Check FRU-based product information if available

WKS and server cards have an FRU with product information, so try to use
that for product name and product SKU if it exists.

Signed-off-by: Kent Russell <kent.russell@amd.com>
Change-Id: I40bbd3bf62f4cb02e96015ed1630112691cacbc3


[ROCm/rocm_smi_lib commit: df7c3434cd]

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

@@ -979,6 +979,33 @@ rsmi_status_t rsmi_num_monitor_devices(uint32_t *num_devices);
  */
 rsmi_status_t rsmi_dev_id_get(uint32_t dv_ind, uint16_t *id);
 
+
+/**
+ *  @brief Get the SKU for a desired device associated with the device with
+ *  provided device index.
+ *
+ *  @details Given a device index @p dv_ind and a pointer to a char @p sku,
+ *  this function will attempt to obtain the SKU from the Product Information
+ *  FRU chip, present on server ASICs. It will write the sku value to the
+ *  char array pointed to by @p sku.
+ *
+ *  @param[in] dv_ind a device index
+ *
+ *  @param[inout] sku a pointer to char to which the sku will be written
+ *
+ *  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_sku_get(uint32_t dv_ind, char *sku);
+
 /**
  *  @brief Get the device vendor id associated with the device with provided
  *  device index.

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

@@ -99,6 +99,8 @@ enum DevInfoTypes {
   kDevPerfLevel,
   kDevOverDriveLevel,
   kDevDevID,
+  kDevDevProdName,
+  kDevDevProdNum,
   kDevVendorID,
   kDevSubSysDevID,
   kDevSubSysVendorID,

projects/rocm-smi-lib/python_smi_tools/rocm_smi.py

@@ -1621,13 +1621,15 @@ def showProductName(deviceList):
                 device_model = model.value.decode()
                 printLog(device, 'Card model', '\t\t' + device_model)
             printLog(device, 'Card vendor', '\t\t' + device_vendor)
-            # Retrieve the device SKU from VBIOS
+            # Retrieve the SKU
-            ret = rocmsmi.rsmi_dev_vbios_version_get(device, vbios, 256)
+            ret = rocmsmi.rsmi_dev_sku_get(device, device_sku, 256)
-            if rsmi_ret_ok(ret, device) and vbios.value.decode():
+            if not rsmi_ret_ok(ret, device) or not device_sku.value.decode():
-                # Device SKU is just 6 characters after the first occurance of '-' in vbios_version
+                # Retrieve the device SKU from VBIOS if product_number doesn't exist
-                # TODO: Use Product Info sysfs where possible
+                ret = rocmsmi.rsmi_dev_vbios_version_get(device, vbios, 256)
-                device_sku = vbios.value.decode().split('-')[1][:6]
+                if rsmi_ret_ok(ret, device) and vbios.value.decode():
-                printLog(device, 'Card SKU', '\t\t' + device_sku)
+                    # Device SKU is just 6 characters after the first occurance of '-' in vbios_version
+                    device_sku = vbios.value.decode().split('-')[1][:6]
+            printLog(device, 'Card SKU', '\t\t' + device_sku)
         else:
             printLog(device, 'Incompatible device.\n' \
                              'GPU[%s]\t\t: Expected vendor name: Advanced Micro Devices, Inc. [AMD/ATI]\n'\

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

@@ -722,6 +722,15 @@ rsmi_dev_id_get(uint32_t dv_ind, uint16_t *id) {
   return get_id(dv_ind, amd::smi::kDevDevID, id);
 }
 
+rsmi_status_t
+rsmi_dev_sku_get(uint32_t dv_ind, uint16_t *id) {
+  TRY
+  CHK_SUPPORT_NAME_ONLY(id)
+  DEVICE_MUTEX
+  return get_id(dv_ind, amd::smi::kDevDevProdNum, id);
+  CATCH
+}
+
 rsmi_status_t
 rsmi_dev_subsystem_id_get(uint32_t dv_ind, uint16_t *id) {
   CHK_SUPPORT_NAME_ONLY(id)
@@ -1530,6 +1539,25 @@ static rsmi_status_t get_backup_name(uint16_t id, char *name, size_t len) {
   return RSMI_STATUS_SUCCESS;
 }
 
+static rsmi_status_t get_dev_name_from_file(uint32_t dv_ind, char *name,
+                                               size_t len) {
+  std::string val_str;
+  rsmi_status_t ret = get_dev_value_line(amd::smi::kDevDevProdName, dv_ind, &val_str);
+
+  if (ret != 0) {
+    return amd::smi::ErrnoToRsmiStatus(ret);
+  }
+  size_t ct = val_str.copy(name, len);
+
+  name[std::min(len - 1, ct)] = '\0';
+
+  if (len < (val_str.size() + 1)) {
+    return RSMI_STATUS_INSUFFICIENT_SIZE;
+  }
+
+  return RSMI_STATUS_SUCCESS;
+}
+
 // Parse pci.ids files. Comment lines have # in first column. Otherwise,
 // Syntax:
 // vendor  vendor_name
@@ -1687,7 +1715,11 @@ rsmi_dev_name_get(uint32_t dv_ind, char *name, size_t len) {
 
   DEVICE_MUTEX
 
-  ret = get_dev_name_from_id(dv_ind, name, len, NAME_STR_DEVICE);
+  ret = get_dev_name_from_file(dv_ind, name, len);
+
+  if (ret) {
+    ret = get_dev_name_from_id(dv_ind, name, len, NAME_STR_DEVICE);
+  }
 
   return ret;
   CATCH

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

@@ -72,6 +72,8 @@ namespace smi {
 
 // Device sysfs file names
 static const char *kDevPerfLevelFName = "power_dpm_force_performance_level";
+static const char *kDevDevProdNameFName = "product_name";
+static const char *kDevDevProdNumFName = "product_number";
 static const char *kDevDevIDFName = "device";
 static const char *kDevVendorIDFName = "vendor";
 static const char *kDevSubSysDevIDFName = "subsystem_device";
@@ -212,6 +214,8 @@ static const char *kDevPerfLevelUnknownStr = "unknown";
 static const std::map<DevInfoTypes, const char *> kDevAttribNameMap = {
     {kDevPerfLevel, kDevPerfLevelFName},
     {kDevOverDriveLevel, kDevOverDriveLevelFName},
+    {kDevDevProdName, kDevDevProdNameFName},
+    {kDevDevProdNum, kDevDevProdNumFName},
     {kDevDevID, kDevDevIDFName},
     {kDevVendorID, kDevVendorIDFName},
     {kDevSubSysDevID, kDevSubSysDevIDFName},
@@ -338,7 +342,8 @@ static const std::map<const char *, dev_depends_t> kDevFuncDependsMap = {
   {"rsmi_dev_vendor_id_get",             {{kDevVendorIDFName}, {}}},
 
   {"rsmi_dev_name_get",                  {{kDevVendorIDFName,
-                                                        kDevDevIDFName}, {}}},
+                                           kDevDevIDFName}, {}}},
+  {"rsmi_dev_sku_get",                   {{kDevDevProdNumFName}, {}}},
   {"rsmi_dev_brand_get",                 {{kDevVendorIDFName}, {}}},
   {"rsmi_dev_vendor_name_get",           {{kDevVendorIDFName}, {}}},
   {"rsmi_dev_serial_number_get",         {{kDevSerialNumberFName}, {}}},
@@ -798,6 +803,8 @@ int Device::readDevInfo(DevInfoTypes type, std::string *val) {
     case kDevPerfLevel:
     case kDevUsage:
     case kDevOverDriveLevel:
+    case kDevDevProdName:
+    case kDevDevProdNum:
     case kDevDevID:
     case kDevSubSysDevID:
     case kDevSubSysVendorID: