[SWDEV-495169] Update ROCm SMI CLI and Error handling (#3)

Issues include:

Update ROCm SMI displaying None or Not Supported to N/A
Update ROCm SMI displaying err msg to instead log err

Signed-off-by: Juan Castillo juan.castillo@amd.com
Change-Id: I1a2ce6e4f329666b5666664a7d7b4475d6c1cbc7

Author: dmitrii-galantsev

python_smi_tools/rocm_smi.py

@@ -1980,22 +1980,28 @@ def showAllConcise(deviceList):
                              + getComputePartition(device, silent)
                              + ", " + getPartitionId(device, silent))
         sclk = showCurrentClocks([device], 'sclk', concise=silent)
+        if not sclk:
+            sclk = 'N/A'
         mclk = showCurrentClocks([device], 'mclk', concise=silent)
+        if not mclk:
+            mclk = 'N/A'
         (retCode, fanLevel, fanSpeed) = getFanSpeed(device, silent)
         fan = str(fanSpeed) + '%'
         if getPerfLevel(device, silent) != -1:
             perf = getPerfLevel(device, silent)
         else:
-            perf = 'Unsupported'
+            perf = 'N/A'
         if getMaxPower(device, silent) != -1:
             pwrCap = str(getMaxPower(device, silent)) + 'W'
         else:
-            pwrCap = 'Unsupported'
+            pwrCap = 'N/A'
         if getGpuUse(device, silent) != -1:
             gpu_busy = str(getGpuUse(device, silent)) + '%'
         else:
-            gpu_busy = 'Unsupported'
+            gpu_busy = 'N/A'
         allocated_mem_percent = getAllocatedMemoryPercent(device)
+        if allocated_mem_percent['ret'] != rsmi_status_t.RSMI_STATUS_SUCCESS:
+            allocated_mem_percent['combined'] = 'N/A'
 
         # Top Row - per device data
         values['card%s' % (str(device))] = [device, getNodeId(device),

src/rocm_smi.cc

@@ -368,8 +368,9 @@ static rsmi_status_t get_dev_mon_value(amd::smi::MonitorTypes type,
   }
 
   if (!amd::smi::IsInteger(val_str)) {
-    std::cerr << "Expected integer value from monitor,"
-                                " but got \"" << val_str << "\"" << std::endl;
+    std::ostringstream ss;
+    ss << "Expected integer value from monitor, but got \"" << val_str << "\"";
+    LOG_ERROR(ss);
     return RSMI_STATUS_UNEXPECTED_DATA;
   }
 
@@ -397,8 +398,9 @@ static rsmi_status_t get_dev_mon_value(amd::smi::MonitorTypes type,
   }
 
   if (!amd::smi::IsInteger(val_str)) {
-    std::cerr << "Expected integer value from monitor,"
-                                " but got \"" << val_str << "\"" << std::endl;
+    std::ostringstream ss;
+    ss << "Expected integer value from monitor, but got \"" << val_str << "\"";
+    LOG_ERROR(ss);
     return RSMI_STATUS_UNEXPECTED_DATA;
   }