TMS570LS0914: FEE data corrupted

Part Number: TMS570LS0914
Other Parts Discussed in Thread: HALCOGEN

Hi Team,

I am working on a TMS570LS0914 based application with the following setup:

System Configuration

• MCU: TMS570LS0914
• Bootloader present
• Application start address: 0x00020000
• Tool: HALCoGen (TI FEE driver)
• FEE storage: Flash Bank 7
• FEE configuration:
  • Number of blocks: 16
  • Block size: 64 bytes
  • Block overhead: 24 bytes
  • Virtual sectors:
    • Total: 4
    • Per EEPROM: 2
  • Virtual page size: 8 bytes
  • Variable datasets: Enabled
  • Checksum: Enabled 

Observations

• Without bootloader
  • Application runs correctly
  • FEE read/write works as expected
  • ASW (application software) behavior is normal
• With bootloader (app at 0x20000)
  • BSW works correctly
  • FEE APIs execute without returning explicit errors
  • BUT ASW behavior becomes incorrect
• Key Finding
  • When I disable all NVM/FEE-related parameters, the ASW works correctly
  • This indicates the issue is related to FEE data read/write affecting ASW variables
• Critical Observation (Block-specific behavior)
  • FEE has 16 blocks (64 bytes each)
  • Data is used in multiple blocks (e.g., Block 5 and Block 6) 

 Case 1:Using Block 6 only → ASW works correctly

 Case 2:Using both Block 5 and Block 6 → ASW fails

 Case 3:Disabling Block 5 → ASW works again 

Analysis Done

• Verified linker configuration:
  • Application located at 0x00020000
  • No obvious RAM overlap
• FEE data stored in Bank 7
• Behavior suggests:
  • No direct crash or exception
  • Possible data corruption during FEE read
• Issue seems position/layout dependent, not specific to block content 

  Questions

1. Could this be related to:
   • FEE block packing across virtual sector boundaries?
   • Alignment issues due to 8-byte write granularity?
2. Is there any known issue when:
   • Application is relocated (bootloader case)
   • FEE layout remains unchanged?
3. Does FEE require:
   • Additional free space (beyond configured blocks) for internal operations like garbage collection
4. Could virtual sector size be insufficient leading to:
   • corruption when additional blocks (e.g., Block 5) are enabled?
5. Any known constraints regarding:
   • block ordering
   • block alignment
   • or total usable capacity vs configured capacity? 

Regards,

Prajakta