CCS/TMS570LS0714: F021 library for use with TMS570LS0714

Hello Joe,

1. TMS570LS0714 and TMS570LS1224 use the same F021 API library: F021_API_CortexR4_BE_V3D16.lib

2. TMS570LS0714 flash has only 11 sectors, but the flash on LS1224 has 15 sectors. 

QJ;

Thank you for the response.

The issue we appear to be having is that when we call the Fapi_issueProgrammingCommand() API command, the result is 48 (0x3000) which is FMSTAT.INVDAT, invalid data.

INVDAT states: "When set, this bit indicates that the user attempted to program a 1 where a 0 was already present."  However, before issuing this command we do a blank check which shows that the area which we wish to program is all 1.

Furthermore, if we step through the code during debugging using CCS, this Fapi_issueProgrammingCommand() succeeds on the 714.

Also, as stated earlier, this code has been working perfectly on the 1224 (with or without debug mode) for years.

Here a sanitized version of the code in question.

// for this operation, dst = 0x10000

Flash_Erase_Check(dst, 0x10);  // start address, length, checks for 0xFF in each byte

Fapi_initializeFlashBanks(SYS_CLK_FREQ); 

Fapi_setActiveFlashBank((Fapi_FlashBankType)0);

Fapi_enableMainBankSectors(0xFFFF);

Fapi_issueProgrammingCommand((uint32_t *)dst, (uint8_t *)buffer, (uint32_t)size, 0, 0, Fapi_AutoEccGeneration);

// at this point we read the FAPI status and get 48 on the 714

Any additional insight you might have on the issue would be appreciated.

Thanks in advance,
Joe Shidle

Joe,

Since LS0714 has only 1 flash bank, you need to run the flash AI library from SRAM instead of flash. Did you copy the code to SRAM and execute the code from SRAM?

[REDACTED]

QJ;

Thanks again for your response.

The code in question is the SafetyMCU_Bootloader code from TI and it appears that in the sys_startup.c function _c_int00() there is a call to _copyAPI2RAM_() which, in sys_core.asm appears to "copy the Flash API from flash to SRAM"

So yes, it appears that we are copying and executing the FAPI code from SRAM.

Is there anything else?

Thanks in advance,
Joe Shidle

Hello Joe,

When you initialize the flash bank, did you check if the bank is initialized successfully?

if ((Fapi_initializeFlashBanks((uint32_t)SYS_CLK_FREQ)) == Fapi_Status_Success){

       (void)Fapi_setActiveFlashBank((Fapi_FlashBankType)0);
       (void)Fapi_enableMainBankSectors(0xFFFF); 
}else {
       return (1);
}

What happens if you erase the sector again using F021 flash API before programming data?

Hello QJ;

Thank you for the response.

QJ Wang said:

When you initialize the flash bank, did you check if the bank is initialized successfully?

We do not, as we copy code from TI examples in the original F021 API software bundle and they do not check return values there. 

When I add code to check the return value of the Fapi_initializeFlashBanks(), I receive return value of 8 which per the reference manual is FMSTAT.VOLTSTAT which is a core voltage status.

However please note two things:

1) I receive this same FMSTAT.VOLTSTAT when I am running the same code through the debugger, and the later flash program succeeds.

2) I added check of the return value for the Fapi_setActiveFlashBank() and Fapi_enableMainBankSectors() and both return status = 0 which is success, for both cases of running in debugger when flash write succeeds and running on processor when flash write fails.

So I am skeptical that the VOLTSTAT is a concern, but I am open to suggestions.

QJ Wang said:

What happens if you erase the sector again using F021 flash API before programming data?

When I erase the sector prior to programming the data:

1) Everything is successful (erase, program) when I run through the debugger.

2) The chip resets during the erase sector, when I run it normally on the device.

Thanks in advance for your further help,

Joe Shidle