Re: [avr-gcc-list] It is more than: .byte SYMA - SYMB

[Denis Chertykov]

Joerg Wunsch <address@hidden> writes:

> As Denis Chertykov <address@hidden> wrote:
> 
> >> c: .string  "Hello World"
> >>    .align 2
> >> d: ldi   XH,pm_hi8(d - c)     <- OK
> >>    ldd   r16,Y+(d - c)        <- Error: constant value required
> 
> >This happened because relocations for Y+N or Z+N not supported.
> >Please read GAS info about relocations.
> 
> Still, i don't see (after reading the gas manual) when the assembler
> decides that it'll be a relocation vs. an absolute expression.
> 
> Am i right that changing the code like this would make it an
> absolute expression again?

No, because I have supported special relocation for `ldi' this  reloc
called `R_AVR_HI8_LDI_PM'. You can look at bfd/elf32-avr.c

> 
>    .align 2
> c: .string  "Hello World"
>    .align 2
> d: ldi   XH,pm_hi8(d - c)     <- OK
>    ldd   r16,Y+(d - c)        <- Error: constant value required
> 
> If not, i'm confused, since as far as i can tell, in the above,
> d - c is an assembly-time constant expression.

I don't know why GAS treat (d - c) as not a constant value, but I see
that (d - c) work as relocatable expression.

Denis.
PS: GAS isn't so good as GCC.