SLAZ160J October   2012  – May 2021 MSP430F2121

 

  1. 1Functional Advisories
  2. 2Preprogrammed Software Advisories
  3. 3Debug Only Advisories
  4. 4Fixed by Compiler Advisories
  5. 5Nomenclature, Package Symbolization, and Revision Identification
    1. 5.1 Device Nomenclature
    2. 5.2 Package Markings
      1.      DGV20
      2.      PW20
      3.      DW20
      4.      RGE24
    3. 5.3 Memory-Mapped Hardware Revision (TLV Structure)
  6. 6Advisory Descriptions
    1. 6.1  BCL6
    2. 6.2  BCL8
    3. 6.3  BCL9
    4. 6.4  BCL10
    5. 6.5  BCL11
    6. 6.6  BCL12
    7. 6.7  BCL13
    8. 6.8  BCL14
    9. 6.9  BSL5
    10. 6.10 CPU4
    11. 6.11 CPU5
    12. 6.12 CPU6
    13. 6.13 CPU11
    14. 6.14 CPU12
    15. 6.15 CPU13
    16. 6.16 CPU14
    17. 6.17 CPU19
    18. 6.18 CPU45
    19. 6.19 EEM20
    20. 6.20 FLASH16
    21. 6.21 FLASH17
    22. 6.22 FLASH18
    23. 6.23 FLASH19
    24. 6.24 FLASH20
    25. 6.25 FLASH22
    26. 6.26 FLASH24
    27. 6.27 FLASH27
    28. 6.28 FLASH36
    29. 6.29 JTAG15
    30. 6.30 PORT8
    31. 6.31 PORT10
    32. 6.32 SYS15
    33. 6.33 TA12
    34. 6.34 TA16
    35. 6.35 TA21
    36. 6.36 TAB22
    37. 6.37 XOSC5
    38. 6.38 XOSC8
  7. 7Revision History

CPU6

CPU Module

Category

Compiler-Fixed

Function

Incorrect execution of ADD instruction

Description

When the CPU executes an ADD (.B or .W) instruction using indirect addressing mode with destination R1 or R4 to R15, directly after a RET or RETI instruction, the addition will be executed twice. This bug does not apply in the case that the indirect source working register is R2 or R3 (constant generator access for #2 or #4 are valid).

The instruction word mask that corresponds to the CPU6 condition is as follows:



GUID-20201119-CA0I-ZZC5-PDGK-NR876BSZ4MNL-low.png

Workaround

1. (Preferred) Use indexed addressing with offset 0 instead of indirect addressing:
ADD(.B/.W)   0(R10),R11

Note that some assemblers may convert this instruction to indirect addressing.


Alternatively, when the source is in assembly, making the following changes works around the bug. Perform the following only if Workaround 1 cannot be implemented.

or

2. Place an instruction (for example NOP) between any instances of CALL and ADD(.B/.W) instructions where the ADD immediately follows the CALL.

or

3. Replace all RETI instructions with 'DW 01301h'. The instruction word for RETI is 01300h. By replacing this directly in assembly with 01301h, the bug is avoided and the RETI instruction is executed properly; for example:

WDT_ISR                    ; Exit LPM3 on reti
bic.w   #LPM3,0(SP) ;
;reti                   ; Replace RETI with
DW      01301h          ; this line of code

Refer to the table below for compiler-specific fix implementation information.

IDE/Compiler Version Number Notes
IAR Embedded Workbench IAR EW430 v4.x or later User is required to add the compiler flag option below. --hw_workaround=CPU6 Add -h option for Assembler
TI MSP430 Compiler Tools (Code Composer Studio) Fix not available
MSP430 GNU Compiler (MSP430-GCC) Fix not available