ISSUE In Using the Network call on TMS320C6678LEVM

jagadeesh guptha

Hi,

I am Facing the following issue when I am running the NC_NcNetstart() call in my application by reusing the NDK client(client_evmc6678l) example code.

ERROR LOG:

Ethernet eventId : 48 and vectId (Interrupt) : 7
Timeout waiting for reply from PA to Pa_addMac command
ti.sysbios.heaps.HeapMem: line 354: assertion failure: A_invalidFree: Invalid free

Same error is coming when I was trying to execute the client_evmc6678l.out(Example Code), but after giving the system reset(After referring to the TI forum) ,the error is getting vanished and program is working fine.

Description About the Problem After carrying out the Experiments to my application:

1. I am trying to run the application through core0

CPPI And QMSS are getting initialized properly (since initializations are done by the core0 according to the code )

But above ERROR LOG is coming when I am trying to execute the NC_NetStart() in application.

2. I am trying to run the application through core1

CPPI And QMSS are not getting initialized properly (since initializations are done by the core0 according to the code)

But above ERROR LOG is not coming when I am trying to execute the NC_NetStart() in application.

3. I am trying to run the application through core1

CPPI And QMSS are getting initialized properly (since initializations are changed by me according to the core1)

But above ERROR LOG is coming when I am trying to execute the NC_NetStart() in application.

Code:

if (platform_get_coreid() == 1)

{

qmss_cfg.master_core = 1;

}

else

{

qmss_cfg.master_core = 0;

}

I followed and tried all the suggested solutions from the TI forum of the following links:

Applying System Reset in the debug session

http://e2e.ti.com/support/embedded/tirtos/f/355/p/239367/841042.aspx

Keeping the Processor in No boot mode

http://e2e.ti.com/support/dsp/c6000_multi-core_dsps/f/639/p/306345/1073749.aspx

Loading the GEL file to all the cores

http://e2e.ti.com/support/dsp/c6000_multi-core_dsps/f/639/t/255715.aspx

Running the Global Default setup scripts after keeping the break point in the code

http://e2e.ti.com/support/embedded/tirtos/f/355/p/239367/841042.aspx

Making the memory free check

http://downloads.ti.com/dsps/dsps_public_sw/sdo_sb/targetcontent/bios/sysbios/6_33_06_50/exports/bios_6_33_06_50/docs/cdoc/ti/sysbios/heaps/HeapMem.html#per-instance_creation

But none provided fruitful results to me,

Could you please help me out soon to resolve this issue as soon as possible.

Thanks & Regards

Jagadeesh Guptha

over 12 years ago

0 Pubesh over 12 years ago

TI__Mastermind 20790 points

Jagadeesh,

The original test code(HelloWorld, Client projects) will run on Core0 only. Need to modify the code for Core1 or Core2. And also required some changes in the nimu driver code.

The nimu has to be slightly modified in order to handle a non-core0 run of the client example.
I have tested this out on the client example, and it worked on core 1.
I rebuilt the new nimu by importing the project and pointing the client_example to the newly built nimu in the PDK (C:\ti\pdk_C6678_1_1_2_6\packages\ti\transport\ndk\nimu\lib\debug).
A notable change in the nimu is the addition of the following:

nimu_eth.c
uint32_t PA_ACC_CHANNEL_NUM;

There was also a slight modification in the client example source.
Please take a closer look at this e2e post where it addresses the issue.
http://e2e.ti.com/support/dsp/c6000_multi-core_dsps/f/639/p/165415/858287.aspx

Something to note in the execution of the example is that core0 has to be connected, and the GEL executed before trying to run the client in a core other than core0.