SparkFun Forums

[SteveC]

I have a small but annoying issue when using the USB JTAG Tiny with an Olimex P149 Proto board.

I have an application where the bulk of the code resides in Flash @ 0x7000, but has an algorithm currently located in RAM @ 0x200 because it's continually evolving and I'm simply saving Flash wear cycles.

The issue is whenever code is changed resulting in different code length the first JTAG download fails to verify because the memory word at 0x200 is overwritten with the value 0x200 instead of the mov #0A00h,sp instruction.
Cancelling the Kickstart debug session and restarting it always succeeds.

The first debug download of the day always succeeds.

If the code is unchanged in length the download always succeeds even if the code is different.

Can anybody shed some light on things?

Host System: Win XP
Kickstart: v3.40.1.9
JTAG Tiny drivers: 1.011
2.0m USB cable
Debugger options: Verify download & Retain Memory
--

[OLIMEX]

Correctme if I'm wrong but the Option Retain memory will not do Erase? so why do you wonder that Verify fails when code is changing?

Best regards
Tsvetan

[SteveC]

Correctme if I'm wrong but the Option Retain memory will not do Erase? so why do you wonder that Verify fails when code is changing?

No, that option only erases IF the code has changed. It will still change & rewrite memory if necessary. It's only intended to save unnecessary FLASH erase/write cycles. Strictly it's called "Retain unchanged memory".

In any case the error is that a value, coincidentally 0x200 is erroneously written to address 0x200. That value never changes in the code, it is ALWAYS a stack initialisation instruction.

[OLIMEX]

can you send your project to us to test?
Tsvetan

[SteveC]

Thanks - I'll try and distill a very simple example that displays the same behaviour and send you that.

[SteveC]

OK - I have now distilled a very simple project that demonstrates the problem. The fact most of the code was in FLASH and some is in RAM doesn't seem relevant.

I've isolated the problem to when one of the interrupt vectors points to code in RAM that has moved since the last JTAG download. Very strange indeed and possibly points to an issue with Kickstart.

If the 'nop' at label 'stackclr' is alternately commented out then not between each compilation/download then I get the download verification error every time. Again address 0x200 is overwritten with value 0x200. The next download without changing any code always succeeds.

I can zip up the whole project and send it to you - to what address should I send it?

Code: Select all

#include  "msp430x14x.h"

;******************************************************************************
            RSEG    CSTACK                  ; FORWARD declaration of stack segment

            RSEG    RAM                     ; Program Start
;------------------------------------------------------------------------------
main
start       mov.w   #sfe(CSTACK),SP         ; Initialize stackpointer
StopWDT     mov.w   #WDTPW+WDTHOLD, &WDTCTL ; Stop WDT
            mov     #sfb(CSTACK), r12
stackclr    nop
            jmp     $


TAX_ISR     add     &TAIV, PC               ; Add Timer_A offset vector
            reti                            ; CCR0 - no source
            jmp     CCR1_ISR                ; CCR1
            jmp     CCR2_ISR                ; CCR2
            reti                            ; CCR3
            reti                            ; CCR4
TA_over     xor.b   #001h,&P1OUT            ; Toggle P1.0
            reti                            ; Return from overflow ISR		

CCR1_ISR    XOR     #OUTMOD2, &TACCTL1      ;toggle set/reset next o/p
            BIT     #OUTMOD2, &TACCTL1
;           SNIP

;TACCR2 irq handler
CCR2_ISR    MOV     &TACCR2, r4             ;read capture value from serial pulse stream
;           SNIP

;------------------------------------------------------------------------------
;           Interrupt Vectors Used - MSP430F149
;------------------------------------------------------------------------------
            RSEG    INTVEC:CONST

            ORG     TIMERA1_VECTOR
            DW      TAX_ISR

            ORG     NMI_VECTOR              ;
            DW      start                   ; MSP430 NMI Vector
            DW      start                   ; MSP430 RST Vector

            END

[OLIMEX]

Hi,
please e-mail (info at olimex dot com) your complete project (including your EW project files) only from this assembly code we can't reproduce your result
Tsvetan

[OldCow]

I tried your code, but I got a XLINK error. RSEG RAM is not defined.

Can you show me how your xlink command file look like? The one I use (lnk430f149.xcl) look like this:

• // -------------------------------------------------------------------
  // RAM memory
  // -------------------------------------------------------------------
  
  -Z(DATA)DATA16_I,DATA16_Z,DATA16_N,HEAP+_HEAP_SIZE=0800-09FF
  -Z(DATA)CSTACK+_STACK_SIZE#
  
  // -------------------------------------------------------------------
  // Information memory (FLASH)
  // -------------------------------------------------------------------
  
  -Z(CODE)INFO=1000-10FF
  -Z(CODE)INFOA=1080-10FF
  -Z(CODE)INFOB=1000-107F
  
  
  // -------------------------------------------------------------------
  // ROM memory (FLASH)
  // -------------------------------------------------------------------
  
  -Z(CODE)RAM=0200-07FF
  -Z(CODE)CSTART=1100-FFDF
  -Z(CODE)CODE=1100-FFDF
  
  // Constant data
  
  -Z(CONST)DATA16_C,DATA16_ID,DIFUNCT,CHECKSUM=1100-FFDF
  
  // Interrupt vectors
  
  -Z(CONST)INTVEC=FFE0-FFFF
  -Z(CONST)RESET=FFFE-FFFF

[OldCow]

I tried your code, but I got a XLINK error. RSEG RAM is not defined.

I added a -Z command to the xlc file and use TI parallel port FET tool to test your code. The result is the same as yours.

[SteveC]

Hi OldCow,

I use a custom link file...

Code: Select all

//*****************************************************************
//  XLINK command file for the MSP430 IAR C/C++ Compiler
//
//  This is the XLINK command file for the MSP430F149
//  microprocessor.
//
//  Copyright 1996-2004 IAR Systems. All rights reserved.
//
//  Usage:  xlink  your_file(s)  -f lnkMSP430F149  cl430xxx
//
//  $Revision: 1.0 $
//
//*****************************************************************


//*****************************************************************
//  The following segments are defined in this linker command file:
//
//  Data read/write segments (RAM)
//  ==============================
//
//  segment     address range   usage
//  -------     -------------   --------------------------
//  DATA16_I    0200-09FF       Initialized variables
//  DATA16_Z    0200-09FF       Zero initialized variables
//  DATA16_N    0200-09FF       Uninitialized variables
//  CSTACK      0200-09FF       Run-time stack/auto variables
//  HEAP        0200-09FF       The heap used by malloc and free
//
//
//  Program and non-volatile segments (FLASH)
//  =========================================
//
//  segment     address range   usage
//  -------     -------------   --------------------------
//  INFO        1000-10FF       Information memory
//  CSTART      1100-FFDF       cstartup program code
//  CODE        1100-FFDF       Program code
//  DATA16_C    1100-FFDF       Constant "const" variables AND String literals
//  DATA16_ID   1100-FFDF       Initializers for DATA16_I
//  DIFUNCT     1100-FFDF       Dynamic initialization vector used by C++
//  CHECKSUM    1100-FFDF       The linker places the checksum byte(s) in this segment,
//                              when the -J linker command line option is used.
//
//  INTVEC      FFE0-FFFF       Interrupt vectors
//
//  NOTE:
//  It is not possible to pack the CSTART segment by using the XLINK -P option
//  Special function registers and peripheral modules occupy addresses 0-01FFh
//  Be sure to use end values for the defined addresses
//*****************************************************************

// -------------------------------------------------------------------
// Stack size and heap size
// -------------------------------------------------------------------

// Uncomment for command line use
//-D_STACK_SIZE=50
//-D_HEAP_SIZE=50

// -------------------------------------------------------------------
// Define CPU
// -------------------------------------------------------------------

-cmsp430

// -------------------------------------------------------------------
// RAM memory
// -------------------------------------------------------------------

-Z(CODE)RAM=0200-09FF
-Z(DATA)DATA16_I,DATA16_Z,DATA16_N,DATA,HEAP+_HEAP_SIZE=0200-09FF
-Z(DATA)CSTACK+_STACK_SIZE#

// -------------------------------------------------------------------
//  Information memory (FLASH)
// -------------------------------------------------------------------

-Z(CODE)INFO=1000-10FF
-Z(CODE)INFOA=1080-10FF
-Z(CODE)INFOB=1000-107F

// -------------------------------------------------------------------
// ROM memory (FLASH)
// -------------------------------------------------------------------

//  Code

-Z(CODE)BOOT=1000-11FF
-Z(CODE)LANG=3E00-6FFF
-Z(CODE)CSTART=7000-FFDF
-Z(CODE)CODE=7000-FFDF

// Constant data

-Z(CONST)DATA16_C,DATA16_ID,DIFUNCT,CHECKSUM=7000-FFDF


// Interrupt vectors

-Z(CONST)INTVEC=FFE0-FFFF
-Z(CONST)RESET=FFFE-FFFF


// -------------------------------------------------------------------
// End of File
// -------------------------------------------------------------------

[SteveC]

I tried your code, but I got a XLINK error. RSEG RAM is not defined. I added a -Z command to the xlc file and use TI parallel port FET tool to test your code. The result is the same as yours.

Link file posted above. You mean you get a verify error on download? If so then it's pretty strange, but at least it's consistent and presumably points to a Kickstart error rather than the USB Tiny JTAG.

I'll still mail the project to Tsvetan in case he has some ideas. Thanks for the help so far guys.

[SteveC]

Tsvetan - I've E-Mailed you the zipped project, thanks again for your help...

[SteveC]

I've had a reply from Tsvetan at Olimex about this issue and he's confirmed that he gets the same faulty result with a range of JTAG adaptors for my example project.

So the fault lies elsewhere and not with JTAG Tiny. I just wanted to make it clear.

Thanks for the support guys.