drm/amd/pp: Add auto power profilng switch based on workloads (v2)

Add power profiling mode dynamic switch based on the workloads.
Currently, support Cumpute, VR, Video, 3D,power saving with Cumpute
have highest prority, power saving have lowest prority.

in manual dpm mode, driver will stop auto switch, just save the client's
requests. user can set power profiling mode through sysfs.

when exit manual dpm mode, driver will response the client's requests.
switch based on the client's prority.

v2: squash in fixes from Rex

Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

9 files changed, 66 insertions, 37 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h
index 9c373f8f465c..643d008410c6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h
@@ -341,9 +341,9 @@ enum amdgpu_pcie_gen {
                 ((adev)->powerplay.pp_funcs->reset_power_profile_state(\
                         (adev)->powerplay.pp_handle, request))
 
-#define amdgpu_dpm_switch_power_profile(adev, type) \
+#define amdgpu_dpm_switch_power_profile(adev, type, en) \
                 ((adev)->powerplay.pp_funcs->switch_power_profile(\
-                        (adev)->powerplay.pp_handle, type))
+                        (adev)->powerplay.pp_handle, type, en))
 
 #define amdgpu_dpm_set_clockgating_by_smu(adev, msg_id) \
                 ((adev)->powerplay.pp_funcs->set_clockgating_by_smu(\
diff --git a/drivers/gpu/drm/amd/include/kgd_pp_interface.h b/drivers/gpu/drm/amd/include/kgd_pp_interface.h
index e482daf394d3..15bd0f9acf73 100644
--- a/drivers/gpu/drm/amd/include/kgd_pp_interface.h
+++ b/drivers/gpu/drm/amd/include/kgd_pp_interface.h
@@ -83,20 +83,6 @@ enum amd_vce_level {
         AMD_VCE_LEVEL_DC_GP_HIGH = 5, /* DC, general purpose queue, 1080 >= res > 720 */
 };
 
-enum amd_pp_profile_type {
-        AMD_PP_GFX_PROFILE,
-        AMD_PP_COMPUTE_PROFILE,
-};
-
-struct amd_pp_profile {
-        enum amd_pp_profile_type type;
-        uint32_t min_sclk;
-        uint32_t min_mclk;
-        uint16_t activity_threshold;
-        uint8_t up_hyst;
-        uint8_t down_hyst;
-};
-
 enum amd_fan_ctrl_mode {
         AMD_FAN_CTRL_NONE = 0,
         AMD_FAN_CTRL_MANUAL = 1,
@@ -143,7 +129,6 @@ enum PP_SMC_POWER_PROFILE {
         PP_SMC_POWER_PROFILE_VR           = 0x3,
         PP_SMC_POWER_PROFILE_COMPUTE      = 0x4,
         PP_SMC_POWER_PROFILE_CUSTOM       = 0x5,
-        PP_SMC_POWER_PROFILE_AUTO         = 0x6,
 };
 
 enum {
@@ -252,8 +237,7 @@ struct amd_pm_funcs {
         int (*get_pp_table)(void *handle, char **table);
         int (*set_pp_table)(void *handle, const char *buf, size_t size);
         void (*debugfs_print_current_performance_level)(void *handle, struct seq_file *m);
-        int (*switch_power_profile)(void *handle,
-                        enum amd_pp_profile_type type);
+        int (*switch_power_profile)(void *handle, enum PP_SMC_POWER_PROFILE type, bool en);
 /* export to amdgpu */
         void (*powergate_uvd)(void *handle, bool gate);
         void (*powergate_vce)(void *handle, bool gate);
diff --git a/drivers/gpu/drm/amd/powerplay/amd_powerplay.c b/drivers/gpu/drm/amd/powerplay/amd_powerplay.c
index a8705b5ad264..b989bf3542d6 100644
--- a/drivers/gpu/drm/amd/powerplay/amd_powerplay.c
+++ b/drivers/gpu/drm/amd/powerplay/amd_powerplay.c
@@ -1077,22 +1077,44 @@ static int pp_odn_edit_dpm_table(void *handle, uint32_t type, long *input, uint3
 }
 
 static int pp_dpm_switch_power_profile(void *handle,
-                enum amd_pp_profile_type type)
+                enum PP_SMC_POWER_PROFILE type, bool en)
 {
         struct pp_hwmgr *hwmgr;
-        struct amd_pp_profile request = {0};
         struct pp_instance *pp_handle = (struct pp_instance *)handle;
+        long workload;
+        uint32_t index;
 
         if (pp_check(pp_handle))
                 return -EINVAL;
 
         hwmgr = pp_handle->hwmgr;
 
-        if (hwmgr->current_power_profile != type) {
-                request.type = type;
-                pp_dpm_set_power_profile_state(handle, &request);
+        if (hwmgr->hwmgr_func->set_power_profile_mode == NULL) {
+                pr_info("%s was not implemented.\n", __func__);
+                return -EINVAL;
+        }
+
+        if (!(type < PP_SMC_POWER_PROFILE_CUSTOM))
+                return -EINVAL;
+
+        mutex_lock(&pp_handle->pp_lock);
+
+        if (!en) {
+                hwmgr->workload_mask &= ~(1 << hwmgr->workload_prority[type]);
+                index = fls(hwmgr->workload_mask);
+                index = index > 0 && index <= Workload_Policy_Max ? index - 1 : 0;
+                workload = hwmgr->workload_setting[index];
+        } else {
+                hwmgr->workload_mask |= (1 << hwmgr->workload_prority[type]);
+                index = fls(hwmgr->workload_mask);
+                index = index <= Workload_Policy_Max ? index - 1 : 0;
+                workload = hwmgr->workload_setting[index];
         }
 
+        if (hwmgr->dpm_level != AMD_DPM_FORCED_LEVEL_MANUAL)
+                hwmgr->hwmgr_func->set_power_profile_mode(hwmgr, &workload, 0);
+        mutex_unlock(&pp_handle->pp_lock);
+
         return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c
index aea107643859..4b6cf795243c 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c
@@ -125,6 +125,21 @@ static const struct cgs_irq_src_funcs thermal_irq_src[3] = {
         { .handler = phm_ctf_irq }
 };
 
+static void hwmgr_init_workload_prority(struct pp_hwmgr *hwmgr)
+{
+        hwmgr->workload_prority[PP_SMC_POWER_PROFILE_FULLSCREEN3D] = 2;
+        hwmgr->workload_prority[PP_SMC_POWER_PROFILE_POWERSAVING] = 0;
+        hwmgr->workload_prority[PP_SMC_POWER_PROFILE_VIDEO] = 1;
+        hwmgr->workload_prority[PP_SMC_POWER_PROFILE_VR] = 3;
+        hwmgr->workload_prority[PP_SMC_POWER_PROFILE_COMPUTE] = 4;
+
+        hwmgr->workload_setting[0] = PP_SMC_POWER_PROFILE_POWERSAVING;
+        hwmgr->workload_setting[1] = PP_SMC_POWER_PROFILE_VIDEO;
+        hwmgr->workload_setting[2] = PP_SMC_POWER_PROFILE_FULLSCREEN3D;
+        hwmgr->workload_setting[3] = PP_SMC_POWER_PROFILE_VR;
+        hwmgr->workload_setting[4] = PP_SMC_POWER_PROFILE_COMPUTE;
+}
+
 int hwmgr_early_init(struct pp_instance *handle)
 {
         struct pp_hwmgr *hwmgr;
@@ -151,6 +166,7 @@ int hwmgr_early_init(struct pp_instance *handle)
         hwmgr_set_user_specify_caps(hwmgr);
         hwmgr->fan_ctrl_is_in_default_mode = true;
         hwmgr->reload_fw = 1;
+        hwmgr_init_workload_prority(hwmgr);
 
         switch (hwmgr->chip_family) {
         case AMDGPU_FAMILY_CI:
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/pp_psm.c b/drivers/gpu/drm/amd/powerplay/hwmgr/pp_psm.c
index ed3bd1502c0d..d0ef8f9c1361 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/pp_psm.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/pp_psm.c
@@ -220,6 +220,8 @@ int psm_adjust_power_state_dynamic(struct pp_hwmgr *hwmgr, bool skip,
         struct pp_power_state *pcurrent;
         struct pp_power_state *requested;
         bool equal;
+        uint32_t index;
+        long workload;
 
         if (skip)
                 return 0;
@@ -247,6 +249,15 @@ int psm_adjust_power_state_dynamic(struct pp_hwmgr *hwmgr, bool skip,
         if (!phm_force_dpm_levels(hwmgr, hwmgr->request_dpm_level))
                 hwmgr->dpm_level = hwmgr->request_dpm_level;
 
+        if (hwmgr->dpm_level != AMD_DPM_FORCED_LEVEL_MANUAL) {
+                index = fls(hwmgr->workload_mask);
+                index = index > 0 && index <= Workload_Policy_Max ? index - 1 : 0;
+                workload = hwmgr->workload_setting[index];
+
+                if (hwmgr->power_profile_mode != workload && hwmgr->hwmgr_func->set_power_profile_mode)
+                        hwmgr->hwmgr_func->set_power_profile_mode(hwmgr, &workload, 0);
+        }
+
         return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c
index 58e2e0a10b3d..d8c205cf8f12 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c
@@ -1510,6 +1510,9 @@ static void smu7_init_dpm_defaults(struct pp_hwmgr *hwmgr)
         data->current_profile_setting.mclk_up_hyst = 0;
         data->current_profile_setting.mclk_down_hyst = 100;
         data->current_profile_setting.mclk_activity = SMU7_MCLK_TARGETACTIVITY_DFLT;
+        hwmgr->workload_mask = 1 << hwmgr->workload_prority[PP_SMC_POWER_PROFILE_FULLSCREEN3D];
+        hwmgr->power_profile_mode = PP_SMC_POWER_PROFILE_FULLSCREEN3D;
+        hwmgr->default_power_profile_mode = PP_SMC_POWER_PROFILE_FULLSCREEN3D;
 
         if (hwmgr->chip_id == CHIP_POLARIS12 || hwmgr->is_kicker) {
                 uint8_t tmp1, tmp2;
@@ -2466,9 +2469,6 @@ static int smu7_hwmgr_backend_init(struct pp_hwmgr *hwmgr)
         smu7_patch_voltage_workaround(hwmgr);
         smu7_init_dpm_defaults(hwmgr);
 
-        hwmgr->power_profile_mode = PP_SMC_POWER_PROFILE_FULLSCREEN3D;
-        hwmgr->default_power_profile_mode = PP_SMC_POWER_PROFILE_FULLSCREEN3D;
-
         /* Get leakage voltage based on leakage ID. */
         if (phm_cap_enabled(hwmgr->platform_descriptor.platformCaps,
                         PHM_PlatformCaps_EVV)) {
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c
index d90a0f1dbb55..f23861f2c685 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c
@@ -766,13 +766,12 @@ static int vega10_hwmgr_backend_init(struct pp_hwmgr *hwmgr)
 
         hwmgr->backend = data;
 
+        hwmgr->workload_mask = 1 << hwmgr->workload_prority[PP_SMC_POWER_PROFILE_VIDEO];
         hwmgr->power_profile_mode = PP_SMC_POWER_PROFILE_VIDEO;
         hwmgr->default_power_profile_mode = PP_SMC_POWER_PROFILE_VIDEO;
 
         vega10_set_default_registry_data(hwmgr);
-
         data->disable_dpm_mask = 0xff;
-        data->workload_mask = 0xff;
 
         /* need to set voltage control types before EVV patching */
         data->vddc_control = VEGA10_VOLTAGE_CONTROL_NONE;
@@ -4187,11 +4186,6 @@ static int vega10_dpm_force_dpm_level(struct pp_hwmgr *hwmgr,
                 break;
         case AMD_DPM_FORCED_LEVEL_AUTO:
                 ret = vega10_unforce_dpm_levels(hwmgr);
-                if (hwmgr->default_power_profile_mode != hwmgr->power_profile_mode) {
-                                smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetWorkloadMask,
-                                                1 << hwmgr->default_power_profile_mode);
-                                hwmgr->power_profile_mode = hwmgr->default_power_profile_mode;
-                }
                 break;
         case AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD:
         case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK:
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.h b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.h
index de3219fff2db..8f6c2cb962da 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.h
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.h
@@ -374,9 +374,6 @@ struct vega10_hwmgr {
         /* ---- Overdrive next setting ---- */
         uint32_t                       apply_overdrive_next_settings_mask;
 
-        /* ---- Workload Mask ---- */
-        uint32_t                       workload_mask;
-
         /* ---- SMU9 ---- */
         struct smu_features            smu_features[GNLD_FEATURES_MAX];
         struct vega10_smc_state_table  smc_state_table;
diff --git a/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h b/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h
index 65224dbb550d..94df6eed5384 100644
--- a/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h
+++ b/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h
@@ -696,6 +696,8 @@ enum PP_TABLE_VERSION {
 /**
  * The main hardware manager structure.
  */
+#define Workload_Policy_Max 5
+
 struct pp_hwmgr {
         void *adev;
         uint32_t chip_family;
@@ -757,6 +759,9 @@ struct pp_hwmgr {
         bool od_enabled;
         uint32_t power_limit;
         uint32_t default_power_limit;
+        uint32_t workload_mask;
+        uint32_t workload_prority[Workload_Policy_Max];
+        uint32_t workload_setting[Workload_Policy_Max];
 };
 
 struct cgs_irq_src_funcs {