Merge amd-staging into amd-master 20230710

Signed-off-by: Hao Zhou <Hao.Zhou@amd.com>
Change-Id: I213f4f064b94187a9d157e51fb09c3051a26a449

include/amd_smi/amdsmi.h

@@ -3390,16 +3390,21 @@ amdsmi_get_gpu_vram_usage(amdsmi_processor_handle processor_handle, amdsmi_vram_
 /**
  *  @brief          Returns the list of processes running on a given GPU including itself.
  *
- *  @note           The user provides a buffer to store the list and the
+ *  @note           The user provides a buffer to store the list and the maximum
- *                  maximum number of processes that can be returned. If the user
+ *                  number of processes that can be returned. If the user sets
- *                  sets max_processes to 0, the total number of processes will be
+ *                  max_processes to 0, the current total number of processes will
- *                  returned.
+ *                  replace max_processes param. After that, the function needs to be
+ *                  called again, with updated max_processes, to successfully fill the
+ *                  process list, which was previously allocated with max_processes
  *
  *  @param[in]      processor_handle Device which to query
  *
  *  @param[in,out]  max_processes Reference to the size of the list buffer in
  *                  number of elements. Returns the return number of elements
- *                  in list or the number of running processes if equal to 0.
+ *                  in list or the number of running processes if equal to 0,
+ *                  and if given value in param max_processes is less than
+ *                  number of processes currently running,
+ *                  AMDSMI_STATUS_OUT_OF_RESOURCES will be returned.
  *
  *  @param[out]     list Reference to a user-provided buffer where the process
  *                  list will be returned. This buffer must contain at least

py-interface/amdsmi_interface.py

@@ -28,6 +28,8 @@ from collections.abc import Iterable
 from . import amdsmi_wrapper
 from .amdsmi_exception import *
 
+MAX_NUM_PROCESSES = 1024
+
 class AmdSmiInitFlags(IntEnum):
     INIT_ALL_PROCESSORS = amdsmi_wrapper.AMDSMI_INIT_ALL_PROCESSORS
     INIT_AMD_CPUS = amdsmi_wrapper.AMDSMI_INIT_AMD_CPUS
@@ -845,14 +847,7 @@ def amdsmi_get_gpu_process_list(
             processor_handle, amdsmi_wrapper.amdsmi_processor_handle
         )
 
-    max_processes = ctypes.c_uint32(0)
+    max_processes = ctypes.c_uint32(MAX_NUM_PROCESSES)
-    process_list = (amdsmi_wrapper.amdsmi_process_handle_t *
-                    max_processes.value)()
-    _check_res(
-        amdsmi_wrapper.amdsmi_get_gpu_process_list(
-            processor_handle, ctypes.byref(max_processes), process_list 
-        )
-    )
 
     process_list = (amdsmi_wrapper.amdsmi_process_handle_t *
                     max_processes.value)()
@@ -862,7 +857,8 @@ def amdsmi_get_gpu_process_list(
         )
     )
 
-    return [amdsmi_wrapper.amdsmi_process_handle_t(x) for x in list(process_list)]
+    return [amdsmi_wrapper.amdsmi_process_handle_t(process_list[x])\
+    for x in range(0, max_processes.value)]
 
 
 def amdsmi_get_gpu_process_info(

rocm_smi/src/rocm_smi_monitor.cc

@@ -367,17 +367,19 @@ Monitor::setVoltSensorLabelMap(void) {
   }
   auto add_volt_sensor_entry = [&](uint32_t file_index) {
     ret = readMonitor(kMonVoltLabel, file_index, &type_str);
-    rsmi_voltage_type_t t_type = kVoltSensorNameMap.at(type_str);
+    rsmi_voltage_type_t t_type;
+
     // If readMonitor fails, there is no label file for the file_index.
     // In that case, map the type to file index 0, which is not supported
     // and will fail appropriately later when we check for support.
     if (ret) {
-      volt_type_index_map_.insert({t_type, 0});
       index_volt_type_map_.insert({file_index, RSMI_VOLT_TYPE_INVALID});
     } else {
-      volt_type_index_map_.insert({t_type, file_index});
+      t_type = kVoltSensorNameMap.at(type_str);
+      volt_type_index_map_[t_type] = file_index;
       index_volt_type_map_.insert({file_index, t_type});
     }
+
     return 0;
   };
 
@@ -463,6 +465,8 @@ Monitor::getVoltSensorIndex(rsmi_voltage_type_t type) {
 
 rsmi_voltage_type_t
 Monitor::getVoltSensorEnum(uint64_t ind) {
+  // check if ind is a key or not
+  if (index_volt_type_map_.count(ind) == 0) return RSMI_VOLT_TYPE_INVALID;
   return index_volt_type_map_.at(ind);
 }