TMS570LS1224: TMS570LS3137 EEPROM with FEE lib

Part Number: TMS570LS1224

Tool/software:

Hello,

Regarding FEE Virtual Sectors and block size allocation, I wanted to ask potential advice regarding this two points : 

1) Using a 64KB EEPROM (4 x 16 KB Flash sectors), do you have any recommendations using : 

Option 1 : 4x16 KB Virtual sectors

Option 2 : 2x32 KB ( with 2x16KB flash per Virtual sectors),

in case of course the block to be save in Virtual sectors doesn't exceed 16 Kbyte.

It seems that in Halogen configuration, the advise is to create Virtual sector composed of 2 flash sectors, only if the size of data block do not fit in 1 Virtual sector of minimum size (16Kb here).

2) Regarding the data block size, I personally see advantages to contain them to minimum size (8 Bytes of data) in case the data to save are "variables or parameters for software"

-because the FEE mechanisms to write and read them allow us to identify them (block number, easy to find)

-less write/erase of EEPROM if just a couple of data are changed in one bigger block

-Virtual Sector full slower and so les Virtual Sector change

-less decoding effort and less buffer sizes consumption in the application (For RAM cost)

But am I correct here ? And do you have some points to recommends for block sizes selections in Virtual Sectors ?

Thank you !

Regards,

Marc

  • Hi Marc,

    1) Using a 64KB EEPROM (4 x 16 KB Flash sectors), do you have any recommendations using : 

    I will suggest you use 2 virtual sectors only, because on our recent tastings we found an issue with FEE driver while using 4 virtual sectors. Please refer below thread to understand more details about this issue:

    (+) TMS570LS1224: FEE specific questions - Arm-based microcontrollers forum - Arm-based microcontrollers - TI E2E support forums

    On the first iteration the FEE is utilizing all the four virtual sectors but from second iteration onwards it is just using two virtual sectors only. This issue not yet rectified int he FEE driver, so please use the two virtual sectors only.

    But am I correct here ? And do you have some points to recommends for block sizes selections in Virtual Sectors ?

    You are correct it would be better to keep the data block size as multiples of 8.

    FEE user guide also suggesting the same.

    --
    Thanks & regards,
    Jagadish.

  • Hello again,

    Additional questions regarding the FEE and the size of the data blocks, I gave a try with configuring data block of 4 Bytes with Halogen (there is no restrictions provided by it) : 

    I am correctly writing and reading the blocks as well  :

    --> Block size  (header 24 Bytes + data 4 Bytes) : 28 Bytes / Block data only : 4 Bytes

    So even if it is written inside the manual smallest is 64 bits (8 Bytes), can we also set blocks of 32 bits (4 Bytes)  ?

    This would help my development.

    Thank you,

    Regards,

    Marc

  • Hi Marc,

    Apologies for the delay in the late response. I was stuck with other issues.

    Today i got chance to test this condition.

    It is not giving any error if you are using 4 bytes however in the background, it is assigning the 8bytes even if you configured block size as 4 bytes.

    For example, see the testing of below two configurations:

    In these tastings i configured two blocks with same size

    1. 4 bytes configuration:

    2. 8 bytes configuration:

    As you can see in the both the configurations the block-2 data was starting at same address. And you can also notice 0xFFFFFFFF after block -1 4 bytes data. So that means in the background it is assigning min size of block as 8 bytes.

    --
    Thanks & regards,
    Jagadish.

  • Hello Jagadish,

    Thank you for the feedback, so it is indeed possible but at the cost to EEPOM byte lost and so less capacity of saving.

    Thank you !

    Regards,

    Marc