Re: [avr-gcc-list] ATMega128 RAMPZ behaviour changed ?

[E. Weddington]

On 16 Dec 2003 at 13:06, Simon Eatough wrote:

> Hi
> According to the ATMega128 documentation, bit 0 of RAMPZ selects the high or
> low page  access of Z pointer, with the other bits of RAMPZ unused. However
> it appears that setting RAMPZ to 3 causes elpm return garbage. This
> behaviour seems to break  pgm_read_byte_far(). Anybody else noticed this ?.

pgm_read_byte_far() accepts a uint32_t, which of course can be a value that 
extends beyond the flash address range of a mega128.

The disadvantage to limiting the allowable range of addresses in 
pgm_read_byte_far to those of the mega128 is that when the mega256 is 
released, then pgm_read_byte_far will no longer work. And, yes, I've heard 
that there will be such a beast as the mega256.

So the current response is: don't do that. Test your addresses before you 
use them as a parameter to pgm_read_byte_far().