[SWDEV-493274/SWDEV-514998] Add AMD SMI partition tests + Add Guest amd-smi static --partition (#127)

* [SWDEV-493274/SWDEV-514998] Add AMD SMI partition tests + Add Guest amd-smi static --partition

Changes:
    - Added amd-smi static --partition for guest systems
    - Added C++ tests for memory and compute (accelerator) partitions
    - Added Python tests for amdsmi_get_gpu_vram_info(),
       amdsmi_get_gpu_accelerator_partition_profile_config()
    - Updated Python tests for
      amdsmi_get_gpu_accelerator_partition_profile()
      Now includes more profile and resource detail
    - Added amdsmi_get_gpu_xcd_counter();
      Tests provided for both C++/Python APIs
    - Added AmdSmiVramType & AmdSmiVramVendor: they were missing
      python testing required adding.

Change-Id: Ib6549d8ccc5fb68726f38745b87c78f890186022
Signed-off-by: Charis Poag <Charis.Poag@amd.com>

rocm_smi/src/rocm_smi.cc

@@ -6577,6 +6577,16 @@ rsmi_dev_partition_id_get(uint32_t dv_ind, uint32_t *partition_id) {
   if (ret == RSMI_STATUS_SUCCESS) {
     *partition_id = static_cast<uint32_t>((pci_id >> 28) & 0xf);
   }
+  std::ostringstream bdf_sstream;
+  bdf_sstream << std::hex << std::setfill('0') << std::setw(4)
+  << ((pci_id >> 32) & 0xFFFFFFFF) << ":";
+  bdf_sstream << std::hex << std::setfill('0') << std::setw(2) << ((pci_id >> 8) & 0xFF) << ":";
+  bdf_sstream << std::hex << std::setfill('0') << std::setw(2) << ((pci_id >> 3) & 0xF8) << ".";
+  bdf_sstream << std::hex << std::setfill('0') << +(pci_id & 0x7);
+  bdf_sstream << "\nPartition ID ((pci_id >> 28) & 0xf): " << std::dec
+  << static_cast<int>((pci_id >> 28) & 0xf);
+  bdf_sstream << "\nPartition ID (pci_id & 0x7): " << std::dec << static_cast<int>(pci_id & 0x7);
+  // std::cout << __PRETTY_FUNCTION__ << " BDF: " << bdf_sstream.str() << std::endl;
 
   /**
    * Fall back is required due to driver changes within KFD.
@@ -6603,9 +6613,11 @@ rsmi_dev_partition_id_get(uint32_t dv_ind, uint32_t *partition_id) {
      << " | Success"
      << " | Device #: " << dv_ind
      << " | Type: partition_id"
-     << " | Data: " << *partition_id
+     << " | Data: " << static_cast<int>(*partition_id)
      << " | Returning = "
-     << getRSMIStatusString(RSMI_STATUS_SUCCESS) << " |";
+     << getRSMIStatusString(RSMI_STATUS_SUCCESS) << " |"
+     << "\n BDF: " << bdf_sstream.str() << std::endl;
+  // std::cout << ss.str() << std::endl;
   LOG_INFO(ss);
   return ret;
   CATCH

rocm_smi/src/rocm_smi_device.cc

@@ -1020,7 +1020,7 @@ int Device::readDevInfoLine(DevInfoTypes type, std::string *line) {
      << get_type_string(type) << "), returning *line = "
      << *line;
   LOG_INFO(ss);
-
+  fs.close();
   return 0;
 }
 
@@ -1103,6 +1103,7 @@ int Device::readDevInfoMultiLineStr(DevInfoTypes type,
   while (std::getline(fs, line)) {
     retVec->push_back(line);
   }
+  fs.close();
 
   if (retVec->empty()) {
     ss << "Read devInfoMultiLineStr for DevInfoType ("
@@ -1771,6 +1772,38 @@ std::string Device::readBootPartitionState<rsmi_memory_partition_type_t>(
   return boot_state;
 }
 
+rsmi_status_t Device::get_smi_device_identifiers(uint32_t device_id,
+        rsmi_device_identifiers_t *device_identifiers) {
+  bool found_device = false;
+  rsmi_status_t ret = RSMI_STATUS_NOT_SUPPORTED;
+  if (device_identifiers == nullptr) {
+    return RSMI_STATUS_INVALID_ARGS;
+  }
+
+  amd::smi::RocmSMI& smi = amd::smi::RocmSMI::getInstance();
+  auto devices = smi.devices();
+
+  for (uint32_t i = 0; i < devices.size(); i++) {
+    if (i != device_id) {
+      continue;
+    }
+    rsmi_device_identifiers_t smi_device;
+    smi_device.card_index = devices[i]->index();
+    smi_device.drm_render_minor = devices[i]->drm_render_minor();
+    smi_device.bdfid = devices[i]->bdfid();
+    smi_device.kfd_gpu_id = devices[i]->kfd_gpu_id();
+    smi_device.partition_id = devices[i]->m_partition_id;
+    smi_device.smi_device_id = i;
+    *device_identifiers = smi_device;
+    found_device = true;
+    break;
+  }
+  if (found_device) {
+    ret = RSMI_STATUS_SUCCESS;
+  }
+  return ret;
+}
+
 
 #undef RET_IF_NONZERO
 }  // namespace smi

rocm_smi/src/rocm_smi_gpu_metrics.cc

@@ -3387,7 +3387,7 @@ AMGpuMetricsPublicLatestTupl_t GpuMetricsBase_v13_t::copy_internal_to_external_m
       metrics_public_init.vcn_activity[0] = metrics_public_init.average_mm_activity;
     }
     // average_mm_activity needs to not be UIN16_MAX and
-    // metrics_public_init.xcp_stats->vcn_busy[0] should also be UIN16_MAX
+    // metrics_public_init.xcp_stats->vcn_busy[0] should also be UINT16_MAX
     if (metrics_public_init.average_mm_activity != UINT16_MAX
         && metrics_public_init.xcp_stats->vcn_busy[0] == UINT16_MAX) {
       metrics_public_init.xcp_stats->vcn_busy[0] = metrics_public_init.average_mm_activity;