message queue

If MessageQ_get on A8 doesn't unblock it means remote core didn't respond. In this case if it is VPSS M3 then VPSS M3 has probably crashed. Check if you are getting any exception or assert msg in the console logs.

MessageQ is not suitable for sending msg at 1 us interval. It is has too much overhead to be suitable for  sending at 1 us rate.

Hi Badri,

Thanks for your reply. May I know which interval is suitable for MsgQ ? (16us ?) 

WHat functionality do you want to realize ? Why are you trying to  send message to M3VPSS every 1 us ?

Hi Badri,

We have several tasks and may have chance to send message at the same time. We know the system has mutex to avoid this situation, but it sometimes hang. To send message to M3VPSS every 1us is to reproduce quickly.

Hi Badri,

Sometimes I will get the below message. What does it mean ?

[m3vpss ] !!!XDC RUNTIME ASSERT FAILED

[m3vpss ] xdc.runtime.Error @ ti.sysbios.gates.GateMutex: line 97:
[m3vpss ] assertion failure: A_badContext: bad calling context. See GateMutex API doc for details.

Assert indicates GateMutex API is called from non-task context like SWI or HWI or withing critical section where HWI/SWI are disabled.Get the exception trace to check if stacktrace is available.

Hi Badri,

This error happens when I add some debug message in utils_mbox. It seems like some HW behaviors have timing issue and need to back quickly. Another issue is why TI disable exception notify in RDK 4.1 ?

Jack 123 said:

This error happens when I add some debug message in utils_mbox. It seems like some HW behaviors have timing issue and need to back quickly

GateMutex assert  is not an issue with hw timing. It is because you have introduced a bug by trying to print from ISR context. Use Vps_rprintf() if you want to print from ISR context.

Jack 123 said:

Another issue is why TI disable exception notify in RDK 4.1 ?

Because not all application register for Exception event callback. You can enable it if you require.

Hi Badri,

Thanks for your quickly reply. I find sometimes A8 wait M3 and M3 queues are full. It seems the communication is break. Do you have any idea ?

thanks.

Jack,

When the symptom "A8 wait M3 and M3 queues are full" happened, have you verified if M3 is still alive? I ask this because the symptom may also be caused by M3 hang.

Best Regards,

Luke Lin

Hi Luke,

M3 is still alive since the debug message is active.

Hi Luke,

It seems M3VPSS is crashed, but M3VIDEO is alive.

Jack, 

Thanks for the update. This seems a power problem that causes memory corruption then results in M3 hang. I suggest to check your h/w first.

Best Regards,

Luke Lin

Hi Luke,

We have sent the board to TI lab and the report of AVS power is ok.