Experiencing strange behaviour of Fapi_doBlankCheck

René K.

Other Parts Discussed in Thread: TMDX570LC43HDK, HALCOGEN

Hi @all,

while developing a firmware that reads/writes calibration values from/to eeprom I got some strange error. I used the same functions for a RM48 implementation but now, while porting the SW to a TMS570LC43xx, this error popped up:

After perfoming a Fapi_doBlankCheck the flashbank 7 appears to be empty (0xFFFFFFFF) even though it is still containing old calibration data. Watching the memory content with the memory browser in CCS shows the calibration data before calling Fapi_doBlankCheck and afterwards the memory seems to be blank...

I am using:

TMDX570LC43HDK

Code Composer Studio Version: 6.0.1.00040

HALCoGen Version: 04.04.00

FLASH API Version: 02.01.01

System configuration:

HCLK 150 MHz

FAPI is running in RAM, copied by copytable

MPU/Cache disabled in HALCoGen

Defined macro _L2FMC in CCS project properties

Short implementation in HL_sys_main.c:

    Fapi_StatusType            f021_Res;
   Fapi_FlashStatusWordType    f021_State;
   uint32                       *f021_Address;

   copy_in(&_flash_ctbl);

   f021_Res = Fapi_initializeFlashBanks( 150 );
       if ( f021_Res != Fapi_Status_Success ){
           DOUT("Unable to initialize EEPROM support!");

       }else{
           DOUT("EEPROM support initialized...");

       }
       //
       f021_Res   =   Fapi_setActiveFlashBank( Fapi_FlashBank7 );
       if ( f021_Res != Fapi_Status_Success ){
           DOUT("Unable to activate flash bank!");

       }

       while( FAPI_CHECK_FSM_READY_BUSY != Fapi_Status_FsmReady );

       f021_Res   =   Fapi_enableEepromBankSectors( 0xFFFFFFFF, 0xFFFFFFFF );

       if ( f021_Res != Fapi_Status_Success ){
           DOUT("Unable to enable sectors!");

       }

       while( FAPI_CHECK_FSM_READY_BUSY != Fapi_Status_FsmReady );

       f021_Address    = (uint32 *) 0xF0200000;

       f021_State.au32StatusWord[3]   =   0;   //   prepare function answer
       Fapi_disableAutoEccCalculation();
       f021_Res                        =   Fapi_doBlankCheck( f021_Address, 2, &f021_State );

       while( 1 );

Can anybody give me a hint, where to look for the reason?

Thanks in advance...

over 10 years ago

0 Vishwanath Reddy over 10 years ago

TI__Mastermind 19365 points

Hi Rene,

Similar post is available on forum.

https://e2e.ti.com/support/microcontrollers/hercules/f/312/t/424082

MPU settings were updated in HALCoGen 04.05.00.

If you still have questions, please come back.

0 René K. over 9 years ago in reply to Vishwanath Reddy

Intellectual 280 points

Hi Vishwanath,

I read the mentioned post and recognized similarities but unfortunately an update to HALCoGen 04.05.00/regenerated/rebuild did not solve my issue.

5518.20150910_FLASH_Test_HALCoGen_04_05.zip

0 Vishwanath Reddy over 9 years ago in reply to René K.

TI__Mastermind 19365 points

Hi Rene,
I have taken your project and added below code after line f021_Address = (uint32 *) 0xF0200000;
uint8_t Data[]={0,1,2,3,4,5,6,7,8,9};
Fapi_issueProgrammingCommand(f021_Address,Data,8,0,0,Fapi_DataOnly);
I was able to see data in Bank 7 even after blankcheck API is executed.

0 René K. over 9 years ago in reply to Vishwanath Reddy

Intellectual 280 points

Hi Vishwanath,

thank you for your answer. Yes, I can reproduce the behaviour you described but isn't it useless to program memory to do a doBlankCheck afterwards? Sorry, maybe I don't understand your suggestion, could you explain in detail the use case for your code extension?

Thanks in advance...

Regards,
René

0 Vishwanath Reddy over 9 years ago in reply to René K.

TI__Mastermind 19365 points

Hi Rene,I thought you are observing following:
-Calbration data is available in Bank7 and visible through CCS memory browser
-Application calls Fapi_doBlankCheck -CCS memory browser shows all F's.
My intention was to see if after blank check is executed, does CCS still shows the data in bank 7 through memory browser.

0 René K. over 9 years ago in reply to Vishwanath Reddy

Intellectual 280 points

Ok, now I got it. Here some Screen shot of the result. Please watch the break Point stepping indicator on left side

After programming the Memory contains an invaild data word:

Call of doBlankCheck "clears" Memory beyond the written word:

0 Vishwanath Reddy over 9 years ago in reply to René K.

TI__Mastermind 19365 points

Hi Rene,
For TMS570LC43xx device, ECC is enabled by default. The random data you see in ccs memory browser is because of ECC errors(ECC area for corresponding Bank7 is also in erased state ). When blank check API is called, it enters a special mode and ignores all ECC errors for Bank 7 and hence you see all F's. Blank check API does not erase bank 7.

0 René K. over 9 years ago in reply to Vishwanath Reddy

Intellectual 280 points

Hi,
ok, that's seems to be plausible but when you are talking about ECC Errors you mean physical errors in memory, right?
That's strange because I am using a new HDK with only a few Flash write/erase cycles... What could be the reason for that?

Thanks for your efforts!

0 Vishwanath Reddy over 9 years ago in reply to René K.

TI__Mastermind 19365 points

Hi Rene,

The erased flash on the LC4357 does not read all 0xFFFFFFFF because the ECC is always enabled. The ECC equation includes a number of address bits and at some addresses the combination of all '1's in the flash and ECC plus the address bits is viewed as a correctable error. The corrected value is what you see in CCS and it's not 0xFFFFFFFF. This is weird looking but it's the way the part is designed so just ignore it.

0 René K. over 9 years ago in reply to Vishwanath Reddy

Intellectual 280 points

Thanks a lot!

Regards,

René

0 René K. over 9 years ago in reply to René K.

Intellectual 280 points

Maybe this is helpful for anyone that runs into the same problem:

www.ti.com/.../litabsmultiplefilelist.tsp

0 René K. over 9 years ago in reply to René K.

Intellectual 280 points

Please use "Fapi_issueProgrammingCommand(f021_Address,Data,8,0,0,Fapi_AutoEccGeneration); " instead of "Fapi_DataOnly" when using above source code... That will fix the problem with "wrong" information content... Hell knows how this got into the code...

Arm-based microcontrollers

Arm-based microcontrollers forum

Experiencing strange behaviour of Fapi_doBlankCheck