topic Re: Saving struct to flash memory using FlashProgram in Kinetis Microcontrollers

Saving struct to flash memory using FlashProgram

piotrcerba — Sun, 25 Oct 2015 18:46:39 GMT

Hello, I have a following problem. I want to save my own struct using the flash_demo example from SDK.

I am using FlashProgram function for that purpose.

What have I tried:

Here is my code (the rest of SDK flash_demo remains unchanged):

typedef struct
{
  int x, y;
  char name[40];
} myStruct;
int main(void)
{
/*rest of example code*/
myStruct first = { .x = 1, .y = 123, .name = "George" };
/*rest of example code*/
   /********************************************************************
    *   For devices without SWAP, program some data for demo purposes
    *********************************************************************/
    destination = flashSSDConfig.PFlashBase + (flashSSDConfig.PFlashSize - 6*FTFx_PSECTOR_SIZE);
    end = flashSSDConfig.PFlashBase + (flashSSDConfig.PFlashSize - 4*FTFx_PSECTOR_SIZE);
    /*for (i = 0; i < BUFFER_SIZE_BYTE; i++)
    {
         Set source buffer
        program_buffer[i] = i + 100;
    }*/
    size = sizeof (myStruct);//BUFFER_SIZE_BYTE;
    /* message for user */
    PRINTF("\n\n\r---->Running FlashProgram() and FlashProgramCheck()...");
    while ((destination + (size)) < end)
    {
        ret = FlashProgram(&flashSSDConfig, destination, size, \
                                       first, g_FlashLaunchCommand);
        if (FTFx_OK != ret)
        {
            ErrorTrap(ret);
        }
/*rest of example code*/

Unfortunately this gives an error:

D:/Freescale/SDK/examples/frdmkl43z/demo_apps/flash_demo/main.c:354:9: error: incompatible type for argument 4 of 'FlashProgram'

1. Does it mean I am not able to save struct in flash memory? (somewhat contrary to what Mike Butcher said here -> Reading flash - why not by normal memory access? )

2. Is there any way to transform any structure to uint8_t* which is required by FlashProgram?

I will be very grateful for any suggestions!

Re: Saving struct to flash memory using FlashProgram

bobpaddock — Mon, 26 Oct 2015 12:40:04 GMT

I don't know what FlashProgram wants. Would it want &first?

However using structures in this way is not recommended due to how the compiler might pack the structures.

A structure may have hidden bytes to maintain alignment. ARM natively normally aligns to four byte boundaries.

offsetof() in <stddef.h> can be helpful in (de)serializing structures for storage and transmission.

Re: Saving struct to flash memory using FlashProgram

Jorge_Gonzalez — Mon, 02 Nov 2020 13:19:47 GMT

Hello Piotr Cerba:

Bob Paddock is correct, the FlashProgram() function of the driver expects a pointer to a raw array of bytes, not a structure. You can cast the parameter to match the uint8_t* expected by the driver, like this:

FlashProgram(&flashSSDConfig, destAdrss, sizeof(first), (uint8_t*)&first, g_FlashLaunchCommand);

This is what I get in KDS with GCC using your code (except that I used address 0x80000) and considering that the ARM Cortex-M core in Kinetis uses little endianness:

Since the structure is stored in Flash in the same order as the variable in RAM, then you can retrieve the data from flash by using a simple pointer to the same kind of structure, something like this:

typedef struct
{
    int x, y;
    char name[40];
} myStruct, *myStructPtr;
//...
myStruct first = { .custom = "ON", .x = 1, .y = 123, .name = "George" };
myStructPtr firstPtr = (myStructPtr)0x80000;
//...
destAdrss = 0x80000;
FlashProgram(&flashSSDConfig, destAdrss, sizeof(first), (uint8_t*)&first, g_FlashLaunchCommand);
//Erasing "first"
first.x = 0;
first.y = 0;
//Recovering "first"
first.x = firstPtr -> x;
first.y = firstPtr -> y;

:smileyalert: Note that the core processor is 32-bit (4 bytes), but the smallest programming size for the flash memory controller in K64 (FTFE) is 8-bytes (phrase). Therefore the flash driver expects that the size parameter is a multiple of 8 and the dest address parameter is aligned to an 8-bytes boundary. The size of your structure is 40 + 4 + 4 = 48 bytes, which is good, but if you add more members and the size is not a multiple of 8, then you may need to add padding bytes, considering that the compiler may add its own padding bytes to a member which is not 4-bytes long, as mentioned by Bob.

I hope this information is helpful.

Regards!,
Jorge Gonzalez

-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------