Handle case where PCIe information is not implemented in system

Also add a new error code for this.


[ROCm/rocm_smi_lib commit: 5e6424cab3]

projects/rocm-smi-lib/include/rocm_smi/rocm_smi.h

@@ -84,9 +84,14 @@ typedef enum {
   RSMI_STATUS_INTERNAL_EXCEPTION,        //!< An internal exception was caught
   RSMI_STATUS_INPUT_OUT_OF_BOUNDS,       //!< The provided input is out of
                                          //!< allowable or safe range
-  RSMI_INITIALIZATION_ERROR,             //!< An error occurred when rsmi
+  RSMI_STATUS_INIT_ERROR,                //!< An error occurred when rsmi
                                          //!< initializing internal data
                                          //!< structures
+  RSMI_INITIALIZATION_ERROR = RSMI_STATUS_INIT_ERROR,
+  RSMI_STATUS_NOT_YET_IMPLEMENTED,       //!< The requested function has not
+                                         //!< yet been implemented in the
+                                         //!< current system for the current
+                                         //!< devices
   RSMI_STATUS_UNKNOWN_ERROR = 0xFFFFFFFF,  //!< An unknown error occurred
 } rsmi_status_t;
 

projects/rocm-smi-lib/src/rocm_smi.cc

@@ -464,6 +464,10 @@ static rsmi_status_t get_frequencies(amd::smi::DevInfoTypes type,
   }
   assert(val_vec.size() <= RSMI_MAX_NUM_FREQUENCIES);
 
+  if (val_vec.size() == 0) {
+    return RSMI_STATUS_NOT_YET_IMPLEMENTED;
+  }
+
   f->num_supported = val_vec.size();
   bool current = false;
   f->current = RSMI_MAX_NUM_FREQUENCIES + 1;  // init to an invalid value
@@ -1092,6 +1096,16 @@ rsmi_status_string(rsmi_status_t status, const char **status_string) {
       *status_string = "The provided input is out of allowable or safe range";
       break;
 
+    case RSMI_STATUS_INIT_ERROR:
+      *status_string = "An error occurred during initialization, during "
+       "monitor discovery or when when initializing internal data structures";
+      break;
+
+    case RSMI_STATUS_NOT_YET_IMPLEMENTED:
+      *status_string = "The called function has not been implemented in this "
+        "system for this device type";
+      break;
+
     default:
       *status_string = "An unknown error occurred";
       return RSMI_STATUS_UNKNOWN_ERROR;

projects/rocm-smi-lib/tests/rocm_smi_test/functional/rsmi_sanity.cc

@@ -611,13 +611,17 @@ void TestSanity::Run(void) {
         print_frequencies(&f);
       }
       err = rsmi_dev_pci_bandwidth_get(i, &b);
-      CHK_ERR_ASRT(err)
-      IF_VERB(STANDARD) {
-        std::cout << "\t**Supported PCIe bandwidths: ";
-        std::cout << b.transfer_rate.num_supported << std::endl;
-        print_frequencies(&b.transfer_rate, b.lanes);
+      if (err == RSMI_STATUS_NOT_YET_IMPLEMENTED) {
+        std::cout << "\t**Get PCIE Bandwidth: Not supported on this machine"
+                                                              << std::endl;
+      } else {
+          CHK_ERR_ASRT(err)
+          IF_VERB(STANDARD) {
+            std::cout << "\t**Supported PCIe bandwidths: ";
+            std::cout << b.transfer_rate.num_supported << std::endl;
+            print_frequencies(&b.transfer_rate, b.lanes);
+          }
       }
-
       err = rsmi_dev_gpu_clk_freq_get(i, RSMI_CLK_TYPE_SYS, &f);
       CHK_ERR_ASRT(err)
       IF_VERB(STANDARD) {
@@ -766,8 +770,12 @@ void TestSanity::Run(void) {
       CHK_RSMI_PERM_ERR(err)
 
       err = test_set_pci_bw(i);
-      CHK_RSMI_PERM_ERR(err)
-
+      if (err == RSMI_STATUS_NOT_YET_IMPLEMENTED) {
+        std::cout << "\t**Set PCIE Bandwidth: Not supported on this machine"
+                                                              << std::endl;
+      } else {
+        CHK_RSMI_PERM_ERR(err)
+      }
       err = test_set_fan_speed(i);
       CHK_RSMI_PERM_ERR(err)