Hi, TI experts
HW:TDA4ECO, Customer Board
Our SDK version:
Linux 08.06.01.02
Rtos: 08.06.01.03
BOOT: SBL HLOS boot
We are doing pressure test, 23 seconds power on, 5 seconds power off
The phenomenon I described may be a bit chaotic, but the situation is really like this
At the beginning, We encountered the following types of GPU error logs:
1. PVRSRV_ERROR_TIMEOUT -- can's see display on the screen -- Issue reproduce about once every 100-200 times
2. Mismatching sizes of RGXFW_ALIGN_CHECKS_INIT -- can't see display on the screen -- Issue reproduce about once every 50-100 times
3. Safety Watchdog Trigger ! [358] -- continue to print this log -- can't see display on the screen -- issue reproduce about one every 200+ (the probability is uncertain, only encountered once)
新的问题--watchdog-trigger.log.txt
After encountering these issues, I searched for similar issues on e2e and tried a patch, but it seems to have not improved my problem, which is a bit strange. If necessary, I can continue to test this patch in the future
patch link: [FAQ] TDA4VL-Q1: What are the GPU driver bug fixes for SDK 8.6 or earlier?
Then I found a post that was helpful for debugging. I removed the patch I had previously apply() and followed the steps in the post to capture the log:
We captured the GPU debug logs by following the steps below:
1. the system will automatically insmod the pvr driver
2. run 'pvrdebug -loggroups main,mts,hwr' before run the gpu app
3. run run our app demo
4. pvrlogdump
The main modification is to execute commands such as ‘pvrdebug -loggroups main, mts, hwr’ before running the GPU demo app (which we wrote ourselves).
GPU demo link: caputre node -> ldc node -> (gpu...) -> csitx node
I have tried to power on and off 300 times but have not encountered the previous error. It seems that a new error has appeared. I will attach the log and video:
pvrdumplog:
log:
[com COM20] (2024-05-11_141349) COM20 (Prolific PL2303GC USB Serial COM Port (COM20)).log
video:
normal boot log:
Please help to further debug.
Tahm