CCS/AM6548: debugging a simple program

Hello John,

unfortunately a lot of things could go wrong.

With some more information about what you're trying to do I might be able to help you:

• I'm guessing that you're using the processor SDK RTOS for the AM65x?
• What does your "simple hello world program" look like? Is it baremetal or does it use TI-RTOS? Are you following these steps maybe? http://software-dl.ti.com/processor-sdk-rtos/esd/docs/latest/rtos/index_examples_demos.html#no-os-bare-metal-example
• I'm assuming you're trying to run your program on the A53 core?
• How did you setup CCS to connect to the AM65x? Did you use the "Basic CCS setup" from the RTOS documentation (http://software-dl.ti.com/processor-sdk-rtos/esd/docs/latest/rtos/index_how_to_guides.html#basic-ccs-setup) or the "Advanced AM65x Debug Setup with DMSC Firmware Load" (http://software-dl.ti.com/processor-sdk-rtos/esd/docs/latest/rtos/index_how_to_guides.html#advanced-am65x-debug-setup-with-dmsc-firmware-load), or something else?
• How did you boot the board, i.e. which boot medium did you use, and what was the core executing at the time you connected?
• You wrote that you've been able to connect to the A53 core - does that mean you could connect the core from the "Debug" view that lists all the cores, and the core stopped?
• What do you mean by "but the debugger does not see it, does not stop."? If you press the suspend button in CCS, does the core enter debug mode?
• Are there any error messages in the "Console" view?

The CCS sets a breakpoint at the main() address, but really execution starts way before that. The programs that you build with the processor SDK contain startup code that configures the processor, e.g. caches, MMU. This code can usually be executed only once, i.e. on a system that is mostly reset to defaults. If any of the assumptions of the startup code are not true, the system crashes in various kinds of ways. Sometimes you're at least able to halt the core, and see that it's stuck in some abort handler, but sometimes you're unable to halt the core (you should get error messages in the "Console" in that case).

Regards,

Dominic

Yes, I am using the SDK RTOS for the AM6548.

I am trying to start with bare metal, no OS. (I have DIP switch 3 set to 0.)

I am trying to run on the first A53 core.

I am using the Basic CCS setup.

Here is my program:

int main(void)
{
    int count = 0;
    printf("Hello World\n");
    while(1)
    {
        count++;
        if(count > 10000)
        {
            count = 0;
            printf("Hello again\n");
        }
    };
}

Yes, I am connecting from the debug window that lists all of the cores.

I first connect to DMSC Cortex M3 and set it to running.

Texas Instruments XDS110 USB Debug Probe/DMSC_Cortex_M3_0 (Running)

Then I connect to Cortex A53_0_0 and set it to running. 

Texas Instruments XDS110 USB Debug Probe/CortexA53_0_0 (Running(Secure))

I put a break point inside the while(1) loop.

The next thing I tried was to load the program from the Run menu

C:/CCS Workspace\hello_world_test\Debug\hello_world_test.out

There is a window (0x200) that reads "Break at address "0x200" with no debug information available, or outside of program code."

But, no messages on the console.

How did you create this hello world program, did you follow the steps from the RTOS guide (e.g. )? Is there an XDC/RTSC config file, and is the platform, target and compiler (note the aarch64) in the CCS project settings configured correctly?

If you want you can attach the .out file and I'll have a look tomorrow.

I've seen this "Running (secure)" myself, and in that case I've been unable to debug code on the A53.  In my case this happens only when I power-cycle the board while the processor is halted via the debugger, so this might be something else.

I haven't used the "Basic Setup" for quite some time, but I can give it a try tomorrow in the office.

You should see debug output in the console view (maybe you need to check different console subviews - there's a dropdown button somewhere for this) when you connect via the basic setup - are there any errors/warnings shown?

Regards,

Dominic

Hi,

John Hoffman said:

Then I connect to Cortex A53_0_0 and set it to running. 

If you are following the exact sequence of events you posted, the step highlighed above is a problem. You can't set the processor to run and then set breakpoints, load code, etc. 

The procedure sent by Dominic is sound and I was just able to run it to completion in my setup (both CCSv9.3 and 10.0).

Hope this helps,

Rafael

I started from section 9.9 instead of section 9.1.

I have restarted now from section 9.1 and have been able to get a little further.

I am able to load the project and step through the code (a little bit).

But, I am not getting anything on the serial terminal.

When using the UART connection, there are 4 ports that are created. Is it always the first one listed that is used?

It appears as though the program goes into Board_init(boardCfg) and never returns.

Hi,

John Hoffman said:

But, I am not getting anything on the serial terminal.

I am a bit confused. The Hello world project above does not send anything through UART port. Perhaps another project?

John Hoffman said:

When using the UART connection, there are 4 ports that are created. Is it always the first one listed that is used?

The COM port selection would be heavily dependent on the project itself. Just keep in mind the built in XDS110 Debug Probe also has an additional UART port. 

I am building the projects of the SDK to see if there is a pre-canned UART project and will report back to this thread. 

Regards,

Rafael 

I am now stepping through the initialization part of the source code from the 9.1.4 Lab in the TI SDK RTOS How to Guide.

In the am65xx_evm.c source file, there is function called :

Board_STATUS Board_init(Board_initCfg cfg)

Inside this file, there is a call to :

ret = Board_moduleClockInit();

This is located in board_clock.c

This then goes into the following for loop and never returns:

for(index = 0; index < loopCount; index++)

{

status = Board_moduleClockEnable(gBoardClkModuleID[index]);

if(status != BOARD_SOK)

{

return BOARD_INIT_CLOCK_FAIL;

}

}

Any thoughts on what is causing this to hang up?

Yes, you have gotten to the same point that I have.

The debugger does not help in this case.

The Sciclient-related code pertains to DMSC.  Are you loading the DMSC firmware:

http://software-dl.ti.com/processor-sdk-rtos/esd/docs/latest/rtos/index_how_to_guides.html#advanced-am65x-debug-setup-with-dmsc-firmware-load

Yes, that was loaded the first time I attempted to connect to the DMSC.

Brad,

Thanks for the heads up. I looked at the javascript and did the steps manually, just to be sure I fully understood the process. It worked for me as it can be seen on the short clip below:

Hope this helps,

Rafael

Rafael -- that's great!  Thanks for recording the video.

John -- are you able to replicate what Rafael did?  We can work on optimizing the flow once you have it working.

John -- can you please update on your current status?  Were you able to reproduce what Rafael did?

Rafael,

Thanks for the video.

I am out of the office until Thursday, but I still may be able try the sequence from the video before then.

Thanks

John