RTOS/TMS320C6678: MessageQ_create() blocked and do not return

Hi,

When I watch rov, I see something below:

I'm using transportqmss. And rov indicates that my core7' s gatemp is entered by core0, which causes it blocked in messageq_create().

My core0 may recieve msg from or send msg to core7, and core7 may recieve and send too.

In my case ,how can I create msgQ successfully when using transportqmss during runtime?

Thanks.

The Memory Browser shows the gate address is within heapMem:

What I have found from the init code about this is :

        /* Create the heap that will be used to allocate messages. */
        GateMP_Params_init(&gateMpParams);
        gateMpParams.localProtect = GateMP_LocalProtect_INTERRUPT;
        gateMpHandle = GateMP_create(&gateMpParams);

        HeapBufMP_Params_init(&heapBufParams);
        heapBufParams.regionId  = 0;
        heapBufParams.name      = QMSS_MSGQ_HEAP_NAME;
        heapBufParams.numBlocks = HOST_DESC_NUM;
        heapBufParams.blockSize = QMSS_MTU_SIZE_BYTES;
        /* GateMP so allocation can take place within QMSS rx interrupt
         * context */
        heapBufParams.gate      = gateMpHandle;
        heapHandle = HeapBufMP_create(&heapBufParams);

what will gateMp_localprotect_interrupt behave?

My core0 receives msgs from all other cores and then process thenm and finally send back.

what may cause it enter the gate and do not leave?

Hi,

I try add MessageQ.maxRuntimeEntries = 40 in my .cfg files, but it didn't work.

Other cores can't return successfully either.

Below is what I see from rov, there are some errors report by bios.

Hi,

I have read ex11_ping and found it's a bit different from my application.The difference between them is that my application call messageQ_create() when there are many other cores who are calling message_put() and message_get(). That is , my application( in a special core) will call messageQ_create() when there is a high concurrency situation (many other cores who works like the ex11_ping sending msg).

Below are some example codes to show how my application works:

core1 - core3:

while(TRUE){
     MessageQ_get(localHandle, &msg);
     processMsg(msg);
     MessageQ_put(CORE0_queue, msg);    //send msg to core0
}

core0:

while(TRUE){
     MessageQ_get(localHandle, &msg);       //recieve msg from other cores
     processMsg(msg);
     MessageQ_put(MessageQ_getReplyQueue(msg), msg);
}

And at sometime during runtime:

core4:

MessageQ_create(msgQueueName, NULL);

In this point ,core4 will be blocked.

However, when I change core1~3's code to 

while(TRUE){
     MessageQ_get(localHandle, &msg);
     processMsg(msg);
     MessageQ_put(CORE0_queue, msg);    //send msg to core0

     for(i = 0; i < 1000; i++)
        for(j = 0; j < 1000; j++);
}

core4 can return from messageQ_create() successfully.

However, I don't want to add some code like meaningless "for" loop. And I don't know how to adjust  i and j 's value to get it not blocked exactly.

Can I check whether messageQ_create() will be blocked or not before I really call it? and How can I do to make messageQ_create() not blocked when there is a high concurrency situation?

Thanks for your reply.

HI,

I find that core4 has created a msgQ with name I give.

Below is what I see from rov:

taskExecutor_CORE4 is the name I pass to messageQ_create() which block my task.

Why the msgQ is created successfully and it still does not return?

Thanks for your help.