefaultDpmTable] failed to get eclk dpm levels!"); + return ret; + } + } else { + single_dpm_table->count = 1; + single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.eclk / 100; + } + vega20_init_single_dpm_state(&(single_dpm_table->dpm_state)); + + /* vclk */ + single_dpm_table = &(dpm_table->vclk_table); + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UVD_BIT)) { + ret = vega20_set_single_dpm_table(smu, single_dpm_table, PPCLK_VCLK); + if (ret) { + pr_err("[SetupDefaultDpmTable] failed to get vclk dpm levels!"); + return ret; + } + } else { + single_dpm_table->count = 1; + single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.vclk / 100; + } + vega20_init_single_dpm_state(&(single_dpm_table->dpm_state)); + + /* dclk */ + single_dpm_table = &(dpm_table->dclk_table); + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UVD_BIT)) { + ret = vega20_set_single_dpm_table(smu, single_dpm_table, PPCLK_DCLK); + if (ret) { + pr_err("[SetupDefaultDpmTable] failed to get dclk dpm levels!"); + return ret; + } + } else { + single_dpm_table->count = 1; + single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.dclk / 100; + } + vega20_init_single_dpm_state(&(single_dpm_table->dpm_state)); + + /* dcefclk */ + single_dpm_table = &(dpm_table->dcef_table); + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT)) { + ret = vega20_set_single_dpm_table(smu, single_dpm_table, + PPCLK_DCEFCLK); + if (ret) { + pr_err("[SetupDefaultDpmTable] failed to get dcefclk dpm levels!"); + return ret; + } + } else { + single_dpm_table->count = 1; + single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.dcefclk / 100; + } + vega20_init_single_dpm_state(&(single_dpm_table->dpm_state)); + + /* pixclk */ + single_dpm_table = &(dpm_table->pixel_table); + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT)) { + ret = vega20_set_single_dpm_table(smu, single_dpm_table, + PPCLK_PIXCLK); + if (ret) { + pr_err("[SetupDefaultDpmTable] failed to get pixclk dpm levels!"); + return ret; + } + } else { + single_dpm_table->count = 0; + } + vega20_init_single_dpm_state(&(single_dpm_table->dpm_state)); + + /* dispclk */ + single_dpm_table = &(dpm_table->display_table); + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT)) { + ret = vega20_set_single_dpm_table(smu, single_dpm_table, + PPCLK_DISPCLK); + if (ret) { + pr_err("[SetupDefaultDpmTable] failed to get dispclk dpm levels!"); + return ret; + } + } else { + single_dpm_table->count = 0; + } + vega20_init_single_dpm_state(&(single_dpm_table->dpm_state)); + + /* phyclk */ + single_dpm_table = &(dpm_table->phy_table); + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT)) { + ret = vega20_set_single_dpm_table(smu, single_dpm_table, + PPCLK_PHYCLK); + if (ret) { + pr_err("[SetupDefaultDpmTable] failed to get phyclk dpm levels!"); + return ret; + } + } else { + single_dpm_table->count = 0; + } + vega20_init_single_dpm_state(&(single_dpm_table->dpm_state)); + + /* fclk */ + single_dpm_table = &(dpm_table->fclk_table); + + if (smu_feature_is_enabled(smu,FEATURE_DPM_FCLK_BIT)) { + ret = vega20_set_single_dpm_table(smu, single_dpm_table, + PPCLK_FCLK); + if (ret) { + pr_err("[SetupDefaultDpmTable] failed to get fclk dpm levels!"); + return ret; + } + } else { + single_dpm_table->count = 0; + } + vega20_init_single_dpm_state(&(single_dpm_table->dpm_state)); + + memcpy(smu_dpm->golden_dpm_context, dpm_table, + sizeof(struct vega20_dpm_table)); + + return 0; +} + +static int vega20_populate_umd_state_clk(struct smu_context *smu) +{ + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + struct vega20_dpm_table *dpm_table = NULL; + struct vega20_single_dpm_table *gfx_table = NULL; + struct vega20_single_dpm_table *mem_table = NULL; + + dpm_table = smu_dpm->dpm_context; + gfx_table = &(dpm_table->gfx_table); + mem_table = &(dpm_table->mem_table); + + smu->pstate_sclk = gfx_table->dpm_levels[0].value; + smu->pstate_mclk = mem_table->dpm_levels[0].value; + + if (gfx_table->count > VEGA20_UMD_PSTATE_GFXCLK_LEVEL && + mem_table->count > VEGA20_UMD_PSTATE_MCLK_LEVEL) { + smu->pstate_sclk = gfx_table->dpm_levels[VEGA20_UMD_PSTATE_GFXCLK_LEVEL].value; + smu->pstate_mclk = mem_table->dpm_levels[VEGA20_UMD_PSTATE_MCLK_LEVEL].value; + } + + smu->pstate_sclk = smu->pstate_sclk * 100; + smu->pstate_mclk = smu->pstate_mclk * 100; + + return 0; +} + +static int vega20_get_clk_table(struct smu_context *smu, + struct pp_clock_levels_with_latency *clocks, + struct vega20_single_dpm_table *dpm_table) +{ + int i, count; + + count = (dpm_table->count > MAX_NUM_CLOCKS) ? MAX_NUM_CLOCKS : dpm_table->count; + clocks->num_levels = count; + + for (i = 0; i < count; i++) { + clocks->data[i].clocks_in_khz = + dpm_table->dpm_levels[i].value * 1000; + clocks->data[i].latency_in_us = 0; + } + + return 0; +} + +static int vega20_print_clk_levels(struct smu_context *smu, + enum smu_clk_type type, char *buf) +{ + int i, now, size = 0; + int ret = 0; + uint32_t gen_speed, lane_width; + struct amdgpu_device *adev = smu->adev; + struct pp_clock_levels_with_latency clocks; + struct vega20_single_dpm_table *single_dpm_table; + struct smu_table_context *table_context = &smu->smu_table; + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + struct vega20_dpm_table *dpm_table = NULL; + struct vega20_od8_settings *od8_settings = + (struct vega20_od8_settings *)smu->od_settings; + OverDriveTable_t *od_table = + (OverDriveTable_t *)(table_context->overdrive_table); + PPTable_t *pptable = (PPTable_t *)table_context->driver_pptable; + + dpm_table = smu_dpm->dpm_context; + + switch (type) { + case SMU_SCLK: + ret = smu_get_current_clk_freq(smu, SMU_GFXCLK, &now); + if (ret) { + pr_err("Attempt to get current gfx clk Failed!"); + return ret; + } + + single_dpm_table = &(dpm_table->gfx_table); + ret = vega20_get_clk_table(smu, &clocks, single_dpm_table); + if (ret) { + pr_err("Attempt to get gfx clk levels Failed!"); + return ret; + } + + for (i = 0; i < clocks.num_levels; i++) + size += sprintf(buf + size, "%d: %uMhz %s\n", i, + clocks.data[i].clocks_in_khz / 1000, + (clocks.data[i].clocks_in_khz == now * 10) + ? "*" : ""); + break; + + case SMU_MCLK: + ret = smu_get_current_clk_freq(smu, SMU_UCLK, &now); + if (ret) { + pr_err("Attempt to get current mclk Failed!"); + return ret; + } + + single_dpm_table = &(dpm_table->mem_table); + ret = vega20_get_clk_table(smu, &clocks, single_dpm_table); + if (ret) { + pr_err("Attempt to get memory clk levels Failed!"); + return ret; + } + + for (i = 0; i < clocks.num_levels; i++) + size += sprintf(buf + size, "%d: %uMhz %s\n", + i, clocks.data[i].clocks_in_khz / 1000, + (clocks.data[i].clocks_in_khz == now * 10) + ? "*" : ""); + break; + + case SMU_SOCCLK: + ret = smu_get_current_clk_freq(smu, SMU_SOCCLK, &now); + if (ret) { + pr_err("Attempt to get current socclk Failed!"); + return ret; + } + + single_dpm_table = &(dpm_table->soc_table); + ret = vega20_get_clk_table(smu, &clocks, single_dpm_table); + if (ret) { + pr_err("Attempt to get socclk levels Failed!"); + return ret; + } + + for (i = 0; i < clocks.num_levels; i++) + size += sprintf(buf + size, "%d: %uMhz %s\n", + i, clocks.data[i].clocks_in_khz / 1000, + (clocks.data[i].clocks_in_khz == now * 10) + ? "*" : ""); + break; + + case SMU_FCLK: + ret = smu_get_current_clk_freq(smu, SMU_FCLK, &now); + if (ret) { + pr_err("Attempt to get current fclk Failed!"); + return ret; + } + + single_dpm_table = &(dpm_table->fclk_table); + for (i = 0; i < single_dpm_table->count; i++) + size += sprintf(buf + size, "%d: %uMhz %s\n", + i, single_dpm_table->dpm_levels[i].value, + (single_dpm_table->dpm_levels[i].value == now / 100) + ? "*" : ""); + break; + + case SMU_DCEFCLK: + ret = smu_get_current_clk_freq(smu, SMU_DCEFCLK, &now); + if (ret) { + pr_err("Attempt to get current dcefclk Failed!"); + return ret; + } + + single_dpm_table = &(dpm_table->dcef_table); + ret = vega20_get_clk_table(smu, &clocks, single_dpm_table); + if (ret) { + pr_err("Attempt to get dcefclk levels Failed!"); + return ret; + } + + for (i = 0; i < clocks.num_levels; i++) + size += sprintf(buf + size, "%d: %uMhz %s\n", + i, clocks.data[i].clocks_in_khz / 1000, + (clocks.data[i].clocks_in_khz == now * 10) ? "*" : ""); + break; + + case SMU_PCIE: + gen_speed = (RREG32_PCIE(smnPCIE_LC_SPEED_CNTL) & + PSWUSP0_PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE_MASK) + >> PSWUSP0_PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE__SHIFT; + lane_width = (RREG32_PCIE(smnPCIE_LC_LINK_WIDTH_CNTL) & + PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_RD_MASK) + >> PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_RD__SHIFT; + for (i = 0; i < NUM_LINK_LEVELS; i++) + size += sprintf(buf + size, "%d: %s %s %dMhz %s\n", i, + (pptable->PcieGenSpeed[i] == 0) ? "2.5GT/s," : + (pptable->PcieGenSpeed[i] == 1) ? "5.0GT/s," : + (pptable->PcieGenSpeed[i] == 2) ? "8.0GT/s," : + (pptable->PcieGenSpeed[i] == 3) ? "16.0GT/s," : "", + (pptable->PcieLaneCount[i] == 1) ? "x1" : + (pptable->PcieLaneCount[i] == 2) ? "x2" : + (pptable->PcieLaneCount[i] == 3) ? "x4" : + (pptable->PcieLaneCount[i] == 4) ? "x8" : + (pptable->PcieLaneCount[i] == 5) ? "x12" : + (pptable->PcieLaneCount[i] == 6) ? "x16" : "", + pptable->LclkFreq[i], + (gen_speed == pptable->PcieGenSpeed[i]) && + (lane_width == pptable->PcieLaneCount[i]) ? + "*" : ""); + break; + + case SMU_OD_SCLK: + if (od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMIN].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMAX].feature_id) { + size = sprintf(buf, "%s:\n", "OD_SCLK"); + size += sprintf(buf + size, "0: %10uMhz\n", + od_table->GfxclkFmin); + size += sprintf(buf + size, "1: %10uMhz\n", + od_table->GfxclkFmax); + } + + break; + + case SMU_OD_MCLK: + if (od8_settings->od8_settings_array[OD8_SETTING_UCLK_FMAX].feature_id) { + size = sprintf(buf, "%s:\n", "OD_MCLK"); + size += sprintf(buf + size, "1: %10uMhz\n", + od_table->UclkFmax); + } + + break; + + case SMU_OD_VDDC_CURVE: + if (od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ1].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ2].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ3].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE1].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE2].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE3].feature_id) { + size = sprintf(buf, "%s:\n", "OD_VDDC_CURVE"); + size += sprintf(buf + size, "0: %10uMhz %10dmV\n", + od_table->GfxclkFreq1, + od_table->GfxclkVolt1 / VOLTAGE_SCALE); + size += sprintf(buf + size, "1: %10uMhz %10dmV\n", + od_table->GfxclkFreq2, + od_table->GfxclkVolt2 / VOLTAGE_SCALE); + size += sprintf(buf + size, "2: %10uMhz %10dmV\n", + od_table->GfxclkFreq3, + od_table->GfxclkVolt3 / VOLTAGE_SCALE); + } + + break; + + case SMU_OD_RANGE: + size = sprintf(buf, "%s:\n", "OD_RANGE"); + + if (od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMIN].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMAX].feature_id) { + size += sprintf(buf + size, "SCLK: %7uMhz %10uMhz\n", + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMIN].min_value, + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMAX].max_value); + } + + if (od8_settings->od8_settings_array[OD8_SETTING_UCLK_FMAX].feature_id) { + single_dpm_table = &(dpm_table->mem_table); + ret = vega20_get_clk_table(smu, &clocks, single_dpm_table); + if (ret) { + pr_err("Attempt to get memory clk levels Failed!"); + return ret; + } + + size += sprintf(buf + size, "MCLK: %7uMhz %10uMhz\n", + clocks.data[0].clocks_in_khz / 1000, + od8_settings->od8_settings_array[OD8_SETTING_UCLK_FMAX].max_value); + } + + if (od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ1].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ2].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ3].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE1].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE2].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE3].feature_id) { + size += sprintf(buf + size, "VDDC_CURVE_SCLK[0]: %7uMhz %10uMhz\n", + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ1].min_value, + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ1].max_value); + size += sprintf(buf + size, "VDDC_CURVE_VOLT[0]: %7dmV %11dmV\n", + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE1].min_value, + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE1].max_value); + size += sprintf(buf + size, "VDDC_CURVE_SCLK[1]: %7uMhz %10uMhz\n", + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ2].min_value, + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ2].max_value); + size += sprintf(buf + size, "VDDC_CURVE_VOLT[1]: %7dmV %11dmV\n", + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE2].min_value, + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE2].max_value); + size += sprintf(buf + size, "VDDC_CURVE_SCLK[2]: %7uMhz %10uMhz\n", + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ3].min_value, + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ3].max_value); + size += sprintf(buf + size, "VDDC_CURVE_VOLT[2]: %7dmV %11dmV\n", + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE3].min_value, + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE3].max_value); + } + + break; + + default: + break; + } + return size; +} + +static int vega20_upload_dpm_level(struct smu_context *smu, bool max, + uint32_t feature_mask) +{ + struct vega20_dpm_table *dpm_table; + struct vega20_single_dpm_table *single_dpm_table; + uint32_t freq; + int ret = 0; + + dpm_table = smu->smu_dpm.dpm_context; + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_GFXCLK_BIT) && + (feature_mask & FEATURE_DPM_GFXCLK_MASK)) { + single_dpm_table = &(dpm_table->gfx_table); + freq = max ? single_dpm_table->dpm_state.soft_max_level : + single_dpm_table->dpm_state.soft_min_level; + ret = smu_send_smc_msg_with_param(smu, + (max ? SMU_MSG_SetSoftMaxByFreq : SMU_MSG_SetSoftMinByFreq), + (PPCLK_GFXCLK << 16) | (freq & 0xffff)); + if (ret) { + pr_err("Failed to set soft %s gfxclk !\n", + max ? "max" : "min"); + return ret; + } + } + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT) && + (feature_mask & FEATURE_DPM_UCLK_MASK)) { + single_dpm_table = &(dpm_table->mem_table); + freq = max ? single_dpm_table->dpm_state.soft_max_level : + single_dpm_table->dpm_state.soft_min_level; + ret = smu_send_smc_msg_with_param(smu, + (max ? SMU_MSG_SetSoftMaxByFreq : SMU_MSG_SetSoftMinByFreq), + (PPCLK_UCLK << 16) | (freq & 0xffff)); + if (ret) { + pr_err("Failed to set soft %s memclk !\n", + max ? "max" : "min"); + return ret; + } + } + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_SOCCLK_BIT) && + (feature_mask & FEATURE_DPM_SOCCLK_MASK)) { + single_dpm_table = &(dpm_table->soc_table); + freq = max ? single_dpm_table->dpm_state.soft_max_level : + single_dpm_table->dpm_state.soft_min_level; + ret = smu_send_smc_msg_with_param(smu, + (max ? SMU_MSG_SetSoftMaxByFreq : SMU_MSG_SetSoftMinByFreq), + (PPCLK_SOCCLK << 16) | (freq & 0xffff)); + if (ret) { + pr_err("Failed to set soft %s socclk !\n", + max ? "max" : "min"); + return ret; + } + } + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_FCLK_BIT) && + (feature_mask & FEATURE_DPM_FCLK_MASK)) { + single_dpm_table = &(dpm_table->fclk_table); + freq = max ? single_dpm_table->dpm_state.soft_max_level : + single_dpm_table->dpm_state.soft_min_level; + ret = smu_send_smc_msg_with_param(smu, + (max ? SMU_MSG_SetSoftMaxByFreq : SMU_MSG_SetSoftMinByFreq), + (PPCLK_FCLK << 16) | (freq & 0xffff)); + if (ret) { + pr_err("Failed to set soft %s fclk !\n", + max ? "max" : "min"); + return ret; + } + } + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT) && + (feature_mask & FEATURE_DPM_DCEFCLK_MASK)) { + single_dpm_table = &(dpm_table->dcef_table); + freq = single_dpm_table->dpm_state.hard_min_level; + if (!max) { + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetHardMinByFreq, + (PPCLK_DCEFCLK << 16) | (freq & 0xffff)); + if (ret) { + pr_err("Failed to set hard min dcefclk !\n"); + return ret; + } + } + } + + return ret; +} + +static int vega20_force_clk_levels(struct smu_context *smu, + enum smu_clk_type clk_type, uint32_t mask) +{ + struct vega20_dpm_table *dpm_table; + struct vega20_single_dpm_table *single_dpm_table; + uint32_t soft_min_level, soft_max_level, hard_min_level; + int ret = 0; + + soft_min_level = mask ? (ffs(mask) - 1) : 0; + soft_max_level = mask ? (fls(mask) - 1) : 0; + + dpm_table = smu->smu_dpm.dpm_context; + + switch (clk_type) { + case SMU_SCLK: + single_dpm_table = &(dpm_table->gfx_table); + + if (soft_max_level >= single_dpm_table->count) { + pr_err("Clock level specified %d is over max allowed %d\n", + soft_max_level, single_dpm_table->count - 1); + ret = -EINVAL; + break; + } + + single_dpm_table->dpm_state.soft_min_level = + single_dpm_table->dpm_levels[soft_min_level].value; + single_dpm_table->dpm_state.soft_max_level = + single_dpm_table->dpm_levels[soft_max_level].value; + + ret = vega20_upload_dpm_level(smu, false, FEATURE_DPM_GFXCLK_MASK); + if (ret) { + pr_err("Failed to upload boot level to lowest!\n"); + break; + } + + ret = vega20_upload_dpm_level(smu, true, FEATURE_DPM_GFXCLK_MASK); + if (ret) + pr_err("Failed to upload dpm max level to highest!\n"); + + break; + + case SMU_MCLK: + single_dpm_table = &(dpm_table->mem_table); + + if (soft_max_level >= single_dpm_table->count) { + pr_err("Clock level specified %d is over max allowed %d\n", + soft_max_level, single_dpm_table->count - 1); + ret = -EINVAL; + break; + } + + single_dpm_table->dpm_state.soft_min_level = + single_dpm_table->dpm_levels[soft_min_level].value; + single_dpm_table->dpm_state.soft_max_level = + single_dpm_table->dpm_levels[soft_max_level].value; + + ret = vega20_upload_dpm_level(smu, false, FEATURE_DPM_UCLK_MASK); + if (ret) { + pr_err("Failed to upload boot level to lowest!\n"); + break; + } + + ret = vega20_upload_dpm_level(smu, true, FEATURE_DPM_UCLK_MASK); + if (ret) + pr_err("Failed to upload dpm max level to highest!\n"); + + break; + + case SMU_SOCCLK: + single_dpm_table = &(dpm_table->soc_table); + + if (soft_max_level >= single_dpm_table->count) { + pr_err("Clock level specified %d is over max allowed %d\n", + soft_max_level, single_dpm_table->count - 1); + ret = -EINVAL; + break; + } + + single_dpm_table->dpm_state.soft_min_level = + single_dpm_table->dpm_levels[soft_min_level].value; + single_dpm_table->dpm_state.soft_max_level = + single_dpm_table->dpm_levels[soft_max_level].value; + + ret = vega20_upload_dpm_level(smu, false, FEATURE_DPM_SOCCLK_MASK); + if (ret) { + pr_err("Failed to upload boot level to lowest!\n"); + break; + } + + ret = vega20_upload_dpm_level(smu, true, FEATURE_DPM_SOCCLK_MASK); + if (ret) + pr_err("Failed to upload dpm max level to highest!\n"); + + break; + + case SMU_FCLK: + single_dpm_table = &(dpm_table->fclk_table); + + if (soft_max_level >= single_dpm_table->count) { + pr_err("Clock level specified %d is over max allowed %d\n", + soft_max_level, single_dpm_table->count - 1); + ret = -EINVAL; + break; + } + + single_dpm_table->dpm_state.soft_min_level = + single_dpm_table->dpm_levels[soft_min_level].value; + single_dpm_table->dpm_state.soft_max_level = + single_dpm_table->dpm_levels[soft_max_level].value; + + ret = vega20_upload_dpm_level(smu, false, FEATURE_DPM_FCLK_MASK); + if (ret) { + pr_err("Failed to upload boot level to lowest!\n"); + break; + } + + ret = vega20_upload_dpm_level(smu, true, FEATURE_DPM_FCLK_MASK); + if (ret) + pr_err("Failed to upload dpm max level to highest!\n"); + + break; + + case SMU_DCEFCLK: + hard_min_level = soft_min_level; + single_dpm_table = &(dpm_table->dcef_table); + + if (hard_min_level >= single_dpm_table->count) { + pr_err("Clock level specified %d is over max allowed %d\n", + hard_min_level, single_dpm_table->count - 1); + ret = -EINVAL; + break; + } + + single_dpm_table->dpm_state.hard_min_level = + single_dpm_table->dpm_levels[hard_min_level].value; + + ret = vega20_upload_dpm_level(smu, false, FEATURE_DPM_DCEFCLK_MASK); + if (ret) + pr_err("Failed to upload boot level to lowest!\n"); + + break; + + case SMU_PCIE: + if (soft_min_level >= NUM_LINK_LEVELS || + soft_max_level >= NUM_LINK_LEVELS) { + ret = -EINVAL; + break; + } + + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetMinLinkDpmByIndex, soft_min_level); + if (ret) + pr_err("Failed to set min link dpm level!\n"); + + break; + + default: + break; + } + + return ret; +} + +static int vega20_get_clock_by_type_with_latency(struct smu_context *smu, + enum smu_clk_type clk_type, + struct pp_clock_levels_with_latency *clocks) +{ + int ret; + struct vega20_single_dpm_table *single_dpm_table; + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + struct vega20_dpm_table *dpm_table = NULL; + + dpm_table = smu_dpm->dpm_context; + + switch (clk_type) { + case SMU_GFXCLK: + single_dpm_table = &(dpm_table->gfx_table); + ret = vega20_get_clk_table(smu, clocks, single_dpm_table); + break; + case SMU_MCLK: + single_dpm_table = &(dpm_table->mem_table); + ret = vega20_get_clk_table(smu, clocks, single_dpm_table); + break; + case SMU_DCEFCLK: + single_dpm_table = &(dpm_table->dcef_table); + ret = vega20_get_clk_table(smu, clocks, single_dpm_table); + break; + case SMU_SOCCLK: + single_dpm_table = &(dpm_table->soc_table); + ret = vega20_get_clk_table(smu, clocks, single_dpm_table); + break; + default: + ret = -EINVAL; + } + + return ret; +} + +static int vega20_overdrive_get_gfx_clk_base_voltage(struct smu_context *smu, + uint32_t *voltage, + uint32_t freq) +{ + int ret; + + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_GetAVFSVoltageByDpm, + ((AVFS_CURVE << 24) | (OD8_HOTCURVE_TEMPERATURE << 16) | freq)); + if (ret) { + pr_err("[GetBaseVoltage] failed to get GFXCLK AVFS voltage from SMU!"); + return ret; + } + + smu_read_smc_arg(smu, voltage); + *voltage = *voltage / VOLTAGE_SCALE; + + return 0; +} + +static int vega20_set_default_od8_setttings(struct smu_context *smu) +{ + struct smu_table_context *table_context = &smu->smu_table; + OverDriveTable_t *od_table = (OverDriveTable_t *)(table_context->overdrive_table); + struct vega20_od8_settings *od8_settings = NULL; + PPTable_t *smc_pptable = table_context->driver_pptable; + int i, ret; + + if (smu->od_settings) + return -EINVAL; + + od8_settings = kzalloc(sizeof(struct vega20_od8_settings), GFP_KERNEL); + + if (!od8_settings) + return -ENOMEM; + + smu->od_settings = (void *)od8_settings; + + ret = vega20_setup_od8_information(smu); + if (ret) { + pr_err("Retrieve board OD limits failed!\n"); + return ret; + } + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_SOCCLK_BIT)) { + if (od8_settings->od_feature_capabilities[ATOM_VEGA20_ODFEATURE_GFXCLK_LIMITS] && + od8_settings->od_settings_max[OD8_SETTING_GFXCLK_FMAX] > 0 && + od8_settings->od_settings_min[OD8_SETTING_GFXCLK_FMIN] > 0 && + (od8_settings->od_settings_max[OD8_SETTING_GFXCLK_FMAX] >= + od8_settings->od_settings_min[OD8_SETTING_GFXCLK_FMIN])) { + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMIN].feature_id = + OD8_GFXCLK_LIMITS; + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMAX].feature_id = + OD8_GFXCLK_LIMITS; + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMIN].default_value = + od_table->GfxclkFmin; + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMAX].default_value = + od_table->GfxclkFmax; + } + + if (od8_settings->od_feature_capabilities[ATOM_VEGA20_ODFEATURE_GFXCLK_CURVE] && + (od8_settings->od_settings_min[OD8_SETTING_GFXCLK_VOLTAGE1] >= + smc_pptable->MinVoltageGfx / VOLTAGE_SCALE) && + (od8_settings->od_settings_max[OD8_SETTING_GFXCLK_VOLTAGE3] <= + smc_pptable->MaxVoltageGfx / VOLTAGE_SCALE) && + (od8_settings->od_settings_min[OD8_SETTING_GFXCLK_VOLTAGE1] <= + od8_settings->od_settings_max[OD8_SETTING_GFXCLK_VOLTAGE3])) { + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ1].feature_id = + OD8_GFXCLK_CURVE; + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE1].feature_id = + OD8_GFXCLK_CURVE; + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ2].feature_id = + OD8_GFXCLK_CURVE; + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE2].feature_id = + OD8_GFXCLK_CURVE; + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ3].feature_id = + OD8_GFXCLK_CURVE; + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE3].feature_id = + OD8_GFXCLK_CURVE; + + od_table->GfxclkFreq1 = od_table->GfxclkFmin; + od_table->GfxclkFreq2 = (od_table->GfxclkFmin + od_table->GfxclkFmax) / 2; + od_table->GfxclkFreq3 = od_table->GfxclkFmax; + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ1].default_value = + od_table->GfxclkFreq1; + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ2].default_value = + od_table->GfxclkFreq2; + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ3].default_value = + od_table->GfxclkFreq3; + + ret = vega20_overdrive_get_gfx_clk_base_voltage(smu, + &od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE1].default_value, + od_table->GfxclkFreq1); + if (ret) + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE1].default_value = 0; + od_table->GfxclkVolt1 = + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE1].default_value + * VOLTAGE_SCALE; + ret = vega20_overdrive_get_gfx_clk_base_voltage(smu, + &od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE2].default_value, + od_table->GfxclkFreq2); + if (ret) + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE2].default_value = 0; + od_table->GfxclkVolt2 = + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE2].default_value + * VOLTAGE_SCALE; + ret = vega20_overdrive_get_gfx_clk_base_voltage(smu, + &od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE3].default_value, + od_table->GfxclkFreq3); + if (ret) + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE3].default_value = 0; + od_table->GfxclkVolt3 = + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE3].default_value + * VOLTAGE_SCALE; + } + } + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT)) { + if (od8_settings->od_feature_capabilities[ATOM_VEGA20_ODFEATURE_UCLK_MAX] && + od8_settings->od_settings_min[OD8_SETTING_UCLK_FMAX] > 0 && + od8_settings->od_settings_max[OD8_SETTING_UCLK_FMAX] > 0 && + (od8_settings->od_settings_max[OD8_SETTING_UCLK_FMAX] >= + od8_settings->od_settings_min[OD8_SETTING_UCLK_FMAX])) { + od8_settings->od8_settings_array[OD8_SETTING_UCLK_FMAX].feature_id = + OD8_UCLK_MAX; + od8_settings->od8_settings_array[OD8_SETTING_UCLK_FMAX].default_value = + od_table->UclkFmax; + } + } + + if (od8_settings->od_feature_capabilities[ATOM_VEGA20_ODFEATURE_POWER_LIMIT] && + od8_settings->od_settings_min[OD8_SETTING_POWER_PERCENTAGE] > 0 && + od8_settings->od_settings_min[OD8_SETTING_POWER_PERCENTAGE] <= 100 && + od8_settings->od_settings_max[OD8_SETTING_POWER_PERCENTAGE] > 0 && + od8_settings->od_settings_max[OD8_SETTING_POWER_PERCENTAGE] <= 100) { + od8_settings->od8_settings_array[OD8_SETTING_POWER_PERCENTAGE].feature_id = + OD8_POWER_LIMIT; + od8_settings->od8_settings_array[OD8_SETTING_POWER_PERCENTAGE].default_value = + od_table->OverDrivePct; + } + + if (smu_feature_is_enabled(smu, SMU_FEATURE_FAN_CONTROL_BIT)) { + if (od8_settings->od_feature_capabilities[ATOM_VEGA20_ODFEATURE_FAN_ACOUSTIC_LIMIT] && + od8_settings->od_settings_min[OD8_SETTING_FAN_ACOUSTIC_LIMIT] > 0 && + od8_settings->od_settings_max[OD8_SETTING_FAN_ACOUSTIC_LIMIT] > 0 && + (od8_settings->od_settings_max[OD8_SETTING_FAN_ACOUSTIC_LIMIT] >= + od8_settings->od_settings_min[OD8_SETTING_FAN_ACOUSTIC_LIMIT])) { + od8_settings->od8_settings_array[OD8_SETTING_FAN_ACOUSTIC_LIMIT].feature_id = + OD8_ACOUSTIC_LIMIT_SCLK; + od8_settings->od8_settings_array[OD8_SETTING_FAN_ACOUSTIC_LIMIT].default_value = + od_table->FanMaximumRpm; + } + + if (od8_settings->od_feature_capabilities[ATOM_VEGA20_ODFEATURE_FAN_SPEED_MIN] && + od8_settings->od_settings_min[OD8_SETTING_FAN_MIN_SPEED] > 0 && + od8_settings->od_settings_max[OD8_SETTING_FAN_MIN_SPEED] > 0 && + (od8_settings->od_settings_max[OD8_SETTING_FAN_MIN_SPEED] >= + od8_settings->od_settings_min[OD8_SETTING_FAN_MIN_SPEED])) { + od8_settings->od8_settings_array[OD8_SETTING_FAN_MIN_SPEED].feature_id = + OD8_FAN_SPEED_MIN; + od8_settings->od8_settings_array[OD8_SETTING_FAN_MIN_SPEED].default_value = + od_table->FanMinimumPwm * smc_pptable->FanMaximumRpm / 100; + } + } + + if (smu_feature_is_enabled(smu, SMU_FEATURE_THERMAL_BIT)) { + if (od8_settings->od_feature_capabilities[ATOM_VEGA20_ODFEATURE_TEMPERATURE_FAN] && + od8_settings->od_settings_min[OD8_SETTING_FAN_TARGET_TEMP] > 0 && + od8_settings->od_settings_max[OD8_SETTING_FAN_TARGET_TEMP] > 0 && + (od8_settings->od_settings_max[OD8_SETTING_FAN_TARGET_TEMP] >= + od8_settings->od_settings_min[OD8_SETTING_FAN_TARGET_TEMP])) { + od8_settings->od8_settings_array[OD8_SETTING_FAN_TARGET_TEMP].feature_id = + OD8_TEMPERATURE_FAN; + od8_settings->od8_settings_array[OD8_SETTING_FAN_TARGET_TEMP].default_value = + od_table->FanTargetTemperature; + } + + if (od8_settings->od_feature_capabilities[ATOM_VEGA20_ODFEATURE_TEMPERATURE_SYSTEM] && + od8_settings->od_settings_min[OD8_SETTING_OPERATING_TEMP_MAX] > 0 && + od8_settings->od_settings_max[OD8_SETTING_OPERATING_TEMP_MAX] > 0 && + (od8_settings->od_settings_max[OD8_SETTING_OPERATING_TEMP_MAX] >= + od8_settings->od_settings_min[OD8_SETTING_OPERATING_TEMP_MAX])) { + od8_settings->od8_settings_array[OD8_SETTING_OPERATING_TEMP_MAX].feature_id = + OD8_TEMPERATURE_SYSTEM; + od8_settings->od8_settings_array[OD8_SETTING_OPERATING_TEMP_MAX].default_value = + od_table->MaxOpTemp; + } + } + + for (i = 0; i < OD8_SETTING_COUNT; i++) { + if (od8_settings->od8_settings_array[i].feature_id) { + od8_settings->od8_settings_array[i].min_value = + od8_settings->od_settings_min[i]; + od8_settings->od8_settings_array[i].max_value = + od8_settings->od_settings_max[i]; + od8_settings->od8_settings_array[i].current_value = + od8_settings->od8_settings_array[i].default_value; + } else { + od8_settings->od8_settings_array[i].min_value = 0; + od8_settings->od8_settings_array[i].max_value = 0; + od8_settings->od8_settings_array[i].current_value = 0; + } + } + + return 0; +} + +static int vega20_get_metrics_table(struct smu_context *smu, + SmuMetrics_t *metrics_table) +{ + struct smu_table_context *smu_table= &smu->smu_table; + int ret = 0; + + if (!smu_table->metrics_time || time_after(jiffies, smu_table->metrics_time + HZ / 1000)) { + 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"); + return ret; + } + smu_table->metrics_time = jiffies; + } + + memcpy(metrics_table, smu_table->metrics_table, sizeof(SmuMetrics_t)); + + return ret; +} + +static int vega20_set_default_od_settings(struct smu_context *smu, + bool initialize) +{ + struct smu_table_context *table_context = &smu->smu_table; + int ret; + + if (initialize) { + if (table_context->overdrive_table) + return -EINVAL; + + table_context->overdrive_table = kzalloc(sizeof(OverDriveTable_t), GFP_KERNEL); + + if (!table_context->overdrive_table) + return -ENOMEM; + + ret = smu_update_table(smu, SMU_TABLE_OVERDRIVE, 0, + table_context->overdrive_table, false); + if (ret) { + pr_err("Failed to export over drive table!\n"); + return ret; + } + + ret = vega20_set_default_od8_setttings(smu); + if (ret) + return ret; + } + + ret = smu_update_table(smu, SMU_TABLE_OVERDRIVE, 0, + table_context->overdrive_table, true); + if (ret) { + pr_err("Failed to import over drive table!\n"); + return ret; + } + + return 0; +} + +static int vega20_get_od_percentage(struct smu_context *smu, + enum smu_clk_type clk_type) +{ + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + struct vega20_dpm_table *dpm_table = NULL; + struct vega20_dpm_table *golden_table = NULL; + struct vega20_single_dpm_table *single_dpm_table; + struct vega20_single_dpm_table *golden_dpm_table; + int value, golden_value; + + dpm_table = smu_dpm->dpm_context; + golden_table = smu_dpm->golden_dpm_context; + + switch (clk_type) { + case SMU_OD_SCLK: + single_dpm_table = &(dpm_table->gfx_table); + golden_dpm_table = &(golden_table->gfx_table); + break; + case SMU_OD_MCLK: + single_dpm_table = &(dpm_table->mem_table); + golden_dpm_table = &(golden_table->mem_table); + break; + default: + return -EINVAL; + break; + } + + value = single_dpm_table->dpm_levels[single_dpm_table->count - 1].value; + golden_value = golden_dpm_table->dpm_levels[golden_dpm_table->count - 1].value; + + value -= golden_value; + value = DIV_ROUND_UP(value * 100, golden_value); + + return value; +} + +static int vega20_get_power_profile_mode(struct smu_context *smu, char *buf) +{ + DpmActivityMonitorCoeffInt_t activity_monitor; + uint32_t i, size = 0; + int16_t workload_type = 0; + static const char *profile_name[] = { + "BOOTUP_DEFAULT", + "3D_FULL_SCREEN", + "POWER_SAVING", + "VIDEO", + "VR", + "COMPUTE", + "CUSTOM"}; + static const char *title[] = { + "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"}; + int result = 0; + + if (!smu->pm_enabled || !buf) + return -EINVAL; + + 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]); + + for (i = 0; i <= PP_SMC_POWER_PROFILE_CUSTOM; i++) { + /* conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT */ + workload_type = smu_workload_get_type(smu, i); + if (workload_type < 0) + return -EINVAL; + + 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) ? "*" : " "); + + 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, + "SOCCLK", + activity_monitor.Soc_FPS, + activity_monitor.Soc_UseRlcBusy, + activity_monitor.Soc_MinActiveFreqType, + activity_monitor.Soc_MinActiveFreq, + activity_monitor.Soc_BoosterFreqType, + activity_monitor.Soc_BoosterFreq, + activity_monitor.Soc_PD_Data_limit_c, + activity_monitor.Soc_PD_Data_error_coeff, + activity_monitor.Soc_PD_Data_error_rate_coeff); + + size += sprintf(buf + size, "%19s %d(%13s) %7d %7d %7d %7d %7d %7d %7d %7d %7d\n", + " ", + 2, + "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); + + size += sprintf(buf + size, "%19s %d(%13s) %7d %7d %7d %7d %7d %7d %7d %7d %7d\n", + " ", + 3, + "FCLK", + activity_monitor.Fclk_FPS, + activity_monitor.Fclk_UseRlcBusy, + activity_monitor.Fclk_MinActiveFreqType, + activity_monitor.Fclk_MinActiveFreq, + activity_monitor.Fclk_BoosterFreqType, + activity_monitor.Fclk_BoosterFreq, + activity_monitor.Fclk_PD_Data_limit_c, + activity_monitor.Fclk_PD_Data_error_coeff, + activity_monitor.Fclk_PD_Data_error_rate_coeff); + } + + return size; +} + +static int vega20_set_power_profile_mode(struct smu_context *smu, long *input, uint32_t size) +{ + DpmActivityMonitorCoeffInt_t activity_monitor; + int workload_type = 0, ret = 0; + + smu->power_profile_mode = input[size]; + + if (!smu->pm_enabled) + return ret; + if (smu->power_profile_mode > PP_SMC_POWER_PROFILE_CUSTOM) { + pr_err("Invalid power profile mode %d\n", smu->power_profile_mode); + return -EINVAL; + } + + if (smu->power_profile_mode == PP_SMC_POWER_PROFILE_CUSTOM) { + 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: /* Socclk */ + activity_monitor.Soc_FPS = input[1]; + activity_monitor.Soc_UseRlcBusy = input[2]; + activity_monitor.Soc_MinActiveFreqType = input[3]; + activity_monitor.Soc_MinActiveFreq = input[4]; + activity_monitor.Soc_BoosterFreqType = input[5]; + activity_monitor.Soc_BoosterFreq = input[6]; + activity_monitor.Soc_PD_Data_limit_c = input[7]; + activity_monitor.Soc_PD_Data_error_coeff = input[8]; + activity_monitor.Soc_PD_Data_error_rate_coeff = input[9]; + break; + case 2: /* 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; + case 3: /* Fclk */ + activity_monitor.Fclk_FPS = input[1]; + activity_monitor.Fclk_UseRlcBusy = input[2]; + activity_monitor.Fclk_MinActiveFreqType = input[3]; + activity_monitor.Fclk_MinActiveFreq = input[4]; + activity_monitor.Fclk_BoosterFreqType = input[5]; + activity_monitor.Fclk_BoosterFreq = input[6]; + activity_monitor.Fclk_PD_Data_limit_c = input[7]; + activity_monitor.Fclk_PD_Data_error_coeff = input[8]; + activity_monitor.Fclk_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 */ + workload_type = smu_workload_get_type(smu, smu->power_profile_mode); + if (workload_type < 0) + return -EINVAL; + smu_send_smc_msg_with_param(smu, SMU_MSG_SetWorkloadMask, + 1 << workload_type); + + return ret; +} + +static int +vega20_get_profiling_clk_mask(struct smu_context *smu, + enum amd_dpm_forced_level level, + uint32_t *sclk_mask, + uint32_t *mclk_mask, + uint32_t *soc_mask) +{ + struct vega20_dpm_table *dpm_table = (struct vega20_dpm_table *)smu->smu_dpm.dpm_context; + struct vega20_single_dpm_table *gfx_dpm_table; + struct vega20_single_dpm_table *mem_dpm_table; + struct vega20_single_dpm_table *soc_dpm_table; + + if (!smu->smu_dpm.dpm_context) + return -EINVAL; + + gfx_dpm_table = &dpm_table->gfx_table; + mem_dpm_table = &dpm_table->mem_table; + soc_dpm_table = &dpm_table->soc_table; + + *sclk_mask = 0; + *mclk_mask = 0; + *soc_mask = 0; + + if (gfx_dpm_table->count > VEGA20_UMD_PSTATE_GFXCLK_LEVEL && + mem_dpm_table->count > VEGA20_UMD_PSTATE_MCLK_LEVEL && + soc_dpm_table->count > VEGA20_UMD_PSTATE_SOCCLK_LEVEL) { + *sclk_mask = VEGA20_UMD_PSTATE_GFXCLK_LEVEL; + *mclk_mask = VEGA20_UMD_PSTATE_MCLK_LEVEL; + *soc_mask = VEGA20_UMD_PSTATE_SOCCLK_LEVEL; + } + + if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK) { + *sclk_mask = 0; + } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK) { + *mclk_mask = 0; + } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) { + *sclk_mask = gfx_dpm_table->count - 1; + *mclk_mask = mem_dpm_table->count - 1; + *soc_mask = soc_dpm_table->count - 1; + } + + return 0; +} + +static int +vega20_set_uclk_to_highest_dpm_level(struct smu_context *smu, + struct vega20_single_dpm_table *dpm_table) +{ + int ret = 0; + struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm); + if (!smu_dpm_ctx->dpm_context) + return -EINVAL; + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT)) { + if (dpm_table->count <= 0) { + pr_err("[%s] Dpm table has no entry!", __func__); + return -EINVAL; + } + + if (dpm_table->count > NUM_UCLK_DPM_LEVELS) { + pr_err("[%s] Dpm table has too many entries!", __func__); + return -EINVAL; + } + + dpm_table->dpm_state.hard_min_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetHardMinByFreq, + (PPCLK_UCLK << 16) | dpm_table->dpm_state.hard_min_level); + if (ret) { + pr_err("[%s] Set hard min uclk failed!", __func__); + return ret; + } + } + + return ret; +} + +static int vega20_pre_display_config_changed(struct smu_context *smu) +{ + int ret = 0; + struct vega20_dpm_table *dpm_table = smu->smu_dpm.dpm_context; + + if (!smu->smu_dpm.dpm_context) + return -EINVAL; + + smu_send_smc_msg_with_param(smu, SMU_MSG_NumOfDisplays, 0); + ret = vega20_set_uclk_to_highest_dpm_level(smu, + &dpm_table->mem_table); + if (ret) + pr_err("Failed to set uclk to highest dpm level"); + return ret; +} + +static int vega20_display_config_changed(struct smu_context *smu) +{ + int ret = 0; + + 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; + } + + if ((smu->watermarks_bitmap & WATERMARKS_EXIST) && + smu_feature_is_supported(smu, SMU_FEATURE_DPM_DCEFCLK_BIT) && + smu_feature_is_supported(smu, SMU_FEATURE_DPM_SOCCLK_BIT)) { + smu_send_smc_msg_with_param(smu, + SMU_MSG_NumOfDisplays, + smu->display_config->num_display); + } + + return ret; +} + +static int vega20_apply_clocks_adjust_rules(struct smu_context *smu) +{ + struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm); + struct vega20_dpm_table *dpm_ctx = (struct vega20_dpm_table *)(smu_dpm_ctx->dpm_context); + struct vega20_single_dpm_table *dpm_table; + bool vblank_too_short = false; + bool disable_mclk_switching; + uint32_t i, latency; + + disable_mclk_switching = ((1 < smu->display_config->num_display) && + !smu->display_config->multi_monitor_in_sync) || vblank_too_short; + latency = smu->display_config->dce_tolerable_mclk_in_active_latency; + + /* gfxclk */ + dpm_table = &(dpm_ctx->gfx_table); + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[0].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + dpm_table->dpm_state.hard_min_level = dpm_table->dpm_levels[0].value; + dpm_table->dpm_state.hard_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + + if (VEGA20_UMD_PSTATE_GFXCLK_LEVEL < dpm_table->count) { + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[VEGA20_UMD_PSTATE_GFXCLK_LEVEL].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[VEGA20_UMD_PSTATE_GFXCLK_LEVEL].value; + } + + if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK) { + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[0].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[0].value; + } + + if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) { + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + } + + /* memclk */ + dpm_table = &(dpm_ctx->mem_table); + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[0].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + dpm_table->dpm_state.hard_min_level = dpm_table->dpm_levels[0].value; + dpm_table->dpm_state.hard_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + + if (VEGA20_UMD_PSTATE_MCLK_LEVEL < dpm_table->count) { + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[VEGA20_UMD_PSTATE_MCLK_LEVEL].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[VEGA20_UMD_PSTATE_MCLK_LEVEL].value; + } + + if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK) { + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[0].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[0].value; + } + + if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) { + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + } + + /* honour DAL's UCLK Hardmin */ + if (dpm_table->dpm_state.hard_min_level < (smu->display_config->min_mem_set_clock / 100)) + dpm_table->dpm_state.hard_min_level = smu->display_config->min_mem_set_clock / 100; + + /* Hardmin is dependent on displayconfig */ + if (disable_mclk_switching) { + dpm_table->dpm_state.hard_min_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + for (i = 0; i < smu_dpm_ctx->mclk_latency_table->count - 1; i++) { + if (smu_dpm_ctx->mclk_latency_table->entries[i].latency <= latency) { + if (dpm_table->dpm_levels[i].value >= (smu->display_config->min_mem_set_clock / 100)) { + dpm_table->dpm_state.hard_min_level = dpm_table->dpm_levels[i].value; + break; + } + } + } + } + + if (smu->display_config->nb_pstate_switch_disable) + dpm_table->dpm_state.hard_min_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + + /* vclk */ + dpm_table = &(dpm_ctx->vclk_table); + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[0].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + dpm_table->dpm_state.hard_min_level = dpm_table->dpm_levels[0].value; + dpm_table->dpm_state.hard_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + + if (VEGA20_UMD_PSTATE_UVDCLK_LEVEL < dpm_table->count) { + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[VEGA20_UMD_PSTATE_UVDCLK_LEVEL].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[VEGA20_UMD_PSTATE_UVDCLK_LEVEL].value; + } + + if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) { + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + } + + /* dclk */ + dpm_table = &(dpm_ctx->dclk_table); + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[0].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + dpm_table->dpm_state.hard_min_level = dpm_table->dpm_levels[0].value; + dpm_table->dpm_state.hard_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + + if (VEGA20_UMD_PSTATE_UVDCLK_LEVEL < dpm_table->count) { + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[VEGA20_UMD_PSTATE_UVDCLK_LEVEL].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[VEGA20_UMD_PSTATE_UVDCLK_LEVEL].value; + } + + if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) { + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + } + + /* socclk */ + dpm_table = &(dpm_ctx->soc_table); + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[0].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + dpm_table->dpm_state.hard_min_level = dpm_table->dpm_levels[0].value; + dpm_table->dpm_state.hard_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + + if (VEGA20_UMD_PSTATE_SOCCLK_LEVEL < dpm_table->count) { + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[VEGA20_UMD_PSTATE_SOCCLK_LEVEL].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[VEGA20_UMD_PSTATE_SOCCLK_LEVEL].value; + } + + if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) { + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + } + + /* eclk */ + dpm_table = &(dpm_ctx->eclk_table); + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[0].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + dpm_table->dpm_state.hard_min_level = dpm_table->dpm_levels[0].value; + dpm_table->dpm_state.hard_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + + if (VEGA20_UMD_PSTATE_VCEMCLK_LEVEL < dpm_table->count) { + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[VEGA20_UMD_PSTATE_VCEMCLK_LEVEL].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[VEGA20_UMD_PSTATE_VCEMCLK_LEVEL].value; + } + + if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) { + dpm_table->dpm_state.soft_min_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + dpm_table->dpm_state.soft_max_level = dpm_table->dpm_levels[dpm_table->count - 1].value; + } + return 0; +} + +static int +vega20_notify_smc_dispaly_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; + struct smu_clocks min_clocks = {0}; + struct pp_display_clock_request clock_req; + int ret = 0; + + min_clocks.dcef_clock = smu->display_config->min_dcef_set_clk; + min_clocks.dcef_clock_in_sr = smu->display_config->min_dcef_deep_sleep_set_clk; + min_clocks.memory_clock = smu->display_config->min_mem_set_clock; + + if (smu_feature_is_supported(smu, SMU_FEATURE_DPM_DCEFCLK_BIT)) { + clock_req.clock_type = amd_pp_dcef_clock; + clock_req.clock_freq_in_khz = min_clocks.dcef_clock * 10; + if (!smu_v11_0_display_clock_voltage_request(smu, &clock_req)) { + if (smu_feature_is_supported(smu, SMU_FEATURE_DS_DCEFCLK_BIT)) { + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetMinDeepSleepDcefclk, + min_clocks.dcef_clock_in_sr/100); + if (ret) { + pr_err("Attempt to set divider for DCEFCLK Failed!"); + return ret; + } + } + } else { + pr_info("Attempt to set Hard Min for DCEFCLK Failed!"); + } + } + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT)) { + memtable->dpm_state.hard_min_level = min_clocks.memory_clock/100; + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_SetHardMinByFreq, + (PPCLK_UCLK << 16) | memtable->dpm_state.hard_min_level); + if (ret) { + pr_err("[%s] Set hard min uclk failed!", __func__); + return ret; + } + } + + return 0; +} + +static uint32_t vega20_find_lowest_dpm_level(struct vega20_single_dpm_table *table) +{ + uint32_t i; + + for (i = 0; i < table->count; i++) { + if (table->dpm_levels[i].enabled) + break; + } + if (i >= table->count) { + i = 0; + table->dpm_levels[i].enabled = true; + } + + return i; +} + +static uint32_t vega20_find_highest_dpm_level(struct vega20_single_dpm_table *table) +{ + int i = 0; + + if (!table) { + pr_err("[%s] DPM Table does not exist!", __func__); + return 0; + } + if (table->count <= 0) { + pr_err("[%s] DPM Table has no entry!", __func__); + return 0; + } + if (table->count > MAX_REGULAR_DPM_NUMBER) { + pr_err("[%s] DPM Table has too many entries!", __func__); + return MAX_REGULAR_DPM_NUMBER - 1; + } + + for (i = table->count - 1; i >= 0; i--) { + if (table->dpm_levels[i].enabled) + break; + } + if (i < 0) { + i = 0; + table->dpm_levels[i].enabled = true; + } + + return i; +} + +static int vega20_force_dpm_limit_value(struct smu_context *smu, bool highest) +{ + uint32_t soft_level; + int ret = 0; + struct vega20_dpm_table *dpm_table = + (struct vega20_dpm_table *)smu->smu_dpm.dpm_context; + + if (highest) + soft_level = vega20_find_highest_dpm_level(&(dpm_table->gfx_table)); + else + soft_level = vega20_find_lowest_dpm_level(&(dpm_table->gfx_table)); + + dpm_table->gfx_table.dpm_state.soft_min_level = + dpm_table->gfx_table.dpm_state.soft_max_level = + dpm_table->gfx_table.dpm_levels[soft_level].value; + + if (highest) + soft_level = vega20_find_highest_dpm_level(&(dpm_table->mem_table)); + else + soft_level = vega20_find_lowest_dpm_level(&(dpm_table->mem_table)); + + dpm_table->mem_table.dpm_state.soft_min_level = + dpm_table->mem_table.dpm_state.soft_max_level = + dpm_table->mem_table.dpm_levels[soft_level].value; + + if (highest) + soft_level = vega20_find_highest_dpm_level(&(dpm_table->soc_table)); + else + soft_level = vega20_find_lowest_dpm_level(&(dpm_table->soc_table)); + + dpm_table->soc_table.dpm_state.soft_min_level = + dpm_table->soc_table.dpm_state.soft_max_level = + dpm_table->soc_table.dpm_levels[soft_level].value; + + ret = vega20_upload_dpm_level(smu, false, 0xFFFFFFFF); + if (ret) { + pr_err("Failed to upload boot level to %s!\n", + highest ? "highest" : "lowest"); + return ret; + } + + ret = vega20_upload_dpm_level(smu, true, 0xFFFFFFFF); + if (ret) { + pr_err("Failed to upload dpm max level to %s!\n!", + highest ? "highest" : "lowest"); + return ret; + } + + return ret; +} + +static int vega20_unforce_dpm_levels(struct smu_context *smu) +{ + uint32_t soft_min_level, soft_max_level; + int ret = 0; + struct vega20_dpm_table *dpm_table = + (struct vega20_dpm_table *)smu->smu_dpm.dpm_context; + + soft_min_level = vega20_find_lowest_dpm_level(&(dpm_table->gfx_table)); + soft_max_level = vega20_find_highest_dpm_level(&(dpm_table->gfx_table)); + dpm_table->gfx_table.dpm_state.soft_min_level = + dpm_table->gfx_table.dpm_levels[soft_min_level].value; + dpm_table->gfx_table.dpm_state.soft_max_level = + dpm_table->gfx_table.dpm_levels[soft_max_level].value; + + soft_min_level = vega20_find_lowest_dpm_level(&(dpm_table->mem_table)); + soft_max_level = vega20_find_highest_dpm_level(&(dpm_table->mem_table)); + dpm_table->mem_table.dpm_state.soft_min_level = + dpm_table->gfx_table.dpm_levels[soft_min_level].value; + dpm_table->mem_table.dpm_state.soft_max_level = + dpm_table->gfx_table.dpm_levels[soft_max_level].value; + + soft_min_level = vega20_find_lowest_dpm_level(&(dpm_table->soc_table)); + soft_max_level = vega20_find_highest_dpm_level(&(dpm_table->soc_table)); + dpm_table->soc_table.dpm_state.soft_min_level = + dpm_table->soc_table.dpm_levels[soft_min_level].value; + dpm_table->soc_table.dpm_state.soft_max_level = + dpm_table->soc_table.dpm_levels[soft_max_level].value; + + ret = vega20_upload_dpm_level(smu, false, 0xFFFFFFFF); + if (ret) { + pr_err("Failed to upload DPM Bootup Levels!"); + return ret; + } + + ret = vega20_upload_dpm_level(smu, true, 0xFFFFFFFF); + if (ret) { + pr_err("Failed to upload DPM Max Levels!"); + return ret; + } + + return ret; +} + +static int vega20_update_specified_od8_value(struct smu_context *smu, + uint32_t index, + uint32_t value) +{ + struct smu_table_context *table_context = &smu->smu_table; + OverDriveTable_t *od_table = + (OverDriveTable_t *)(table_context->overdrive_table); + struct vega20_od8_settings *od8_settings = + (struct vega20_od8_settings *)smu->od_settings; + + switch (index) { + case OD8_SETTING_GFXCLK_FMIN: + od_table->GfxclkFmin = (uint16_t)value; + break; + + case OD8_SETTING_GFXCLK_FMAX: + if (value < od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMAX].min_value || + value > od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMAX].max_value) + return -EINVAL; + od_table->GfxclkFmax = (uint16_t)value; + break; + + case OD8_SETTING_GFXCLK_FREQ1: + od_table->GfxclkFreq1 = (uint16_t)value; + break; + + case OD8_SETTING_GFXCLK_VOLTAGE1: + od_table->GfxclkVolt1 = (uint16_t)value; + break; + + case OD8_SETTING_GFXCLK_FREQ2: + od_table->GfxclkFreq2 = (uint16_t)value; + break; + + case OD8_SETTING_GFXCLK_VOLTAGE2: + od_table->GfxclkVolt2 = (uint16_t)value; + break; + + case OD8_SETTING_GFXCLK_FREQ3: + od_table->GfxclkFreq3 = (uint16_t)value; + break; + + case OD8_SETTING_GFXCLK_VOLTAGE3: + od_table->GfxclkVolt3 = (uint16_t)value; + break; + + case OD8_SETTING_UCLK_FMAX: + if (value < od8_settings->od8_settings_array[OD8_SETTING_UCLK_FMAX].min_value || + value > od8_settings->od8_settings_array[OD8_SETTING_UCLK_FMAX].max_value) + return -EINVAL; + od_table->UclkFmax = (uint16_t)value; + break; + + case OD8_SETTING_POWER_PERCENTAGE: + od_table->OverDrivePct = (int16_t)value; + break; + + case OD8_SETTING_FAN_ACOUSTIC_LIMIT: + od_table->FanMaximumRpm = (uint16_t)value; + break; + + case OD8_SETTING_FAN_MIN_SPEED: + od_table->FanMinimumPwm = (uint16_t)value; + break; + + case OD8_SETTING_FAN_TARGET_TEMP: + od_table->FanTargetTemperature = (uint16_t)value; + break; + + case OD8_SETTING_OPERATING_TEMP_MAX: + od_table->MaxOpTemp = (uint16_t)value; + break; + } + + return 0; +} + +static int vega20_update_od8_settings(struct smu_context *smu, + uint32_t index, + uint32_t value) +{ + struct smu_table_context *table_context = &smu->smu_table; + int ret; + + ret = smu_update_table(smu, SMU_TABLE_OVERDRIVE, 0, + table_context->overdrive_table, false); + if (ret) { + pr_err("Failed to export over drive table!\n"); + return ret; + } + + ret = vega20_update_specified_od8_value(smu, index, value); + if (ret) + return ret; + + ret = smu_update_table(smu, SMU_TABLE_OVERDRIVE, 0, + table_context->overdrive_table, true); + if (ret) { + pr_err("Failed to import over drive table!\n"); + return ret; + } + + return 0; +} + +static int vega20_set_od_percentage(struct smu_context *smu, + enum smu_clk_type clk_type, + uint32_t value) +{ + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + struct vega20_dpm_table *dpm_table = NULL; + struct vega20_dpm_table *golden_table = NULL; + struct vega20_single_dpm_table *single_dpm_table; + struct vega20_single_dpm_table *golden_dpm_table; + uint32_t od_clk, index; + int ret = 0; + int feature_enabled; + PPCLK_e clk_id; + + dpm_table = smu_dpm->dpm_context; + golden_table = smu_dpm->golden_dpm_context; + + switch (clk_type) { + case SMU_OD_SCLK: + single_dpm_table = &(dpm_table->gfx_table); + golden_dpm_table = &(golden_table->gfx_table); + feature_enabled = smu_feature_is_enabled(smu, SMU_FEATURE_DPM_GFXCLK_BIT); + clk_id = PPCLK_GFXCLK; + index = OD8_SETTING_GFXCLK_FMAX; + break; + case SMU_OD_MCLK: + single_dpm_table = &(dpm_table->mem_table); + golden_dpm_table = &(golden_table->mem_table); + feature_enabled = smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT); + clk_id = PPCLK_UCLK; + index = OD8_SETTING_UCLK_FMAX; + break; + default: + ret = -EINVAL; + break; + } + + if (ret) + goto set_od_failed; + + od_clk = golden_dpm_table->dpm_levels[golden_dpm_table->count - 1].value * value; + od_clk /= 100; + od_clk += golden_dpm_table->dpm_levels[golden_dpm_table->count - 1].value; + + ret = vega20_update_od8_settings(smu, index, od_clk); + if (ret) { + pr_err("[Setoverdrive] failed to set od clk!\n"); + goto set_od_failed; + } + + if (feature_enabled) { + ret = vega20_set_single_dpm_table(smu, single_dpm_table, + clk_id); + if (ret) { + pr_err("[Setoverdrive] failed to refresh dpm table!\n"); + goto set_od_failed; + } + } else { + single_dpm_table->count = 1; + single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.gfxclk / 100; + } + + ret = smu_handle_task(smu, smu_dpm->dpm_level, + AMD_PP_TASK_READJUST_POWER_STATE, + false); + +set_od_failed: + return ret; +} + +static int vega20_odn_edit_dpm_table(struct smu_context *smu, + enum PP_OD_DPM_TABLE_COMMAND type, + long *input, uint32_t size) +{ + struct smu_table_context *table_context = &smu->smu_table; + OverDriveTable_t *od_table = + (OverDriveTable_t *)(table_context->overdrive_table); + struct smu_dpm_context *smu_dpm = &smu->smu_dpm; + struct vega20_dpm_table *dpm_table = NULL; + struct vega20_single_dpm_table *single_dpm_table; + struct vega20_od8_settings *od8_settings = + (struct vega20_od8_settings *)smu->od_settings; + struct pp_clock_levels_with_latency clocks; + int32_t input_index, input_clk, input_vol, i; + int od8_id; + int ret = 0; + + dpm_table = smu_dpm->dpm_context; + + if (!input) { + pr_warn("NULL user input for clock and voltage\n"); + return -EINVAL; + } + + switch (type) { + case PP_OD_EDIT_SCLK_VDDC_TABLE: + if (!(od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMIN].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMAX].feature_id)) { + pr_info("Sclk min/max frequency overdrive not supported\n"); + return -EOPNOTSUPP; + } + + for (i = 0; i < size; i += 2) { + if (i + 2 > size) { + pr_info("invalid number of input parameters %d\n", size); + return -EINVAL; + } + + input_index = input[i]; + input_clk = input[i + 1]; + + if (input_index != 0 && input_index != 1) { + pr_info("Invalid index %d\n", input_index); + pr_info("Support min/max sclk frequency settingonly which index by 0/1\n"); + return -EINVAL; + } + + if (input_clk < od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMIN].min_value || + input_clk > od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMAX].max_value) { + pr_info("clock freq %d is not within allowed range [%d - %d]\n", + input_clk, + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMIN].min_value, + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FMAX].max_value); + return -EINVAL; + } + + if (input_index == 0 && od_table->GfxclkFmin != input_clk) { + od_table->GfxclkFmin = input_clk; + od8_settings->od_gfxclk_update = true; + } else if (input_index == 1 && od_table->GfxclkFmax != input_clk) { + od_table->GfxclkFmax = input_clk; + od8_settings->od_gfxclk_update = true; + } + } + + break; + + case PP_OD_EDIT_MCLK_VDDC_TABLE: + if (!od8_settings->od8_settings_array[OD8_SETTING_UCLK_FMAX].feature_id) { + pr_info("Mclk max frequency overdrive not supported\n"); + return -EOPNOTSUPP; + } + + single_dpm_table = &(dpm_table->mem_table); + ret = vega20_get_clk_table(smu, &clocks, single_dpm_table); + if (ret) { + pr_err("Attempt to get memory clk levels Failed!"); + return ret; + } + + for (i = 0; i < size; i += 2) { + if (i + 2 > size) { + pr_info("invalid number of input parameters %d\n", + size); + return -EINVAL; + } + + input_index = input[i]; + input_clk = input[i + 1]; + + if (input_index != 1) { + pr_info("Invalid index %d\n", input_index); + pr_info("Support max Mclk frequency setting only which index by 1\n"); + return -EINVAL; + } + + if (input_clk < clocks.data[0].clocks_in_khz / 1000 || + input_clk > od8_settings->od8_settings_array[OD8_SETTING_UCLK_FMAX].max_value) { + pr_info("clock freq %d is not within allowed range [%d - %d]\n", + input_clk, + clocks.data[0].clocks_in_khz / 1000, + od8_settings->od8_settings_array[OD8_SETTING_UCLK_FMAX].max_value); + return -EINVAL; + } + + if (input_index == 1 && od_table->UclkFmax != input_clk) { + od8_settings->od_gfxclk_update = true; + od_table->UclkFmax = input_clk; + } + } + + break; + + case PP_OD_EDIT_VDDC_CURVE: + if (!(od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ1].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ2].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_FREQ3].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE1].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE2].feature_id && + od8_settings->od8_settings_array[OD8_SETTING_GFXCLK_VOLTAGE3].feature_id)) { + pr_info("Voltage curve calibrate not supported\n"); + return -EOPNOTSUPP; + } + + for (i = 0; i < size; i += 3) { + if (i + 3 > size) { + pr_info("invalid number of input parameters %d\n", + size); + return -EINVAL; + } + + input_index = input[i]; + input_clk = input[i + 1]; + input_vol = input[i + 2]; + + if (input_index > 2) { + pr_info("Setting for point %d is not supported\n", + input_index + 1); + pr_info("Three supported points index by 0, 1, 2\n"); + return -EINVAL; + } + + od8_id = OD8_SETTING_GFXCLK_FREQ1 + 2 * input_index; + if (input_clk < od8_settings->od8_settings_array[od8_id].min_value || + input_clk > od8_settings->od8_settings_array[od8_id].max_value) { + pr_info("clock freq %d is not within allowed range [%d - %d]\n", + input_clk, + od8_settings->od8_settings_array[od8_id].min_value, + od8_settings->od8_settings_array[od8_id].max_value); + return -EINVAL; + } + + od8_id = OD8_SETTING_GFXCLK_VOLTAGE1 + 2 * input_index; + if (input_vol < od8_settings->od8_settings_array[od8_id].min_value || + input_vol > od8_settings->od8_settings_array[od8_id].max_value) { + pr_info("clock voltage %d is not within allowed range [%d- %d]\n", + input_vol, + od8_settings->od8_settings_array[od8_id].min_value, + od8_settings->od8_settings_array[od8_id].max_value); + return -EINVAL; + } + + switch (input_index) { + case 0: + od_table->GfxclkFreq1 = input_clk; + od_table->GfxclkVolt1 = input_vol * VOLTAGE_SCALE; + break; + case 1: + od_table->GfxclkFreq2 = input_clk; + od_table->GfxclkVolt2 = input_vol * VOLTAGE_SCALE; + break; + case 2: + od_table->GfxclkFreq3 = input_clk; + od_table->GfxclkVolt3 = input_vol * VOLTAGE_SCALE; + break; + } + } + + break; + + case PP_OD_RESTORE_DEFAULT_TABLE: + ret = smu_update_table(smu, SMU_TABLE_OVERDRIVE, 0, table_context->overdrive_table, false); + if (ret) { + pr_err("Failed to export over drive table!\n"); + return ret; + } + + break; + + case PP_OD_COMMIT_DPM_TABLE: + ret = smu_update_table(smu, SMU_TABLE_OVERDRIVE, 0, table_context->overdrive_table, true); + if (ret) { + pr_err("Failed to import over drive table!\n"); + return ret; + } + + /* retrieve updated gfxclk table */ + if (od8_settings->od_gfxclk_update) { + od8_settings->od_gfxclk_update = false; + single_dpm_table = &(dpm_table->gfx_table); + + if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_GFXCLK_BIT)) { + ret = vega20_set_single_dpm_table(smu, single_dpm_table, + PPCLK_GFXCLK); + if (ret) { + pr_err("[Setoverdrive] failed to refresh dpm table!\n"); + return ret; + } + } else { + single_dpm_table->count = 1; + single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.gfxclk / 100; + } + } + + break; + + default: + return -EINVAL; + } + + if (type == PP_OD_COMMIT_DPM_TABLE) { + ret = smu_handle_task(smu, smu_dpm->dpm_level, + AMD_PP_TASK_READJUST_POWER_STATE, + false); + } + + return ret; +} + +static int vega20_dpm_set_uvd_enable(struct smu_context *smu, bool enable) +{ + if (!smu_feature_is_supported(smu, SMU_FEATURE_DPM_UVD_BIT)) + return 0; + + if (enable == smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UVD_BIT)) + return 0; + + return smu_feature_set_enabled(smu, SMU_FEATURE_DPM_UVD_BIT, enable); +} + +static int vega20_dpm_set_vce_enable(struct smu_context *smu, bool enable) +{ + if (!smu_feature_is_supported(smu, SMU_FEATURE_DPM_VCE_BIT)) + return 0; + + if (enable == smu_feature_is_enabled(smu, SMU_FEATURE_DPM_VCE_BIT)) + return 0; + + return smu_feature_set_enabled(smu, SMU_FEATURE_DPM_VCE_BIT, enable); +} + +static bool vega20_is_dpm_running(struct smu_context *smu) +{ + int ret = 0; + uint32_t feature_mask[2]; + unsigned long feature_enabled; + ret = smu_feature_get_enabled_mask(smu, feature_mask, 2); + feature_enabled = (unsigned long)((uint64_t)feature_mask[0] | + ((uint64_t)feature_mask[1] << 32)); + return !!(feature_enabled & SMC_DPM_FEATURE); +} + +static int vega20_set_thermal_fan_table(struct smu_context *smu) +{ + int ret; + struct smu_table_context *table_context = &smu->smu_table; + PPTable_t *pptable = table_context->driver_pptable; + + ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetFanTemperatureTarget, + (uint32_t)pptable->FanTargetTemperature); + + return ret; +} + +static int vega20_get_fan_speed_rpm(struct smu_context *smu, + uint32_t *speed) +{ + int ret; + + ret = smu_send_smc_msg(smu, SMU_MSG_GetCurrentRpm); + + if (ret) { + pr_err("Attempt to get current RPM from SMC Failed!\n"); + return ret; + } + + smu_read_smc_arg(smu, speed); + + return 0; +} + +static int vega20_get_fan_speed_percent(struct smu_context *smu, + uint32_t *speed) +{ + int ret = 0; + uint32_t current_rpm = 0, percent = 0; + PPTable_t *pptable = smu->smu_table.driver_pptable; + + ret = vega20_get_fan_speed_rpm(smu, ¤t_rpm); + if (ret) + return ret; + + percent = current_rpm * 100 / pptable->FanMaximumRpm; + *speed = percent > 100 ? 100 : percent; + + return 0; +} + +static int vega20_get_gpu_power(struct smu_context *smu, uint32_t *value) +{ + uint32_t smu_version; + int ret = 0; + SmuMetrics_t metrics; + + if (!value) + return -EINVAL; + + ret = vega20_get_metrics_table(smu, &metrics); + if (ret) + return ret; + + ret = smu_get_smc_version(smu, NULL, &smu_version); + if (ret) + return ret; + + /* For the 40.46 release, they changed the value name */ + if (smu_version == 0x282e00) + *value = metrics.AverageSocketPower << 8; + else + *value = metrics.CurrSocketPower << 8; + + return 0; +} + +static int vega20_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 = vega20_get_metrics_table(smu, &metrics); + if (ret) + return ret; + + switch (sensor) { + case AMDGPU_PP_SENSOR_GPU_LOAD: + *value = metrics.AverageGfxActivity; + break; + case AMDGPU_PP_SENSOR_MEM_LOAD: + *value = metrics.AverageUclkActivity; + break; + default: + pr_err("Invalid sensor for retrieving clock activity\n"); + return -EINVAL; + } + + return 0; +} + +static int vega20_thermal_get_temperature(struct smu_context *smu, + enum amd_pp_sensors sensor, + uint32_t *value) +{ + struct amdgpu_device *adev = smu->adev; + SmuMetrics_t metrics; + uint32_t temp = 0; + int ret = 0; + + if (!value) + return -EINVAL; + + ret = vega20_get_metrics_table(smu, &metrics); + if (ret) + return ret; + + switch (sensor) { + case AMDGPU_PP_SENSOR_HOTSPOT_TEMP: + temp = RREG32_SOC15(THM, 0, mmCG_MULT_THERMAL_STATUS); + temp = (temp & CG_MULT_THERMAL_STATUS__CTF_TEMP_MASK) >> + CG_MULT_THERMAL_STATUS__CTF_TEMP__SHIFT; + + temp = temp & 0x1ff; + temp *= SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + + *value = temp; + break; + case AMDGPU_PP_SENSOR_EDGE_TEMP: + *value = metrics.TemperatureEdge * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + break; + case AMDGPU_PP_SENSOR_MEM_TEMP: + *value = metrics.TemperatureHBM * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + break; + default: + pr_err("Invalid sensor for retrieving temp\n"); + return -EINVAL; + } + + return 0; +} +static int vega20_read_sensor(struct smu_context *smu, + enum amd_pp_sensors sensor, + void *data, uint32_t *size) +{ + int ret = 0; + struct smu_table_context *table_context = &smu->smu_table; + PPTable_t *pptable = table_context->driver_pptable; + + if(!data || !size) + return -EINVAL; + + mutex_lock(&smu->sensor_lock); + switch (sensor) { + case AMDGPU_PP_SENSOR_MAX_FAN_RPM: + *(uint32_t *)data = pptable->FanMaximumRpm; + *size = 4; + break; + case AMDGPU_PP_SENSOR_MEM_LOAD: + case AMDGPU_PP_SENSOR_GPU_LOAD: + ret = vega20_get_current_activity_percent(smu, + sensor, + (uint32_t *)data); + *size = 4; + break; + case AMDGPU_PP_SENSOR_GPU_POWER: + ret = vega20_get_gpu_power(smu, (uint32_t *)data); + *size = 4; + break; + case AMDGPU_PP_SENSOR_HOTSPOT_TEMP: + case AMDGPU_PP_SENSOR_EDGE_TEMP: + case AMDGPU_PP_SENSOR_MEM_TEMP: + ret = vega20_thermal_get_temperature(smu, sensor, (uint32_t *)data); + *size = 4; + break; + default: + ret = smu_v11_0_read_sensor(smu, sensor, data, size); + } + mutex_unlock(&smu->sensor_lock); + + return ret; +} + +static int vega20_set_watermarks_table(struct smu_context *smu, + void *watermarks, struct + dm_pp_wm_sets_with_clock_ranges_soc15 + *clock_ranges) +{ + int i; + Watermarks_t *table = watermarks; + + if (!table || !clock_ranges) + return -EINVAL; + + if (clock_ranges->num_wm_dmif_sets > 4 || + clock_ranges->num_wm_mcif_sets > 4) + return -EINVAL; + + for (i = 0; i < clock_ranges->num_wm_dmif_sets; i++) { + table->WatermarkRow[1][i].MinClock = + cpu_to_le16((uint16_t) + (clock_ranges->wm_dmif_clocks_ranges[i].wm_min_dcfclk_clk_in_khz / + 1000)); + table->WatermarkRow[1][i].MaxClock = + cpu_to_le16((uint16_t) + (clock_ranges->wm_dmif_clocks_ranges[i].wm_max_dcfclk_clk_in_khz / + 1000)); + table->WatermarkRow[1][i].MinUclk = + cpu_to_le16((uint16_t) + (clock_ranges->wm_dmif_clocks_ranges[i].wm_min_mem_clk_in_khz / + 1000)); + table->WatermarkRow[1][i].MaxUclk = + cpu_to_le16((uint16_t) + (clock_ranges->wm_dmif_clocks_ranges[i].wm_max_mem_clk_in_khz / + 1000)); + table->WatermarkRow[1][i].WmSetting = (uint8_t) + clock_ranges->wm_dmif_clocks_ranges[i].wm_set_id; + } + + for (i = 0; i < clock_ranges->num_wm_mcif_sets; i++) { + table->WatermarkRow[0][i].MinClock = + cpu_to_le16((uint16_t) + (clock_ranges->wm_mcif_clocks_ranges[i].wm_min_socclk_clk_in_khz / + 1000)); + table->WatermarkRow[0][i].MaxClock = + cpu_to_le16((uint16_t) + (clock_ranges->wm_mcif_clocks_ranges[i].wm_max_socclk_clk_in_khz / + 1000)); + table->WatermarkRow[0][i].MinUclk = + cpu_to_le16((uint16_t) + (clock_ranges->wm_mcif_clocks_ranges[i].wm_min_mem_clk_in_khz / + 1000)); + table->WatermarkRow[0][i].MaxUclk = + cpu_to_le16((uint16_t) + (clock_ranges->wm_mcif_clocks_ranges[i].wm_max_mem_clk_in_khz / + 1000)); + table->WatermarkRow[0][i].WmSetting = (uint8_t) + clock_ranges->wm_mcif_clocks_ranges[i].wm_set_id; + } + + return 0; +} + +static int vega20_get_thermal_temperature_range(struct smu_context *smu, + struct smu_temperature_range *range) +{ + struct smu_table_context *table_context = &smu->smu_table; + ATOM_Vega20_POWERPLAYTABLE *powerplay_table = table_context->power_play_table; + PPTable_t *pptable = smu->smu_table.driver_pptable; + + if (!range || !powerplay_table) + return -EINVAL; + + range->max = powerplay_table->usSoftwareShutdownTemp * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + range->edge_emergency_max = (pptable->TedgeLimit + CTF_OFFSET_EDGE) * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + range->hotspot_crit_max = pptable->ThotspotLimit * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + range->hotspot_emergency_max = (pptable->ThotspotLimit + CTF_OFFSET_HOTSPOT) * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + range->mem_crit_max = pptable->ThbmLimit * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + range->mem_emergency_max = (pptable->ThbmLimit + CTF_OFFSET_HBM) * + SMU_TEMPERATURE_UNITS_PER_CENTIGRADES; + + + return 0; +} + +static int vega20_set_df_cstate(struct smu_context *smu, + enum pp_df_cstate state) +{ + uint32_t smu_version; + int ret; + + ret = smu_get_smc_version(smu, NULL, &smu_version); + if (ret) { + pr_err("Failed to get smu version!\n"); + return ret; + } + + /* PPSMC_MSG_DFCstateControl is supported with 40.50 and later fws */ + if (smu_version < 0x283200) { + pr_err("Df cstate control is supported with 40.50 and later SMC fw!\n"); + return -EINVAL; + } + + return smu_send_smc_msg_with_param(smu, SMU_MSG_DFCstateControl, state); +} + +static int vega20_update_pcie_parameters(struct smu_context *smu, + uint32_t pcie_gen_cap, + uint32_t pcie_width_cap) +{ + PPTable_t *pptable = smu->smu_table.driver_pptable; + int ret, i; + uint32_t smu_pcie_arg; + + for (i = 0; i < NUM_LINK_LEVELS; i++) { + smu_pcie_arg = (i << 16) | + ((pptable->PcieGenSpeed[i] <= pcie_gen_cap) ? (pptable->PcieGenSpeed[i] << 8) : + (pcie_gen_cap << 8)) | ((pptable->PcieLaneCount[i] <= pcie_width_cap) ? + pptable->PcieLaneCount[i] : pcie_width_cap); + ret = smu_send_smc_msg_with_param(smu, + SMU_MSG_OverridePcieParameters, + smu_pcie_arg); + } + + return ret; +} + + +static const struct pptable_funcs vega20_ppt_funcs = { + .tables_init = vega20_tables_init, + .alloc_dpm_context = vega20_allocate_dpm_context, + .store_powerplay_table = vega20_store_powerplay_table, + .check_powerplay_table = vega20_check_powerplay_table, + .append_powerplay_table = vega20_append_powerplay_table, + .get_smu_msg_index = vega20_get_smu_msg_index, + .get_smu_clk_index = vega20_get_smu_clk_index, + .get_smu_feature_index = vega20_get_smu_feature_index, + .get_smu_table_index = vega20_get_smu_table_index, + .get_smu_power_index = vega20_get_pwr_src_index, + .get_workload_type = vega20_get_workload_type, + .run_btc = vega20_run_btc_afll, + .get_allowed_feature_mask = vega20_get_allowed_feature_mask, + .get_current_power_state = vega20_get_current_power_state, + .set_default_dpm_table = vega20_set_default_dpm_table, + .set_power_state = NULL, + .populate_umd_state_clk = vega20_populate_umd_state_clk, + .print_clk_levels = vega20_print_clk_levels, + .force_clk_levels = vega20_force_clk_levels, + .get_clock_by_type_with_latency = vega20_get_clock_by_type_with_latency, + .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_od_percentage = vega20_set_od_percentage, + .set_default_od_settings = vega20_set_default_od_settings, + .od_edit_dpm_table = vega20_odn_edit_dpm_table, + .dpm_set_uvd_enable = vega20_dpm_set_uvd_enable, + .dpm_set_vce_enable = vega20_dpm_set_vce_enable, + .read_sensor = vega20_read_sensor, + .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, + .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, + .is_dpm_running = vega20_is_dpm_running, + .set_thermal_fan_table = vega20_set_thermal_fan_table, + .get_fan_speed_percent = vega20_get_fan_speed_percent, + .get_fan_speed_rpm = vega20_get_fan_speed_rpm, + .set_watermarks_table = vega20_set_watermarks_table, + .get_thermal_temperature_range = vega20_get_thermal_temperature_range, + .set_df_cstate = vega20_set_df_cstate, + .update_pcie_parameters = vega20_update_pcie_parameters, + .init_microcode = smu_v11_0_init_microcode, + .load_microcode = smu_v11_0_load_microcode, + .init_smc_tables = smu_v11_0_init_smc_tables, + .fini_smc_tables = smu_v11_0_fini_smc_tables, + .init_power = smu_v11_0_init_power, + .fini_power = smu_v11_0_fini_power, + .check_fw_status = smu_v11_0_check_fw_status, + .setup_pptable = smu_v11_0_setup_pptable, + .get_vbios_bootup_values = smu_v11_0_get_vbios_bootup_values, + .get_clk_info_from_vbios = smu_v11_0_get_clk_info_from_vbios, + .check_pptable = smu_v11_0_check_pptable, + .parse_pptable = smu_v11_0_parse_pptable, + .populate_smc_tables = smu_v11_0_populate_smc_pptable, + .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_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, + .send_smc_msg = smu_v11_0_send_msg, + .send_smc_msg_with_param = smu_v11_0_send_msg_with_param, + .read_smc_arg = smu_v11_0_read_arg, + .init_display_count = smu_v11_0_init_display_count, + .set_allowed_mask = smu_v11_0_set_allowed_mask, + .get_enabled_mask = smu_v11_0_get_enabled_mask, + .notify_display_change = smu_v11_0_notify_display_change, + .set_power_limit = smu_v11_0_set_power_limit, + .get_current_clk_freq = smu_v11_0_get_current_clk_freq, + .init_max_sustainable_clocks = smu_v11_0_init_max_sustainable_clocks, + .start_thermal_control = smu_v11_0_start_thermal_control, + .stop_thermal_control = smu_v11_0_stop_thermal_control, + .set_deep_sleep_dcefclk = smu_v11_0_set_deep_sleep_dcefclk, + .display_clock_voltage_request = smu_v11_0_display_clock_voltage_request, + .get_fan_control_mode = smu_v11_0_get_fan_control_mode, + .set_fan_control_mode = smu_v11_0_set_fan_control_mode, + .set_fan_speed_percent = smu_v11_0_set_fan_speed_percent, + .set_fan_speed_rpm = smu_v11_0_set_fan_speed_rpm, + .set_xgmi_pstate = smu_v11_0_set_xgmi_pstate, + .gfx_off_control = smu_v11_0_gfx_off_control, + .register_irq_handler = smu_v11_0_register_irq_handler, + .set_azalia_d3_pme = smu_v11_0_set_azalia_d3_pme, + .get_max_sustainable_clocks_by_dc = smu_v11_0_get_max_sustainable_clocks_by_dc, + .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, + .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, +}; + +void vega20_set_ppt_funcs(struct smu_context *smu) +{ + smu->ppt_funcs = &vega20_ppt_funcs; +} diff --git a/drivers/gpu/drm/amd/powerplay/vega20_ppt.h b/drivers/gpu/drm/amd/powerplay/vega20_ppt.h new file mode 100644 index 000000000000..2dc10e47b767 --- /dev/null +++ b/drivers/gpu/drm/amd/powerplay/vega20_ppt.h @@ -0,0 +1,179 @@ +/* + * Copyright 2019 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __VEGA20_PPT_H__ +#define __VEGA20_PPT_H__ + +#define VEGA20_UMD_PSTATE_GFXCLK_LEVEL 0x3 +#define VEGA20_UMD_PSTATE_SOCCLK_LEVEL 0x3 +#define VEGA20_UMD_PSTATE_MCLK_LEVEL 0x2 +#define VEGA20_UMD_PSTATE_UVDCLK_LEVEL 0x3 +#define VEGA20_UMD_PSTATE_VCEMCLK_LEVEL 0x3 + +#define MAX_REGULAR_DPM_NUMBER 16 +#define MAX_PCIE_CONF 2 + +#define VOLTAGE_SCALE 4 +#define AVFS_CURVE 0 +#define OD8_HOTCURVE_TEMPERATURE 85 + +#define SMU_FEATURES_LOW_MASK 0x00000000FFFFFFFF +#define SMU_FEATURES_LOW_SHIFT 0 +#define SMU_FEATURES_HIGH_MASK 0xFFFFFFFF00000000 +#define SMU_FEATURES_HIGH_SHIFT 32 + +enum { + GNLD_DPM_PREFETCHER = 0, + GNLD_DPM_GFXCLK, + GNLD_DPM_UCLK, + GNLD_DPM_SOCCLK, + GNLD_DPM_UVD, + GNLD_DPM_VCE, + GNLD_ULV, + GNLD_DPM_MP0CLK, + GNLD_DPM_LINK, + GNLD_DPM_DCEFCLK, + GNLD_DS_GFXCLK, + GNLD_DS_SOCCLK, + GNLD_DS_LCLK, + GNLD_PPT, + GNLD_TDC, + GNLD_THERMAL, + GNLD_GFX_PER_CU_CG, + GNLD_RM, + GNLD_DS_DCEFCLK, + GNLD_ACDC, + GNLD_VR0HOT, + GNLD_VR1HOT, + GNLD_FW_CTF, + GNLD_LED_DISPLAY, + GNLD_FAN_CONTROL, + GNLD_DIDT, + GNLD_GFXOFF, + GNLD_CG, + GNLD_DPM_FCLK, + GNLD_DS_FCLK, + GNLD_DS_MP1CLK, + GNLD_DS_MP0CLK, + GNLD_XGMI, + GNLD_ECC, + + GNLD_FEATURES_MAX +}; + +struct vega20_dpm_level { + bool enabled; + uint32_t value; + uint32_t param1; +}; + +struct vega20_dpm_state { + uint32_t soft_min_level; + uint32_t soft_max_level; + uint32_t hard_min_level; + uint32_t hard_max_level; +}; + +struct vega20_single_dpm_table { + uint32_t count; + struct vega20_dpm_state dpm_state; + struct vega20_dpm_level dpm_levels[MAX_REGULAR_DPM_NUMBER]; +}; + +struct vega20_pcie_table { + uint16_t count; + uint8_t pcie_gen[MAX_PCIE_CONF]; + uint8_t pcie_lane[MAX_PCIE_CONF]; + uint32_t lclk[MAX_PCIE_CONF]; +}; + +struct vega20_dpm_table { + struct vega20_single_dpm_table soc_table; + struct vega20_single_dpm_table gfx_table; + struct vega20_single_dpm_table mem_table; + struct vega20_single_dpm_table eclk_table; + struct vega20_single_dpm_table vclk_table; + struct vega20_single_dpm_table dclk_table; + struct vega20_single_dpm_table dcef_table; + struct vega20_single_dpm_table pixel_table; + struct vega20_single_dpm_table display_table; + struct vega20_single_dpm_table phy_table; + struct vega20_single_dpm_table fclk_table; + struct vega20_pcie_table pcie_table; +}; + +enum OD8_FEATURE_ID +{ + OD8_GFXCLK_LIMITS = 1 << 0, + OD8_GFXCLK_CURVE = 1 << 1, + OD8_UCLK_MAX = 1 << 2, + OD8_POWER_LIMIT = 1 << 3, + OD8_ACOUSTIC_LIMIT_SCLK = 1 << 4, //FanMaximumRpm + OD8_FAN_SPEED_MIN = 1 << 5, //FanMinimumPwm + OD8_TEMPERATURE_FAN = 1 << 6, //FanTargetTemperature + OD8_TEMPERATURE_SYSTEM = 1 << 7, //MaxOpTemp + OD8_MEMORY_TIMING_TUNE = 1 << 8, + OD8_FAN_ZERO_RPM_CONTROL = 1 << 9 +}; + +enum OD8_SETTING_ID +{ + OD8_SETTING_GFXCLK_FMIN = 0, + OD8_SETTING_GFXCLK_FMAX, + OD8_SETTING_GFXCLK_FREQ1, + OD8_SETTING_GFXCLK_VOLTAGE1, + OD8_SETTING_GFXCLK_FREQ2, + OD8_SETTING_GFXCLK_VOLTAGE2, + OD8_SETTING_GFXCLK_FREQ3, + OD8_SETTING_GFXCLK_VOLTAGE3, + OD8_SETTING_UCLK_FMAX, + OD8_SETTING_POWER_PERCENTAGE, + OD8_SETTING_FAN_ACOUSTIC_LIMIT, + OD8_SETTING_FAN_MIN_SPEED, + OD8_SETTING_FAN_TARGET_TEMP, + OD8_SETTING_OPERATING_TEMP_MAX, + OD8_SETTING_AC_TIMING, + OD8_SETTING_FAN_ZERO_RPM_CONTROL, + OD8_SETTING_COUNT +}; + +struct vega20_od8_single_setting { + uint32_t feature_id; + int32_t min_value; + int32_t max_value; + int32_t current_value; + int32_t default_value; +}; + +struct vega20_od8_settings { + struct vega20_od8_single_setting od8_settings_array[OD8_SETTING_COUNT]; + uint8_t *od_feature_capabilities; + uint32_t *od_settings_max; + uint32_t *od_settings_min; + void *od8_settings; + bool od_gfxclk_update; + bool od_memclk_update; +}; + +extern void vega20_set_ppt_funcs(struct smu_context *smu); + +#endif |