WATCHDOG_DebugSop¶
REVISION HISTORY¶
| Revision No. | Description |
Date |
|---|---|---|
| 1.0 | 5/10/2025 |
1. Troubleshooting of system restart caused by watchdog during system startup process¶
After starting the device watchdog, if the watchdog fails to complete the feeding operation before the timeout time, the watchdog will trigger a system restart. This type of problem can be preliminarily classified into two categories based on the characteristics of dog feeding operations:
1.Watchdog causes system restart during system startup process
The characteristic of this type of problem is that when the system starts from one stage to another, due to various issues, the feeding operation of the next stage of the system cannot be completed in a timely manner, causing the watchdog timeout to trigger the system restart.
For example, if a user starts the watchdog during the uboot phase and then starts the kernel from uboot, but due to some reasons, the startup process gets stuck at a certain stage of the kernel startup phase and fails to progress to the kernel watchdog driver reset watchdog or user layer dog feeding operation phase, causing the watchdog to trigger a system restart.
2.Watchdog causes system restart during system operation
The characteristic of this type of problem is that the system is in a stable running phase, and due to system blocking and other issues, the process of feeding the dog periodically cannot be scheduled, resulting in dog feeding failure and watchdog timeout leading to system restart.
For example, in the kernel phase, the system opens a process to feed the dog every 10 seconds. However, due to the system blocking the feeding operation, the process could not be scheduled for execution, causing the watchdog to trigger a system restart.
1.1 Sort out the watchdog operation flowchart during the startup phase¶
The core of identifying such issues is to clarify the operation process of the watchdog during the startup phase. For each stage involved in the startup process, the following questions need to be clarified
-
Has watchdog been activated in this stage
-
The time node for starting watchdog in this stage and the set timeout duration
-
Is there any dog feeding operation and feeding cycle in this stage
1.2 Open Debug logs for each stage¶
Many of these issues are caused by the system getting stuck at a certain stage, so opening Debug logs for each stage can help us better identify the problem.
For example, in the Kernel stage, the loglevel can be set to 8.
1.3 Seeking help from FAE¶
When seeking help from FAE, please provide the results of the appeal process first.
2. Troubleshooting of system restart caused by watchdog during system operation¶
2.1 Open the debug log of the system¶
Due to the fact that this type of problem is mainly caused by system blocking, there is no unified template for locating the problem due to various reasons. The core of solving this type of problem is to reproduce the problem and collect more logs. Therefore, when reproducing the problem, try to open the debug logs that can be opened as much as possible, which can help us discover the cause of the problem faster.
For example, set the basic value of kernel printk to 8.
2.2 Seeking help from FAE¶
When seeking help from FAE, please provide the results of the above steps first.