MSP430FR5994: Issues with #pragma NOINIT #pragma PERSISTENT FRAM

In "Project->Properties->Debug->MSP430 Flash Settings->Erase Options", make sure the option "Erase and download necessary segments only" is checked.

I have tried that but its not that. Once the .cmd file is changed so the variables are in fram I can not change the values in my program even if disabling the MPU the variables are at address 4000 and 4004 must be something else

0x1E15 is in RAM, and 0x7B7A is not as close to 0x4000 as I would expect, so I think something else is going on.

My test case:

1) I started with this code:

#include <msp430.h> 

#pragma NOINIT(damping)
unsigned char damping;

#pragma PERSISTENT(factor)
float factor=1;
/**
 * main.c
 */
int main(void)
{
	WDTCTL = WDTPW | WDTHOLD;	// stop watchdog timer
	PM5CTL0 &= ~LOCKLPM5;       // Avoid warning
	++damping;
	++factor;
	while (1) {LPM0;}
	/*NOTREACHED*/
	return 0;
}

2) I made the .cmd change you mentioned:

        GROUP(READ_WRITE_MEMORY)
        {

            .TI.persistent : {}              /* For #pragma persistent            */
            .TI.noinit  : {} /*> RAM */      /* For #pragma noinit                */
            .cio           : {}              /* C I/O Buffer                      */
            .sysmem        : {}              /* Dynamic memory allocation area    */
        } PALIGN(0x0400), RUN_START(fram_rw_start)

3) I changed the Erase Options as above

4) I didn't do anything with the MPU (I let the compiler/linker figure it out)

I went through a few combinations of restart and re-download, and saw what I expected -- factor was re-initialized on each download, but retained its (incremented) value over a restart, and damping started out at some random value and kept incrementing (both cases).

I'm not sure what we're doing differently.

thanks address were from an earlier test they are 4000 and 4004, ok so must be something else in my code or a setting stopping it?

One possibility that comes to mind (not that *I* have ever done this) is that, after a series of experiments, there's some forgotten leftover setting that's getting in your way.

If feasible you might consider (a) creating a new project and copying your code into it and/or (b) following my test case above (it took me about 20 minutes) to see if you get the expected results.