• State
  • Locked Locked
  • Replies 3 replies
  • Subscribers 30 subscribers
  • Views 1024 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

RTOS/TM4C123GH6PZ: OS task can not be put into run mode after a burst of CAN packages came in

Eric Fu1
Intellectual 1580 points
Part Number: TM4C123GH6PZ

Tool/software: TI-RTOS

Hi Masters,

My customer is using SYS/BIOS with TM4C123GH6PZ(LM4F232H5QC) MCU and the product is T-BOX for Automotive. In this application, there are 2 Tasks, task#1 is used to process CAN packages and task#2 is used to do other jobs such as power control and so on. 2 Hardware interrupts are implemented in this application, one is GPIO, the other one is CAN interface. events are pending in Task#1 for waiting can packages. Related events are posted in CAN interface Hwi isr to wake up Task#1 for further processing. There are 2 CAN channels used in this application. The issue is that, when the CAN interupt happens with very high frequency, Task#1 will be stuck and can not be scheduled by OS for running. I suspect it is an OS issue and do following workaround: remove event post activity in the CAN HWi, replace with flag setting, and create another task#3 to check the CAN package flag with 5ms interval, if the flag is set, Task#3 will post the event that Task#1 is waiting for, this workaround can work well, but the customer still want to know the root cause, since there are some projects in mass production with the HWi triggered event architecture.

I attach a demo project that can reproduce this issue. The testing code was in pm_os.c file, there are 7 tasks and 3 system tick timer in this project. The event pending function is located in TAK_APP_POWER task, event post function was called from hwi_CAN1_fxn function in mycan.c file. Below you can also find the CAN message format that is used in this testing project:

To reproduce the issue, use a CAN massage generator to send above messages to this node with 5ms interval or smaller interval, by doing this the issue can be reproduced with less than 5min testing. I also attach the demo project.

Please help to support this case and thank you very much.

B&R

Eric

NWC FAE

https://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/908/Demo.7z

  • 0
    TI__Guru* 96960 points
    Hi Eric,

    When task1 is "stuck" can you look at ROV->Tasks->Callstacks to see where it is? What's the priority of the CAN task? Is it lower than any other tasks. Could it be getting starved by another task?

    Todd
  • 0 in reply to ToddMullanix
    TI__Intellectual 1580 points
    Hi Todd,

    I checked the ROV data of Tasks when the issue happens, the result shows that the CAN processing task is always in "Ready" state, but not blocked or running. The priority of this task is set to 5, which is the same with other tasks, while other tasks can be scheduled to run normally, but the CAN task can not run. The stack size of this task is set to 960B, peak utilization got from the ROV is 600B. Everything seems under limitation, so I suspect if it is caused by a bug of this OS. Please help on this case. The CCS version used is 5.4.0, SYS/BIOS version is SYS/BIOS 6.35.1.29. The more frequent CAN Hwi was triggered, the easier that the issue can be reproduced.

    Please give your comments on the debugging, if experiment is needed, you can find the attached project file in the first post. Looking forward to your feedback. Your support is very important to close this urgent case.

    Thanks.

    Eric
  • 0 in reply to Eric Fu1
    TI__Guru* 96960 points
    What is running? Is it a higher priority task, a Swi or Hwi?