Part Number:AM5716
Tool/software: TI-RTOS
(This is a repost, because the other post got locked/closed before solved).
Due to other priorities, I did not have time to follow up this issue the last month, but I was hoping to solve this with you now Garrett Ding.
Original Question:
Hello.
We have some trouble in our project with the network. We have custom board based on both AM5716 and AM5728, but the same issue is seen also on the evaluation modules.
It seems like that when the task running the network is preempted, and then continued, we get an error (data abort) in cpsw_nimu_eth.c line 606. (in pdk 1.0.8 , 1.0.9 and 1.0.10)
We have also had this similar problem when suspending the program in CCS, and then resuming again. This is the cause for a difficult debug procedure for us (we have to unplug the network cable before debugging, or at least suspend the program for less than a few seconds). debugging on network functionality is therefore difficult.
The last time, I sent a project based on our PDK_1_0_10, where we had changed some files. Therefore, you had some trouble building. Now, in the attachment, I have used the original PDK_1_0_10, with only 1 line changed. This line is ti\drv\pm\Power.h:59, and looks like this after the change: #include "src/pmrtos/common/List.h"
The last time, when stripping our project for sensitive data, I mistakenly forgot to remove a semaphore post, resulting in a data abort when the semaphore posted 2^16 times, and resulted in some confusing when you tested my project for the real error we are looking for.
I would like to emphasize that the data abort we are looking for could be easier found when having a setup with a ethernet cable connected to a switch/router with heavy traffic. This is the setup I am using.
I would much appreciate if someone tested my project again with the setup I have mentioned, and do the following: run the program for a few seconds, confirm that you can ping the board. Then, suspend the program for more than 10 seconds (the yellow pause button in CCS). Then, resume the program again. I would be surprised if the data abort is not found then, as I now have used the original PDK_1_0_10.
Attaching 2 files: zipped project folder and my version of PDK_1_0_10 (with only 1 line change in the entire PDK).
Thanks for your continuous support,
Anders Viken.(Please visit the site to view this file)