[SWDEV-516592] Add python interface API for Bad Page Threshold (#141)

- Added python interface APIs for amdsmi_get_gpu_bad_page_threshold()
 - Updated the docs and changelog.

---------

Signed-off-by: Kanangot Balakrishnan, Bindhiya <Bindhiya.KanangotBalakrishnan@amd.com>
Signed-off-by: Arif, Maisam <Maisam.Arif@amd.com>

CHANGELOG.md

@@ -8,6 +8,21 @@ Full documentation for amd_smi_lib is available at [https://rocm.docs.amd.com/pr
 
 ### Added
 
+- **Added bad page threshold count**.  
+  - Added `amdsmi_get_gpu_bad_page_threshold` to Python API and CLI; root/sudo permissions required to display the count.
+
+### Changed
+
+### Removed
+
+### Optimized
+
+### Resolved issues
+
+### Upcoming changes
+
+### Known issues
+
 - **Added cpu model name for RDC**.  
   - Added new C and Python API `amdsmi_get_cpu_model_name`
   - Not sourced from esmi library.

amdsmi_cli/amdsmi_commands.py

@@ -712,6 +712,7 @@ class AMDSMICommands():
         if 'ras' in current_platform_args:
             if args.ras:
                 ras_dict = {"eeprom_version": "N/A",
+                            "bad_page_threshold": "N/A",
                             "parity_schema" : "N/A",
                             "single_bit_schema" : "N/A",
                             "double_bit_schema" : "N/A",
@@ -735,6 +736,10 @@ class AMDSMICommands():
                     ras_dict.update(ras_info)
                 except amdsmi_exception.AmdSmiLibraryException as e:
                     logging.debug("Failed to get ras info for gpu %s | %s", gpu_id, e.get_error_info())
+                try:
+                    ras_dict["bad_page_threshold"] = amdsmi_interface.amdsmi_get_gpu_bad_page_threshold(args.gpu)
+                except amdsmi_exception.AmdSmiLibraryException as e:
+                    logging.debug("Failed to get bad page threshold count for gpu %s | %s", gpu_id, e.get_error_info())
 
                 try:
                     ras_states = amdsmi_interface.amdsmi_get_gpu_ras_block_features_enabled(args.gpu)

amdsmi_cli/amdsmi_parser.py

@@ -682,7 +682,8 @@ class AMDSMIParser(argparse.ArgumentParser):
         clock_help = f"Show one or more valid clock frequency levels. Available options:\n\t{clk_option_str}"
 
         # Options arguments help text for Hypervisors and Baremetal
-        ras_help = "Displays RAS features information"
+        # Might be able to remove Sudo requirement in ROCm 7.0
+        ras_help = "Displays RAS features information;\n\tSudo may be required for some features"
         numa_help = "All numa node information" # Linux Baremetal only
         partition_help = "Partition information"
 

docs/reference/amdsmi-py-api.md

@@ -990,6 +990,38 @@ except AmdSmiException as e:
     print(e)
 ```
 
+### amdsmi_get_gpu_bad_page_threshold
+
+Description:  Returns bad page threshold for the given GPU; Requires root level access to display bad page threshold count; otherwise will return "N/A".
+It is not supported on virtual machine guest
+
+Input parameters:
+
+* `processor_handle` device which to query
+
+Output: Bad page threshold value
+
+Exceptions that can be thrown by `amdsmi_get_gpu_bad_page_threshold` function:
+
+* `AmdSmiLibraryException`
+* `AmdSmiRetryException`
+* `AmdSmiParameterException`
+
+Example:
+
+```python
+try:
+    devices = amdsmi_get_processor_handles()
+    if len(devices) == 0:
+        print("No GPUs on machine")
+    else:
+        for device in devices:
+            threshold = amdsmi_get_gpu_bad_page_threshold(device)
+            print(bad_page["threshold"])
+except AmdSmiException as e:
+    print(e)
+```
+
 ### amdsmi_get_gpu_memory_reserved_pages
 
 Description: Returns reserved memory page info for the given GPU.

py-interface/__init__.py

@@ -109,6 +109,7 @@ from .amdsmi_interface import amdsmi_get_clock_info
 
 from .amdsmi_interface import amdsmi_get_pcie_info
 from .amdsmi_interface import amdsmi_get_gpu_bad_page_info
+from .amdsmi_interface import amdsmi_get_gpu_bad_page_threshold
 from .amdsmi_interface import amdsmi_get_violation_status
 from .amdsmi_interface import amdsmi_get_gpu_xgmi_link_status
 

py-interface/amdsmi_interface.py

@@ -2214,6 +2214,22 @@ def amdsmi_get_gpu_bad_page_info(
 
     return _format_bad_page_info(bad_pages, num_pages)
 
+def amdsmi_get_gpu_bad_page_threshold(
+    processor_handle: amdsmi_wrapper.amdsmi_processor_handle,
+) -> Dict[str, Any]:
+    if not isinstance(processor_handle, amdsmi_wrapper.amdsmi_processor_handle):
+        raise AmdSmiParameterException(
+            processor_handle, amdsmi_wrapper.amdsmi_processor_handle
+        )
+
+    threshold = ctypes.c_uint32()
+    _check_res(
+        amdsmi_wrapper.amdsmi_get_gpu_bad_page_threshold(
+            processor_handle, ctypes.byref(threshold)
+        )
+    )
+
+    return threshold.value
 
 def amdsmi_get_violation_status(
     processor_handle: amdsmi_wrapper.amdsmi_processor_handle,