# NPU驱动升级失败如何解决？
在NPU驱动升级过程中，可能因版本兼容性或环境配置等问题导致命令报错。本文整理了NPU驱动升级过程中的典型报错场景及其解决方案，您可根据实际报错代码或错误描述快速解决问题。
#### 报错场景一：查询NPU驱动信息时，报错-8001
- 执行代码：
  ```
  npu-smi info
  ```
  
- 报错信息：
  ```
  dcmi module initialize failed. ret is -8001
  ```
  
- 解决方案：提前在[文档中心](https://support.huawei.com/enterprise/zh/category/ascend-computing-pid-1557196528909?submodel=doc)下载对应的驱动软件包，再利用**./{product name}-npu-driver_x.x.x_linux-{arch}.run--upgrade** 命令升级驱动，具体示例如下：
  ```
  ./A300-3010-npu-driver_24.1.rc3.1_linux-aarch64.run --upgrade
  ```
  若回显结果如下，则说明升级成功。
  ```
  Verifying archive integrity...  100%   SHA256 checksums are OK. All good.
  Uncompressing ASCEND DRIVER RUN PACKAGE  100%  
  [Driver] [2025-04-17 09:39:16] [INFO]Start time: 2025-04-17 09:39:16
  [Driver] [2025-04-17 09:39:16] [INFO]LogFile: /var/log/ascend_seclog/ascend_install.log
  [Driver] [2025-04-17 09:39:16] [INFO]OperationLogFile: /var/log/ascend_seclog/operation.log
  [Driver] [2025-04-17 09:39:16] [INFO]base version is 24.1.rc3.1.
  [Driver] [2025-04-17 09:39:16] [WARNING]Do not power off or restart the system during the installation/upgrade
  [Driver] [2025-04-17 09:39:16] [INFO]set username and usergroup, HwHiAiUser:HwHiAiUser
  [Driver] [2025-04-17 09:39:18] [INFO]device hot reset start
  /usr/local/Ascend/driver/script/device_hot_reset.sh: line 269: echo: write error: No such device
  [Driver] [2025-04-17 09:39:23] [INFO]driver install type: Direct
  [Driver] [2025-04-17 09:39:23] [INFO]upgradePercentage:10%
  [Driver] [2025-04-17 09:39:29] [INFO]upgradePercentage:30%
  [Driver] [2025-04-17 09:39:29] [INFO]upgradePercentage:40%
  [Driver] [2025-04-17 09:39:30] [INFO]upgradePercentage:90%
  [Driver] [2025-04-17 09:39:33] [INFO]upgradePercentage:100%
  [Driver] [2025-04-17 09:39:34] [INFO]Driver package upgraded finished! 
  [Driver] [2025-04-17 09:39:34] [WARNING]Kernel modules can not be removed, reboot needed for installation/upgrade to take effect! 
  [Driver] [2025-04-17 09:39:34] [INFO]End time: 2025-04-17 09:39:34
  ```
  
 
#### 报错场景二：执行upgrade命令时，报错0x0094
- 执行代码：
  ```
  ./A300-3010-npu-driver_24.1.rc3.1_linux-x86_64.run --upgrade
  ```
  
- 报错信息：
  ```
  Verifying archive integrity...  100%   SHA256 checksums are OK. All good.
  Uncompressing ASCEND DRIVER RUN PACKAGE  100%  
  [Driver] [2025-04-17 09:43:09] [ERROR]ERR_NO:0x0094;ERR_DES:Operation failed, An all-in-one RUN package is found in /usr/local/HiAI, which needs to be uninstalled before proceeding.
  ```
  
- 解决方案：请手动卸载驱动后，重新执行upgrade命令，具体示例如下：
  ```
  ./A300-3010-npu-driver_24.1.rc3.1_linux-x86_64.run --uninstall
  ```
  回显结果如下，则说明卸载成功。
  ```
  Verifying archive integrity...  100%   SHA256 checksums are OK. All good.
  Uncompressing ASCEND DRIVER RUN PACKAGE  100%  
  [Driver] [2025-04-22 15:18:20] [INFO]Start time: 2025-04-22 15:18:20
  [Driver] [2025-04-22 15:18:20] [INFO]LogFile: /var/log/ascend_seclog/ascend_install.log
  [Driver] [2025-04-22 15:18:20] [INFO]OperationLogFile: /var/log/ascend_seclog/operation.log
  [Driver] [2025-04-22 15:18:20] [INFO]base version is 24.1.rc3.1.
  [Driver] [2025-04-22 15:18:45] [INFO]Driver package uninstalled successfully! Reboot needed for uninstallation to take effect! 
  [Driver] [2025-04-22 15:18:45] [INFO]End time: 2025-04-22 15:18:45
  ```
  
 
#### 报错场景三：执行install命令时，报错0x0091
- 执行代码：
  ```
  ./A300-3010-npu-driver_24.1.rc3.1_linux-aarch64.run --install-for-all --full
  ```
  
- 报错信息：
  ```
  Verifying archive integrity...  100%   SHA256 checksums are OK. All good.
  Uncompressing ASCEND DRIVER RUN PACKAGE  100%  
  [Driver] [2025-04-17 09:18:42] [INFO]Start time: 2025-04-17 09:18:42
  [Driver] [2025-04-17 09:18:42] [INFO]LogFile: /var/log/ascend_seclog/ascend_install.log
  [Driver] [2025-04-17 09:18:42] [INFO]OperationLogFile: /var/log/ascend_seclog/operation.log
  [Driver] [2025-04-17 09:18:42] [WARNING]Do not power off or restart the system during the installation/upgrade
  [Driver] [2025-04-17 09:18:42] [ERROR]ERR_NO:0x0091;ERR_DES:HwHiAiUser not exists! Please add HwHiAiUser
  [Driver] [2025-04-17 09:18:42] [INFO]End time: 2025-04-17 09:18:42
  ```
  
- 解决方案：请手动添加名为HwHiAiUser的账户，添加后重新执行install命令。执行以下命令，创建名为HwHiAiUser的账户，作为昇腾NPU AI加速平台的默认系统账户，用于安全地管理和运行NPU相关服务及应用程序。
  ```
  adduser HwHiAiUser
  ```
  
 
#### 报错场景四：执行install命令时，提醒reboot
- 执行代码：
  ```
  ./A300-3010-npu-driver_24.1.rc3.1_linux-aarch64.run --install-for-all --full
  ```
  
- 存在以下两种报错信息：
  ```
  [Driver] [2025-04-17 10:09:30] [WARNING]Kernel modules can not be removed, reboot needed for installation/upgrade to take effect!
  ```
  ```
  [Driver] [2025-04-17 10:18:47] [INFO]Driver package installed successfully! Reboot needed for installation/upgrade to take effect!
  ```
  
- 解决方案：此时驱动已安装成功，您需要重启节点才可生效。建议在重启节点前进行排水操作，避免对当前节点的现存业务产生影响，具体操作请参见[节点排水](https://support.huaweicloud.com/usermanual-cce/cce_10_0605.html)。