ROCm SMI LIB: Add Default GPU Power Cap

Implement default GPU power cap functionality in the LIB.

Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
Change-Id: Ia6b3420beb0e4df5559c3e6d11d0667972590b53

include/rocm_smi/rocm_smi.h

@@ -1653,6 +1653,30 @@ rsmi_dev_energy_count_get(uint32_t dv_ind, uint64_t *power,
 rsmi_status_t
 rsmi_dev_power_cap_get(uint32_t dv_ind, uint32_t sensor_ind, uint64_t *cap);
 
+/**
+ *  @brief Get the default power cap for the device specified by @p dv_ind.
+ *
+ *  @details The maximum power cap be temporarily changed by the user. However,
+ *  this function always returns the default reset power cap. The power level
+ *  returned through @p power will be in microWatts.
+ *
+ *  @param[in] dv_ind a device index
+ *
+ *  @param[inout] default_cap a pointer to a uint64_t that indicates the default
+ *  power cap, in microwatts
+ *  If this parameter is nullptr, this function will return
+ *  ::RSMI_STATUS_INVALID_ARGS if the function is supported with the provided,
+ *  arguments and ::RSMI_STATUS_NOT_SUPPORTED if it is not supported with the
+ *  provided arguments.
+ *
+ *  @retval ::RSMI_STATUS_SUCCESS call was successful
+ *  @retval ::RSMI_STATUS_NOT_SUPPORTED installed software or hardware does not
+ *  support this function with the given arguments
+ *  @retval ::RSMI_STATUS_INVALID_ARGS the provided arguments are not valid
+ */
+rsmi_status_t
+rsmi_dev_power_cap_default_get(uint32_t dv_ind, uint64_t *default_cap);
+
 /**
  *  @brief Get the range of valid values for the power cap
  *

include/rocm_smi/rocm_smi_monitor.h

@@ -63,6 +63,7 @@ enum MonitorTypes {
   kMonFanRPMs,
   kMonFanCntrlEnable,
   kMonPowerCap,
+  kMonPowerCapDefault,
   kMonPowerCapMax,
   kMonPowerCapMin,
   kMonPowerAve,

src/rocm_smi.cc

@@ -2632,6 +2632,22 @@ rsmi_dev_energy_count_get(uint32_t dv_ind, uint64_t *power,
   CATCH
 }
 
+rsmi_status_t
+rsmi_dev_power_cap_default_get(uint32_t dv_ind, uint64_t *default_cap) {
+  TRY
+
+  uint32_t sensor_ind = 1; // power sysfs files have 1-based indices
+  CHK_SUPPORT_SUBVAR_ONLY(default_cap, sensor_ind)
+
+  rsmi_status_t ret;
+
+  DEVICE_MUTEX
+  ret = get_dev_mon_value(amd::smi::kMonPowerCapDefault, dv_ind, sensor_ind, default_cap);
+
+  return ret;
+  CATCH
+}
+
 rsmi_status_t
 rsmi_dev_power_cap_get(uint32_t dv_ind, uint32_t sensor_ind, uint64_t *cap) {
   TRY

src/rocm_smi_monitor.cc

@@ -73,6 +73,7 @@ static const char *kMonMaxFanSpeedFName = "pwm#_max";
 static const char *kMonFanRPMsName = "fan#_input";
 static const char *kMonFanControlEnableName = "pwm#_enable";
 static const char *kMonNameFName = "name";
+static const char *kMonPowerCapDefaultName = "power#_cap_default";
 static const char *kMonPowerCapName = "power#_cap";
 static const char *kMonPowerCapMaxName = "power#_cap_max";
 static const char *kMonPowerCapMinName = "power#_cap_min";
@@ -128,6 +129,7 @@ static const std::map<MonitorTypes, const char *> kMonitorNameMap = {
     {kMonMaxFanSpeed, kMonMaxFanSpeedFName},
     {kMonFanRPMs, kMonFanRPMsName},
     {kMonPowerCap, kMonPowerCapName},
+    {kMonPowerCapDefault, kMonPowerCapDefaultName},
     {kMonPowerCapMax, kMonPowerCapMaxName},
     {kMonPowerCapMin, kMonPowerCapMinName},
     {kMonPowerAve, kMonPowerAveName},
@@ -198,6 +200,10 @@ static const std::map<const char *, monitor_depends_t> kMonFuncDependsMap = {
                                       .variants = {kMonInvalid},
                                     }
   },
+  {"rsmi_dev_power_cap_default_get",        { .mandatory_depends = {kMonPowerCapDefaultName},
+                                      .variants = {kMonInvalid},
+                                    }
+  },
   {"rsmi_dev_power_cap_range_get",  { .mandatory_depends =
                                                         {kMonPowerCapMaxName,
                                                          kMonPowerCapMinName},