TMS320F28388D: How to write a binary file into flash in my own application code?

Hi Zhou,

Regarding the address/data format for the hex output, please take a look at bottom of the read me file at C2000Ware_x_xx_xx_xx\utilities\flash_programmers\serial_flash_programmer\README.txt.

Yes, you can integrate the kernel code as needed - I will loop in our kernel expert to guide you.

Thanks and regards,

Vamsi

Hi Zhou, 

Yes, the sciLoadApplication function can be used to write the application into Flash, you can integrate it into your code. The readme for the serial_flash_programmer outlines how the binary file is formatted. 

Thanks

Anu

Dear Anu,

Yes, I have read the readme file, which is helpful for me to understand the .txt file structure.

But I am confronted with a delicate problem.

The kernel example is running on RAM, which is very OK to update the flash with new firmware.

Our application is running on FLASH, when it's running, there is always something in address 0x00080000,

and my new firmware should also be written to this address.

so when I initialize the FlashAPI, it always says error.

Specifically, the error always happens at following function.

oReturnCheck = Fapi_doBlankCheck((uint32 *)FirstBlockBaseAdd,
Sector64KB_u32length,
&oFlashStatusWord);

so it looks like that  " I can not write myself ".

Is there any workaround?

Thank you!

zhou ping

What error are you getting? Are you erasing the sector first, and are you passing in the start of the sector as the address?

Dear Anu,

When I write the flash, if the data buffer length is larger than 8, there is always error as follows,

 Fapi_Error_AsyncIncorrectDataBufferLength

What should I do if the length is larger than 8?

use a  for loop, and call the  Fapi_issueProgrammingCommand. again and again?

Thanks very much

Actually I have gone through the erase procedure successfully.

Zhou, 

Yes, you would need to call the Fapi_issueProgrammingCommand in a loop, the F2838x Flash API usage example provides an example on how to do this. 

Thanks

Anu

Dear Anu,

Actually most of the functions in the copyData() function is OK, I digged into the code, and the code is really awesome to write the specific data into specific address.

But when I integrate the copyData() function code into my application, a strange problem occured.

please see the attached picture.  The return value of my SPIgetWordData() function is assigned to a temp_short variable.

but the BlockHeader.BlockSize can never be assigned a value !

How can it be?

Thanks!

Perhaps the Disassembly view would be of help here.  Where is BlockHeader getting allocated?

Thanks

Anu

Dear Anu,

Thank you very much for your help!

I am really not good at reading the disassembly view.

Please see the below screenshot.

What can we say from the disassembly view?

Dear Anu,

Actually I use a big array to store the binary file, which is generated by a led example, just for demo.

I can write the memory now, but the written data is not totally consistent with the original data.

There are so many "0000 0000 0000 0000" inserted into the memory.

I use the same copyData() function in kernel project.

Could you please some clues?

Can you look in the map file to see where BlockHeader is allocated? And did you modify copyData to pad the data in any way? It could be helpful to step through the writing of those data blocks to see if this is the case. 

Thanks

Anu