Re: [Amforth-devel] amforth trunk atmel168

[Matthias T. <mt...@we...>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Nathaniel,

Nathaniel Haggard wrote:

> I'm trying to compile amforth for an atmel168 using ubuntu hardy,
> amforth trunk, and AvrStudio416 under wine.

trunk may not always work with all assemblers and host systems. You
may get better results with the released versions.

btw: how does the studio work under wine?

> It fails to compile with this command: "wine ~/avrasm2.exe -I
> ~/Appnotes/ -I ../../core  -e template.hex.eep -l template.hex.lst -m
> template.hex.map template.asm"
> 
> The first error was "No newline at the end of noop.asm" and that was
> easy to fix.

Fixed as well before last release ;=)

> 
> These other two errors are more tricky:

Not really since the are caused by the same reason.
Atmel hast changed the definition files over time and
esp the label names for the interrupt vectors are
not part of the XML part description files. The
easy fix is to guess the right name

> forward referenced symbol 'ADCaddr' in .org

is the 'ADCCaddr' in some versions of the include files.

> template.asm(13): info: '../../core\devices/atmega168.asm' included from here
> template.asm(33): Including file '../../core\amforth.asm'
> ../../core\amforth.asm(8): error: Overlap in .cseg: addr=0x0 conflicts
> with 0x0:0x1
> template.asm(33): info: '../../core\amforth.asm' included from here

This is a follow-up error of the first one is will be gone with the
proper value für the ADC-Interrupt..

For a few chips I already used the converter script that produces the
amforth device definition file and use the address values instead
of some symbolic names. But the generator is not perfect so I did not
change the existing device files. I should do so for the next release.

Matthias
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFJpZ+C9bEHdGEMFjMRAubmAKDIVth5H9liEM3GHjGtwSjUiojdMQCgq9BR
ANJ7RjE2wc39weoYIC/xSLw=
=xDif
-----END PGP SIGNATURE-----