Merge amd-dev into amd-master 20241107

Signed-off-by: Zhang Ava <niandong.zhang@amd.com> Change-Id: I06979911d0e335c142eea73f95dc3801611ac275 [ROCm/amdsmi commit: e0b913b30d]
2024-11-08 10:57:45 +08:00
parent b5b5e26843 f162de9efd
commit 7c05ee67e4
10 changed files with 113 additions and 60 deletions
@@ -591,6 +591,8 @@ GPU: 0

 ### Resolved issues

+- **Fixed `amd-smi reset` commands showing an AttributeError**.  
+
 - **Improved Offline install process & lowered dependency for PyYAML**.  

 - **Fixed CPX not showing total number of logical GPUs**.  
@@ -24,6 +24,9 @@ installed to query firmware information and hardware IPs.
 ### Requirements

 * python 3.6.8+ 64-bit
+  - prerequisite modules:
+    - python3-wheel
+    - python3-setuptools
 * amdgpu driver must be loaded for amdsmi_init() to pass

 ### Installation
@@ -91,6 +94,14 @@ Type "help", "copyright", "credits" or "license" for more information.
 >>>
 ```

+### Installing the Python Prerequisite Modules
+
+Python3-setuptools and python3-wheel can both be installed through the pip installer as shown below:
+
+```bash
+python3 -m pip install setuptools wheel
+```
+
 ## Usage Basics for the C Library

 ### Device/Socket handles
@@ -4270,7 +4270,7 @@ class AMDSMICommands():
        # Error if no subcommand args are passed
        if self.helpers.is_baremetal():
            if not any([args.gpureset, args.clocks, args.fans, args.profile, args.xgmierr, \
-                        args.perf_determinism, args.compute_partition, args.memory_partition, \
+                        args.perf_determinism, \
                        args.power_cap, args.clean_local_data]):
                command = " ".join(sys.argv[1:])
                raise AmdSmiRequiredCommandException(command, self.logger.format)
@@ -4337,8 +4337,8 @@ class AMDSMICommands():
                    logging.debug("Failed to reset fans on gpu %s | %s", gpu_id, e.get_error_info())
                self.logger.store_output(args.gpu, 'reset_fans', result)
            if args.profile:
-                reset_profile_results = {'power_profile' : '',
-                                        'performance_level': ''}
+                reset_profile_results = {'power_profile' : 'N/A',
+                                        'performance_level': 'N/A'}
                try:
                    power_profile_mask = amdsmi_interface.AmdSmiPowerProfilePresetMasks.BOOTUP_DEFAULT
                    amdsmi_interface.amdsmi_set_gpu_power_profile(args.gpu, 0, power_profile_mask)
@@ -4349,48 +4349,48 @@ class AMDSMICommands():
                    reset_profile_results['power_profile'] = "N/A"
                    logging.debug("Failed to reset power profile on gpu %s | %s", gpu_id, e.get_error_info())

-            try:
-                level_auto = amdsmi_interface.AmdSmiDevPerfLevel.AUTO
-                amdsmi_interface.amdsmi_set_gpu_perf_level(args.gpu, level_auto)
-                reset_profile_results['performance_level'] = 'Successfully reset Performance Level'
-            except amdsmi_exception.AmdSmiLibraryException as e:
-                if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
-                    raise PermissionError('Command requires elevation') from e
-                reset_profile_results['performance_level'] = "N/A"
-                logging.debug("Failed to reset perf level on gpu %s | %s", gpu_id, e.get_error_info())
+                try:
+                    level_auto = amdsmi_interface.AmdSmiDevPerfLevel.AUTO
+                    amdsmi_interface.amdsmi_set_gpu_perf_level(args.gpu, level_auto)
+                    reset_profile_results['performance_level'] = 'Successfully reset Performance Level'
+                except amdsmi_exception.AmdSmiLibraryException as e:
+                    if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
+                        raise PermissionError('Command requires elevation') from e
+                    reset_profile_results['performance_level'] = "N/A"
+                    logging.debug("Failed to reset perf level on gpu %s | %s", gpu_id, e.get_error_info())

-            self.logger.store_output(args.gpu, 'reset_profile', reset_profile_results)
-        if args.xgmierr:
-            try:
-                amdsmi_interface.amdsmi_reset_gpu_xgmi_error(args.gpu)
-                result = 'Successfully reset XGMI Error count'
-            except amdsmi_exception.AmdSmiLibraryException as e:
-                if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
-                    raise PermissionError('Command requires elevation') from e
-                result = "N/A"
-                logging.debug("Failed to reset xgmi error count on gpu %s | %s", gpu_id, e.get_error_info())
-            self.logger.store_output(args.gpu, 'reset_xgmi_err', result)
-        if args.perf_determinism:
-            try:
-                level_auto = amdsmi_interface.AmdSmiDevPerfLevel.AUTO
-                amdsmi_interface.amdsmi_set_gpu_perf_level(args.gpu, level_auto)
-                result = 'Successfully disabled performance determinism'
-            except amdsmi_exception.AmdSmiLibraryException as e:
-                if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
-                    raise PermissionError('Command requires elevation') from e
-                result = "N/A"
-                logging.debug("Failed to set perf level on gpu %s | %s", gpu_id, e.get_error_info())
-            self.logger.store_output(args.gpu, 'reset_perf_determinism', result)
-        if args.power_cap:
-            try:
-                power_cap_info = amdsmi_interface.amdsmi_get_power_cap_info(args.gpu)
-                logging.debug(f"Power cap info for gpu {gpu_id} | {power_cap_info}")
-                default_power_cap_in_w = power_cap_info["default_power_cap"]
-                default_power_cap_in_w = self.helpers.convert_SI_unit(default_power_cap_in_w, AMDSMIHelpers.SI_Unit.MICRO)
-                current_power_cap_in_w = power_cap_info["power_cap"]
-                current_power_cap_in_w = self.helpers.convert_SI_unit(current_power_cap_in_w, AMDSMIHelpers.SI_Unit.MICRO)
-            except amdsmi_exception.AmdSmiLibraryException as e:
-                raise ValueError(f"Unable to get power cap info from {gpu_id}") from e
+                self.logger.store_output(args.gpu, 'reset_profile', reset_profile_results)
+            if args.xgmierr:
+                try:
+                    amdsmi_interface.amdsmi_reset_gpu_xgmi_error(args.gpu)
+                    result = 'Successfully reset XGMI Error count'
+                except amdsmi_exception.AmdSmiLibraryException as e:
+                    if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
+                        raise PermissionError('Command requires elevation') from e
+                    result = "N/A"
+                    logging.debug("Failed to reset xgmi error count on gpu %s | %s", gpu_id, e.get_error_info())
+                self.logger.store_output(args.gpu, 'reset_xgmi_err', result)
+            if args.perf_determinism:
+                try:
+                    level_auto = amdsmi_interface.AmdSmiDevPerfLevel.AUTO
+                    amdsmi_interface.amdsmi_set_gpu_perf_level(args.gpu, level_auto)
+                    result = 'Successfully disabled performance determinism'
+                except amdsmi_exception.AmdSmiLibraryException as e:
+                    if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
+                        raise PermissionError('Command requires elevation') from e
+                    result = "N/A"
+                    logging.debug("Failed to set perf level on gpu %s | %s", gpu_id, e.get_error_info())
+                self.logger.store_output(args.gpu, 'reset_perf_determinism', result)
+            if args.power_cap:
+                try:
+                    power_cap_info = amdsmi_interface.amdsmi_get_power_cap_info(args.gpu)
+                    logging.debug(f"Power cap info for gpu {gpu_id} | {power_cap_info}")
+                    default_power_cap_in_w = power_cap_info["default_power_cap"]
+                    default_power_cap_in_w = self.helpers.convert_SI_unit(default_power_cap_in_w, AMDSMIHelpers.SI_Unit.MICRO)
+                    current_power_cap_in_w = power_cap_info["power_cap"]
+                    current_power_cap_in_w = self.helpers.convert_SI_unit(current_power_cap_in_w, AMDSMIHelpers.SI_Unit.MICRO)
+                except amdsmi_exception.AmdSmiLibraryException as e:
+                    raise ValueError(f"Unable to get power cap info from {gpu_id}") from e

                if current_power_cap_in_w == default_power_cap_in_w:
                    self.logger.store_output(args.gpu, 'powercap', f"Power cap is already set to {default_power_cap_in_w}")
@@ -5239,6 +5239,9 @@ class AMDSMICommands():
                    current_mem_cap = "N/A"
                    logging.debug("Failed to get current memory partition capabilties for GPU %s | %s", gpu_id, e.get_error_info())

+                if profile_type == 0:
+                    profile_type = "N/A"
+
                tabular_output_dict = {"gpu_id": gpu_id,
                                       "memory": current_mem_cap,
                                       "accelerator_type": profile_type,
@@ -5298,6 +5301,7 @@ class AMDSMICommands():
                self.logger.store_multiple_device_output()
            self.logger.print_output(multiple_device_enabled=True)
            self.logger.clear_multiple_devices_ouput()
+
        if args.accelerator:
            self.logger.table_header = ''.rjust(7)
            current_header = "GPU_ID".ljust(13) + \
@@ -5361,6 +5365,9 @@ class AMDSMICommands():
                    mem_caps_str = "N/A"
                    logging.debug("Failed to get accelerator partition profile for GPU %s | %s", gpu_id, e.get_error_info())

+                if profile_type == 0:
+                    profile_type = "N/A"
+
                tabular_output_dict = {"gpu_id": gpu_id,
                                       "profile_index": profile_index,
                                       "memory_partition_caps": mem_caps_str,
@@ -691,11 +691,11 @@ typedef union {
 typedef struct {
  amdsmi_accelerator_partition_type_t  profile_type;   // SPX, DPX, QPX, CPX and so on
  uint32_t num_partitions;  // On MI300X, SPX: 1, DPX: 2, QPX: 4, CPX: 8, length of resources array
-  uint32_t profile_index;
  amdsmi_nps_caps_t memory_caps;             // Possible memory partition capabilities
+  uint32_t profile_index;
  uint32_t num_resources;                    // length of index_of_resources_profile
  uint32_t resources[AMDSMI_MAX_ACCELERATOR_PARTITIONS][AMDSMI_MAX_CP_PROFILE_RESOURCES];
-  uint64_t reserved[6];
+  uint64_t reserved[13];
 } amdsmi_accelerator_partition_profile_t;

 typedef enum {
@@ -82,6 +82,7 @@ int SameFile(const std::string fileA, const std::string fileB);
 bool FileExists(char const *filename);
 std::vector<std::string> globFilesExist(const std::string& filePattern);
 int isRegularFile(std::string fname, bool *is_reg);
+int isReadOnlyForAll(const std::string& fname, bool *is_read_only);
 int ReadSysfsStr(std::string path, std::string *retStr);
 int WriteSysfsStr(std::string path, std::string val);
 bool IsInteger(const std::string & n_str);
@@ -71,14 +71,17 @@ validClockNames.sort()
 def driverInitialized():
    """ Returns true if amdgpu is found in the list of initialized modules
    """
-    driverInitialized = ''
-    try:
-        driverInitialized = str(subprocess.check_output("cat /sys/module/amdgpu/initstate |grep live", shell=True))
-    except subprocess.CalledProcessError:
-        pass
-    if len(driverInitialized) > 0:
-        return True
-    return False
+    driverInitialized = False
+    if os.path.exists("/sys/module/amdgpu") :
+        if os.path.exists("/sys/module/amdgpu/initstate"):
+            # amdgpu is loadable module
+            with open("/sys/module/amdgpu/initstate") as initstate:
+                if 'live' in initstate.read():
+                    driverInitialized = True
+        else:
+            # amdgpu is built into the kernel
+            driverInitialized = True
+    return driverInitialized


 def formatJson(device, log):
@@ -2054,7 +2054,7 @@ rsmi_dev_gpu_clk_freq_set(uint32_t dv_ind,
    return ret;
  }

-  int ret_i;
+  rsmi_status_t status;
  amd::smi::DevInfoTypes dev_type;

  const auto & clk_type_it = kClkTypeMap.find(clk_type);
@@ -2064,9 +2064,20 @@ rsmi_dev_gpu_clk_freq_set(uint32_t dv_ind,
    return RSMI_STATUS_INVALID_ARGS;
  }

-  ret_i = dev->writeDevInfo(dev_type, freq_enable_str);
-  return amd::smi::ErrnoToRsmiStatus(ret_i);
+  status =  amd::smi::ErrnoToRsmiStatus(dev->writeDevInfo(dev_type, freq_enable_str));

+  // If an operation is not supported, the dev file, ie /sys/class/drm/card1/device/pp_dpm_pcie
+  // will have read-only perms, and the OS will deny access, before the request hits the driver level
+  if (status == RSMI_STATUS_PERMISSION){
+    bool read_only = false;
+    int perms = amd::smi::isReadOnlyForAll(dev->path(), &read_only);
+    if(read_only){
+      return RSMI_STATUS_NOT_SUPPORTED;
+    }
+  }
+
+  return status;
+  
  CATCH
 }

@@ -173,6 +173,24 @@ int isRegularFile(std::string fname, bool *is_reg) {
  return 0;
 }

+int isReadOnlyForAll(const std::string& fname, bool *is_read_only){
+  struct stat file_stat;
+  int ret;
+
+  ret = stat(fname.c_str(), &file_stat);
+  if (ret) {
+    return errno;
+  }
+
+  if (is_read_only != nullptr) {
+    *is_read_only = (file_stat.st_mode & (S_IRUSR | S_IRGRP | S_IROTH)) && !(file_stat.st_mode & (S_IWUSR | S_IWGRP | S_IWOTH));
+  } else {
+    ret = 1;
+  }
+
+  return ret;
+}
+
 int WriteSysfsStr(std::string path, std::string val) {
  //  On success, zero is returned.  On error, -1 is returned, and
  //  errno is set to indicate the error.
@@ -576,7 +576,7 @@ amdsmi_status_t smi_amdgpu_get_market_name_from_dev_id(uint32_t device_id, char
            break;
        case 0x74a1:
        case 0x74b5:
-            strcpy(market_name, "MI300X-O");
+            strcpy(market_name, "AMD Instinct MI300X");
            break;
        case 0x74a2:
        case 0x74b6:
@@ -4,7 +4,7 @@ option(INSTALL_GTEST "Install GTest (only useful if GTest is not already install
 # Help tests find libraries at runtime
 set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -Wl,--enable-new-dtags")
 set(CMAKE_INSTALL_RPATH
-    "\$ORIGIN:\$ORIGIN/../../../lib"
+    "\$ORIGIN:\$ORIGIN/../../../lib:CPACK_PACKAGING_INSTALL_PREFIX/lib"
    CACHE STRING "RUNPATH for tests. Helps find libgtest.so and libamd_smi.so")

 # Download and compile googletest
@@ -76,6 +76,6 @@ install(

 # Install googletest libraries with tests
 install(TARGETS gtest gtest_main
-    DESTINATION ${SHARE_INSTALL_PREFIX}/tests
+    DESTINATION ${CPACK_PACKAGING_INSTALL_PREFIX}/lib
    COMPONENT ${TESTS_COMPONENT})