TMS320C6678: NDK crashing during Verify_init (during NC_NetStart) when integrating example in a system

Hi,

The NDK team is notified. They will post their feedback directly here.

Best Regards,
Yordan

Hi,

Sorry for the late response. For the memory placement, the critical ones are put into the fast memory in L2, you can find out this from .cfg file:
Memory.defaultHeapInstance = Program.global.heap0;
Program.sectMap["sharedL2"] = "DDR3";
Program.sectMap["systemHeap"] = "DDR3";
Program.sectMap[".sysmem"] = "DDR3";
Program.sectMap[".args"] = "DDR3";
Program.sectMap[".cio"] = "DDR3";
Program.sectMap[".far"] = "DDR3";
Program.sectMap[".rodata"] = "DDR3";
Program.sectMap[".neardata"]= "DDR3";
Program.sectMap[".cppi"] = "DDR3";
Program.sectMap[".init_array"] = "DDR3";
Program.sectMap[".qmss"] = "DDR3";
Program.sectMap[".cinit"] = "DDR3";
Program.sectMap[".bss"] = "DDR3";
Program.sectMap[".const"] = "DDR3";
Program.sectMap[".text"] = "DDR3";
Program.sectMap[".code"] = "DDR3";
Program.sectMap[".switch"] = "DDR3";
Program.sectMap[".data"] = "DDR3";
Program.sectMap[".fardata"] = "DDR3";
Program.sectMap[".args"] = "DDR3";
Program.sectMap[".cio"] = "DDR3";
Program.sectMap[".vecs"] = "DDR3";
Program.sectMap[".DbgSection"] = "DDR3";
Program.sectMap[".far:taskStackSection"] = "L2SRAM";
Program.sectMap[".stack"] = "L2SRAM";
Program.sectMap[".nimu_eth_ll2"] = "L2SRAM";
Program.sectMap[".resmgr_memregion"] = {loadSegment: "L2SRAM", loadAlign:128}; /* QMSS descriptors region */
Program.sectMap[".resmgr_handles"] = {loadSegment: "L2SRAM", loadAlign:16}; /* CPPI/QMSS/PA Handles */
Program.sectMap[".resmgr_pa"] = {loadSegment: "L2SRAM", loadAlign:8}; /* PA Memory */
Program.sectMap[".far:IMAGEDATA"] = {loadSegment: "L2SRAM", loadAlign: 8};
Program.sectMap[".far:NDK_OBJMEM"] = {loadSegment: "L2SRAM", loadAlign: 8};
Program.sectMap[".far:NDK_PACKETMEM"] = {loadSegment: "L2SRAM", loadAlign: 128};

For the SWI, HWI and Semaphore usage, you can look at the RTOS view for the NDK NIMU example, it is under CCS ---->tools---->ROV classic, there are hwi/swi/semaphore.

For the integration, it looks that you run NDK on other cores than core 0? Did it work on core 0 or not? There is a thread on how to run NDK NIMU example on other cores than 0 for MCSDK package: e2e.ti.com/.../858287

Regards, Eric

Hi,

Thank you for your reply.

we managed to avoid core crash by commenting the for in the previous post:

#if 0
    for (i = 0; i < max_queue_number; i++ ) {
        if ( (i == gRxFreeQHnd) || (i == gTxFreeQHnd) || (i == gTxCmdFreeQHnd))
            continue;

        count = Qmss_getQueueEntryCount (i);
        if (count != 0) {
            NIMU_drv_log2 ("Verify_Init: Expected 0 entry count for Queue number = %d, found %d entries\n", i, count);
        }
    }
#endif

At the moment we are facing a different issue:

we moved the NDK stack task on core 7 (core 0 still do all the initializations in the GEL file).

From the ROV it seems that the NDK is up but it is not possible to ping the device.

Any suggestion about this?

Thank you in advance.

Best Regards,

Fabrizio

Hi,

Can you confirm after removing that code and use NDK on core 0 with your application, everything worked fine? If yes, when you move NDK to core 7, did you check the suggestion and the example we did on the link I providede2e.ti.com/.../858287 ?

Regards, Eric

Hi,

thank you for your reply.
I confirm that NDK on core 0 works fine. When I move just the NDK task initialization to core 7 everything looks fine: NDK brings up fine but the board does not ping.
I looked up the differences between the files in the solution you linked and the one I am using:

the one in that solution refer to the old libraries (MCSDK) but it seems that the new libraries already have this fix.
I am using:

pdk v 2.0.8

Bios 6.52.00.12

ndk 2_26_00_08

sdk 4.02.00.09

any other help?

Thank you

Best Regards,
Fabrizio

Hi,

i tried to do the same thing we want to do on our custom board on the EVM: I connected on Core 0 to run the GEL file initializations, after that I connected on core 7 and uploaded a version a little modified of the client example. The same example works on Core 0. I found on another post that this should be working.

to let the NDK work on core 7 i did these changes:

• Change the qmss and cppi master core to core 7:

 if (CSL_chipReadDNUM() == 7)
    {
        qmss_cfg.master_core        = 1;
    }

...

    if (CSL_chipReadDNUM() == 7)
    {
        cppi_cfg.master_core        = 1;
    }

• I added this line to my .cfg

Global.multiCoreStackRunMode = Global.STACKRUN_FLG_CORE_7;

Still the application brings up the NDK (I can see the IP assigned in the ROV view) but the board does not ping.

Any update?

Thank you in advance.

Best Regards,

Fabrizio

Hi,

When testing on core 7, if you ping it, can you set breakpoints on EmacRxPktISR and EmacSend functions to see if this is Rx or Tx problem?

Regards, Eric

Hi,

Thank you for you reply.
I've put the breakpoints where you asked and it seems that both rx and tx were successful (they both returned 0) of course the ping was not successful for timeout reasons.
When I tried to ping again to have a second look the application didn't stop again at the breakpoint, which is strange to me. Still the ping is not successful

Is there anything I can do?

Thank you in advance.
Best Regards,

Fabrizio

Hi,

I have compared the solution provided 02/12/2013 with the MCSDK 2.1.2.6 installation. The difference are:
1) In the NIMU driver side:
nimu_eth.c, PA_ACC_CHANNEL_NUM = coreNum; //so this changed to coreNum instead of core 0

2) the test application: The core you run needs to be the master
qmss_cfg.master_core = 1;
cppi_cfg.master_core = 1;

3) There is some discussion to add the Global.multiCoreStackRunMode = STACKRUN_FLG_CORE_7 in the .cfg

The first change was already inside the NIMU driver of Processor SK RTOS: pdk_c667x_2_0_7\packages\ti\transport\ndk\nimu\src\v1
accChannelNum = PA_ACC_CHANNEL_NUM + coreNum;

You also added the 2nd and 3rd changes. All the changes are supposed to be here. I don't understand why the TX and RX breakpoints hit in the first time when you ping, then didn't the next time. I need look at this further.

Regards, Eric

Hi,

Thank you for your reply.

It is strange for me too. Also if I don't use the breakpoints the ping is not successful anyway.

Could you please get back to me ASAP? This task is very critical for us.

Thank you very much.

Best Regards,

Fabrizio

Hi,

Regarding the Global.multiCoreStackRunMode = STACKRUN_FLG_CORE_7 in the .cfg, I did a run keeping this in the file but uploading the NDK on core 0. In this case it pings too.
Is it possible that in some point there is some code that gives full responsability to core 7 missing?

Thank you in advance.
Best Regards,

Fabrizio