Merge amd-staging into amd-master 20230630

Signed-off-by: Hao Zhou <Hao.Zhou@amd.com>
Change-Id: I97ea2b65f4ac2ae87bbadf7d0f5b3408b44afa9a

amdsmi_cli/amdsmi_commands.py

@@ -298,7 +298,7 @@ class AMDSMICommands():
                     temp_vram_limit = amdsmi_interface.amdsmi_get_temp_metric(args.gpu,
                         amdsmi_interface.AmdSmiTemperatureType.VRAM, amdsmi_interface.AmdSmiTemperatureMetric.CRITICAL)
                 except amdsmi_exception.AmdSmiLibraryException as e:
-                    temp_junction_limit = e.get_error_info()
+                    temp_vram_limit = e.get_error_info()
                     if not self.all_arguments:
                         raise e
 

rocm_smi/src/rocm_smi_utils.cc

@@ -120,6 +120,9 @@ int WriteSysfsStr(std::string path, std::string val) {
 
   fs << val;
   fs.close();
+  if (!fs) {
+    return ENOENT;  // Map to NOT_SUPPORT if errors
+  }
   return ret;
 }
 

tests/amd_smi_test/functional/fan_read_write.cc

@@ -137,6 +137,12 @@ void TestFanReadWrite::Run(void) {
     }
 
     ret = amdsmi_set_gpu_fan_speed(processor_handles_[dv_ind], 0, new_speed);
+
+    // When you can read fan speed, it is not always can set fan speed.
+    if (ret == AMDSMI_STATUS_NOT_SUPPORTED) {
+      std::cout << "***System fan set is not supported." << std::endl;
+      continue;
+    }
     CHK_ERR_ASRT(ret)
 
     sleep(4);