Re: [avr-gcc-list] Link script question

[Dave N6NZ]

I would take very slightly different approach.  Make that a section of 
it's own, and put it in the linker script just after your exported symbol:

  __user_prog_start = .;
  *.(.userprog.*) ;

then create a simple .S

  .section .userprog
  .byte 0xff ; is this the correct directive? I forget
  .end

anyway, that should put the 0xff into .userprog, which I presume will
then be overwritten by your user data at a later time using the 
mechanism that you have currently defined.

-dave

Gre7g Luterman wrote:
> I've got an avrgcc application that uses free Flash blocks to hold user code. 
> To maximize the space available for the user, I start this area at the 
> beginning of the first block following my application code. To locate this, I 
> added the following to my linker script:
>
> --- mega32.orig 2008-10-09 17:14:10.921875000 -0600
> +++ mega32.ld   2008-10-09 12:48:58.281250000 -0600
> @@ -170,6 +170,10 @@
>       _edata = . ;
>       PROVIDE (__data_end = .) ;
>    }  > data
> +  /* User-programming space */
> +  . = _etext + SIZEOF(.data);
> +  . = ALIGN(128);
> +  __user_prog_start = .;
>    .bss  SIZEOF(.data) + ADDR(.data) :
>    {
>       PROVIDE (__bss_start = .) ;
>
> This works great. It creates __user_prog_start at the correct location. Then, 
> in my application, I can initialize a variable to this value with:
>
> prog_uint8_t *VPC = (prog_uint8_t *) &__user_prog_start;
>
> This also works great. The problem is that this user space is completely 
> uninitialized. I need to have at least one 0xFF at this location in my .hex 
> file to insure that uploading a new version of my code will wipe out any user 
> code that might have been there previously.
>
> Can anyone suggest an easy way to do this? Is there something I can add to my 
> script to include an 0xFF at location __user_prog_start? Perhaps something I 
> can do in one of the .c files .S files to stuff a single 0xFF there? Otherwise 
> I'm going to have to write a program to alter the .hex file after its creation 
> and that's awful Kludgey.
>
> Thanks!
> Gre7g
>
>
>       
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> AVR-GCC-list mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/avr-gcc-list