TMS320F28379D: Trying to run two CPUs in stand alone mode

Hello,

Ehsan Qiyassi said:

on the card A;SW1 is on '10' position. Position 1 is ON and position 2 is off. This allows debugging using XDS100V2 onboard emulator.

From your boot flow, it sounds like you're trying to execute a standalone boot to flash? If so, you need A:SW1 position 1 to be OFF.

Please follow the controlCard user's guide for configuring the device for a standalone boot from Flash:

Can you also share a picture of your controlCard switches?

Best,

Matt

Hi Matt,

Thanks for the response. I'd like to be able to execute the program From the flash (Is it different from booting To flash?). I imagined the steps to be: 1) program the flash through XDS100V2; 2) Run the program from the flash

I do turn A:SW1 off after I program the processor. (as described in step 6-a)

I have also tried leaving it as is and just unplug the cable. This is what I meant by saying trying all combinations (step 6-c)

Here's a photo of the control card with the A:SW1, position 1, in ON position:

Again, I change the position 1 of the A:SW1 to 'off' after the board is programmed through XDS100V2.

Hi,

Ehsan Qiyassi said:

I'd like to be able to execute the program From the flash (Is it different from booting To flash?)

Apologies, boot to flash == boot from flash

I replicated your controlCard switch configuration on my setup and was able to get an LED blink example working. Can you confirm the location of the codestart for the application? It must be at 0x80000 for the device to boot the correct flash entry point.

Best,

Matt

Hi Matt,

Thanks for looking into this. codestart for both CPUs is at "00080000". At least according to the map files.

Is there anywhere else I should check?

Hi,

Can you try this procedure with the simple LED blink example in C2000Ware (C2000Ware_5_04_00_00\driverlib\f2837xd\examples\cpu1\led)? I want to rule out the possibility of application error.

Best,

Matt

Sure, will do and keep you posted. What kind of application error could cause this to happen? I'll have to look into that as well.

Matt,

I tried an LED project from the examples and I only had to make it to program to flash instead of ram, and it seems to work even without flipping A:SW1, position 1, to off or even disconnecting the cable.

Any suggestions on where to look next?

Hi Ehsan,

I recommend emulating a boot to flash by configuring the emulation boot registers (EMU_BOOTCTRL). By emulating boot to flash, you can view the current state of the device and debug why the application isn't working.

You can do this easily by connecting to the device and navigating to Scripts > EMU Boot Mode Select > EMU_BOOT_FLASH. Then, reset the CPU and the device will emulate boot to flash. You can load the symbols for the application to further debug.

Best,

Matt

Hi Matt,

Thanks for the suggestion.
I followed the steps. After cpu reset, program counter points to _System_post_cinit() (@0x08369) which I believe is past the codestart. I stepped over the assembly instructions which seemed to loop. I checked the value of "CPU1 EMU_BOOTCTRL Register" through the memeory browser and it seems to have the expected value:

0x0B5A is equivalent to EMU_KEY of 0x5A and EMU_BMODE of 0x0B :

Even though stepping through the assembly code did not seem to progress, Resuming it, will work. which at any rate tells me after reset I have been past any problematic symbol.

Did I understand your suggestion correctly? What are your thoughts?

Hi,

Can you try loading the symbols (i.e. the *.out file) for your application? When you reset the device, you will be executing in the boot ROM. You can then set a break point in your application to resume debugging from there.

Best,
Matt

Hi Matt,

I tried various permutation of connecting to the device, loading the symbols and resetting the CPU (from debug menu and power cycle) but I'm not sure what to expect from taking these steps. Eventually all these permutations bring me to the beginning of the main() which runs fine and had always run fine.

To reiterate, my only problem is that the program doesn't run after power cycle, without reconnecting the emulator. i.e. the program itself doesn't seem to have any issues. I am using SysConfig for creating the linker file

Could you please elaborate what is the expected outcome of following these steps? Maybe I'm skipping over something without knowing it.

Would that help if I upload the .out or .map files for you?

Hi,

Ehsan Qiyassi said:

To reiterate, my only problem is that the program doesn't run after power cycle, without reconnecting the emulator. i.e. the program itself doesn't seem to have any issues. I am using SysConfig for creating the linker file

I'm providing these steps so you can emulate a boot to flash to debug why the device isn't able to run the program after a power cycle. These steps take you from device reset all the way to running the application. It's unusual that you're able to run an LED blink example after power cycling with the same external switch configuration, which had me suspect something isn't right with the application.

If you could share the .out and .map files, that would be fantastic.

Best,

Matt

Thanks,

I've uploaded the files here:
https://windliftllc-my.sharepoint.com/personal/ehsan_windliftllc_onmicrosoft_com/_layouts/15/onedrive.aspx?id=%2Fpersonal%2Fehsan%5Fwindliftllc%5Fonmicrosoft%5Fcom%2FDocuments%2FStandaloneBoot&ga=1

If you wanted to see the application working, there need to be two uart terminals connected to GPIO22/23 (TX/RX) and GPIO28/29 (RX/TX), with these settings:

BR = 115200 ;

word length = 8 ;

stop mode = 1;

no parity.

To reach to that stage, both CPUs need to be running because they Sync at one point.

Hi Ehsan,

I'm happy to hear that the resolution was simple! Thank you for updating the thread with the solution.

Best,

Matt