[SWDEV-513807] Fix amd-smi partition --accelerator not returning AMDSMI_STATUS_NO_PERM (#192)
* [SWDEV-513807] Fix amd-smi partition --accelerator not returning AMDSMI_STATUS_NO_PERM Changes: - Fixed amdsmi_get_gpu_accelerator_partition_profile_config() from not returning AMDSMI_STATUS_NO_PERM - Changed amd-smi partition --accelerator to provide user with a warning if users does not use sudo or root permissions. - Updated changelog for fixes planned for 6.4.1 release Signed-off-by: Charis Poag <Charis.Poag@amd.com>
Этот коммит содержится в:
Poag, Charis
коммит произвёл
GitHub
GitHub
родитель
4a3c70136f
Коммит
0402bb4d75
+107
-2
@@ -18,8 +18,101 @@ Full documentation for amd_smi_lib is available at [https://rocm.docs.amd.com/pr
|
||||
- Increasing available JPEG engines to 40.
|
||||
Current ASICs may not support all 40. These will be indicated as UINT16_MAX or N/A in CLI.
|
||||
|
||||
|
||||
## amd_smi_lib for ROCm 6.4.1
|
||||
|
||||
### Added
|
||||
|
||||
- **Added `amdsmi_get_power_info_v2()` with `sensor_ind`**.
|
||||
|
||||
### Changed
|
||||
|
||||
- **Changed amd-smi partition --accelerator & `amdsmi_get_gpu_accelerator_partition_profile_config()` detect users running without root/sudo privledges**
|
||||
- Updated `amdsmi_get_gpu_accelerator_partition_profile_config()` to return `AMDSMI_STATUS_NO_PERM` immediately
|
||||
if users run without root/sudo permissions.
|
||||
- Updated `amd-smi partition --accelerator` to provide a warning for users without root/sudo permissions (see example below, ***output subject to change***).
|
||||
```shell
|
||||
$ amd-smi partition --accelerator
|
||||
|
||||
ACCELERATOR_PARTITION_PROFILES:
|
||||
|
||||
***************************************************************************
|
||||
** WARNING: **
|
||||
** ACCELERATOR_PARTITION_PROFILES requires sudo/root permissions to run. **
|
||||
** Please run the command with sudo permissions to get accurate results. **
|
||||
***************************************************************************
|
||||
|
||||
GPU_ID PROFILE_INDEX MEMORY_PARTITION_CAPS ACCELERATOR_TYPE PARTITION_ID NUM_PARTITIONS NUM_RESOURCES RESOURCE_INDEX RESOURCE_TYPE RESOURCE_INSTANCES RESOURCES_SHARED
|
||||
N/A N/A N/A N/A 0 N/A N/A N/A N/A N/A N/A
|
||||
N/A N/A N/A N/A 0 N/A N/A N/A N/A N/A N/A
|
||||
N/A N/A N/A N/A 0 N/A N/A N/A N/A N/A N/A
|
||||
N/A N/A N/A N/A 0 N/A N/A N/A N/A N/A N/A
|
||||
N/A N/A N/A N/A 0 N/A N/A N/A N/A N/A N/A
|
||||
N/A N/A N/A N/A 0 N/A N/A N/A N/A N/A N/A
|
||||
N/A N/A N/A N/A 0 N/A N/A N/A N/A N/A N/A
|
||||
N/A N/A N/A N/A 0 N/A N/A N/A N/A N/A N/A
|
||||
|
||||
ACCELERATOR_PARTITION_RESOURCES:
|
||||
RESOURCE_INDEX RESOURCE_TYPE RESOURCE_INSTANCES RESOURCES_SHARED
|
||||
N/A N/A N/A N/A
|
||||
N/A N/A N/A N/A
|
||||
N/A N/A N/A N/A
|
||||
N/A N/A N/A N/A
|
||||
N/A N/A N/A N/A
|
||||
N/A N/A N/A N/A
|
||||
N/A N/A N/A N/A
|
||||
N/A N/A N/A N/A
|
||||
|
||||
|
||||
Legend:
|
||||
* = Current mode
|
||||
```
|
||||
|
||||
- **Changed `amd-smi partition --current`, `amd-smi partition --accelerator`, and `amdsmi_get_gpu_accelerator_partition_profile()` to display partition ID for each individual partition**
|
||||
- Host will continue to display in the full array format, they do not display the individual partitions as Baremetal/Guest setups.
|
||||
- Baremetal and Guest MI3x setups will change to
|
||||
reflect each individual partition ID, now provided in `partition_id[0]` location (as seen in other amd-smi CLI commands).
|
||||
This change was needed for BM/Guest setups due to other related partition outputs seen in (`amd-smi list` and `amd-smi static --partition`) and individual logical partition devices displayed. ***See examples below for reference.***
|
||||
|
||||
Previous output:
|
||||
```shell
|
||||
$ amd-smi partition --current
|
||||
|
||||
CURRENT_PARTITION:
|
||||
GPU_ID MEMORY ACCELERATOR_TYPE ACCELERATOR_PROFILE_INDEX PARTITION_ID
|
||||
0 NPS1 CPX 3 0,1,2,3,4,5,6,7
|
||||
1 NPS1 CPX 3 N/A
|
||||
2 NPS1 CPX 3 N/A
|
||||
3 NPS1 CPX 3 N/A
|
||||
4 NPS1 CPX 3 N/A
|
||||
5 NPS1 CPX 3 N/A
|
||||
6 NPS1 CPX 3 N/A
|
||||
7 NPS1 CPX 3 N/A
|
||||
8 NPS1 CPX 3 0,1,2,3,4,5,6,7
|
||||
9 NPS1 CPX 3 N/A
|
||||
10 NPS1 CPX 3 N/A
|
||||
...
|
||||
```
|
||||
|
||||
New output:
|
||||
```shell
|
||||
amd-smi partition --current
|
||||
CURRENT_PARTITION:
|
||||
GPU_ID MEMORY ACCELERATOR_TYPE ACCELERATOR_PROFILE_INDEX PARTITION_ID
|
||||
0 NPS1 CPX 3 0
|
||||
1 NPS1 CPX 3 1
|
||||
2 NPS1 CPX 3 2
|
||||
3 NPS1 CPX 3 3
|
||||
4 NPS1 CPX 3 4
|
||||
5 NPS1 CPX 3 5
|
||||
6 NPS1 CPX 3 6
|
||||
7 NPS1 CPX 3 7
|
||||
8 NPS1 CPX 3 0
|
||||
9 NPS1 CPX 3 1
|
||||
10 NPS1 CPX 3 2
|
||||
...
|
||||
```
|
||||
|
||||
### Removed
|
||||
|
||||
- **Removed `sensor_ind` in `amdsmi_get_power_info()` for backwards compatibility**.
|
||||
@@ -27,9 +120,21 @@ Full documentation for amd_smi_lib is available at [https://rocm.docs.amd.com/pr
|
||||
- Python API still accepts `sensor_ind` as an optional argument
|
||||
- Changed AMDSMI version from 25.2 to 25.3
|
||||
|
||||
### Added
|
||||
### Optimized
|
||||
|
||||
- **Added `amdsmi_get_power_info_v2()` with `sensor_ind`**.
|
||||
- N/A
|
||||
|
||||
### Resolved issues
|
||||
|
||||
- **Fixed `amd-smi static --partition` for guest systems with MIx ASICs being unable to run**
|
||||
|
||||
### Upcoming changes
|
||||
|
||||
- N/A
|
||||
|
||||
### Known issues
|
||||
|
||||
- N/A
|
||||
|
||||
## amd_smi_lib for ROCm 6.4.0
|
||||
|
||||
|
||||
@@ -5882,7 +5882,7 @@ class AMDSMICommands():
|
||||
tabular_output.append(tabular_output_dict)
|
||||
|
||||
self.logger.multiple_device_output = tabular_output
|
||||
self.logger.table_title = "\nCURRENT_PARTITION"
|
||||
self.logger.table_title = "CURRENT_PARTITION"
|
||||
self.logger.print_output(multiple_device_enabled=True, tabular=True, dynamic=True)
|
||||
self.logger.clear_multiple_devices_output()
|
||||
|
||||
@@ -6054,8 +6054,18 @@ class AMDSMICommands():
|
||||
|
||||
self.logger.multiple_device_output = tabular_output
|
||||
self.logger.table_title = "\nACCELERATOR_PARTITION_PROFILES"
|
||||
# only display warning message if not running as root or with sudo
|
||||
if os.geteuid() != 0:
|
||||
self.logger.warning_message = """
|
||||
***************************************************************************
|
||||
** WARNING: **
|
||||
** ACCELERATOR_PARTITION_PROFILES requires sudo/root permissions to run. **
|
||||
** Please run the command with sudo permissions to get accurate results. **
|
||||
***************************************************************************
|
||||
"""
|
||||
self.logger.print_output(multiple_device_enabled=True, tabular=True, dynamic=True)
|
||||
self.logger.clear_multiple_devices_output()
|
||||
self.logger.warning_message = "" # clear the warning message
|
||||
|
||||
#########################################
|
||||
# print accelerator partition resources #
|
||||
|
||||
@@ -40,6 +40,7 @@ class AMDSMILogger():
|
||||
self.table_header = ""
|
||||
self.secondary_table_title = ""
|
||||
self.secondary_table_header = ""
|
||||
self.warning_message = ""
|
||||
self.helpers = AMDSMIHelpers()
|
||||
|
||||
|
||||
@@ -827,6 +828,8 @@ class AMDSMILogger():
|
||||
primary_table_heading = ''
|
||||
if self.table_title:
|
||||
primary_table_heading = self.table_title + ':\n'
|
||||
if self.warning_message: # Add warning message below the table title
|
||||
primary_table_heading += self.warning_message + '\n'
|
||||
primary_table_heading += self.table_header + '\n'
|
||||
primary_table = primary_table_heading + primary_table
|
||||
|
||||
@@ -884,6 +887,8 @@ class AMDSMILogger():
|
||||
primary_table_heading = ''
|
||||
if self.table_title:
|
||||
primary_table_heading = self.table_title + ':\n'
|
||||
if self.warning_message: # Add warning message below the table title
|
||||
primary_table_heading += self.warning_message + '\n'
|
||||
primary_table_heading += self.table_header + '\n'
|
||||
primary_table = primary_table_heading + primary_table
|
||||
|
||||
|
||||
@@ -5536,8 +5536,7 @@ amdsmi_set_gpu_memory_partition_mode(amdsmi_processor_handle processor_handle,
|
||||
/**
|
||||
* @brief Version 2.0: Returns gpu accelerator partition caps as currently configured in the system
|
||||
* User must use admin/sudo privledges to run this API, or API will not be able to
|
||||
* read resources. Otherwise, API will fill in the structure with as much information as
|
||||
* it can.
|
||||
* read resources.
|
||||
*
|
||||
* @ingroup tagAcceleratorPartition
|
||||
*
|
||||
|
||||
@@ -2003,6 +2003,9 @@ amdsmi_status_t
|
||||
amdsmi_get_gpu_accelerator_partition_profile_config(amdsmi_processor_handle processor_handle,
|
||||
amdsmi_accelerator_partition_profile_config_t *profile_config) {
|
||||
AMDSMI_CHECK_INIT();
|
||||
if (!amd::smi::is_sudo_user()) {
|
||||
return AMDSMI_STATUS_NO_PERM;
|
||||
}
|
||||
std::ostringstream ss;
|
||||
ss << __PRETTY_FUNCTION__
|
||||
<< " | START ";
|
||||
|
||||
Ссылка в новой задаче
Block a user