Re: [avr-gcc-list] avr-libc does not have support for ATxmega16a4u or ATxmega32a4u

[Erik Walthinsen]

On 02/14/2014 12:04 AM, Joerg Wunsch wrote:
> Alas, since the day when Anitha Boyapati has left the Atmel team that
> handles the AVR tools, there has not been much contribution back from
> Atmel to avr-libc.  Instead, they maintain their own set of patches
> (until it will eventually become unmaintainable ...).

In the meantime, I submitted a bug well over a year ago that included 
scripts to walk through all three codebases (binutils, gcc, avr-libc) 
and automatically insert support for every known (at the time) Xmega 
chip, which covers everything except the headers and any per-chip util/ 
actual code.

http://savannah.nongnu.org/bugs/?37850

Though honestly what's needed is an extension of what somebody was doing 
a while back with linker scripts: create a definitions file for every 
single chip, then either rewrite all the toolchain build scripts to 
generate all their own per-chip stuff on the fly, or extend the script I 
posted in that bug report to do it as a pre-processing pass for *all* chips.

As for the headers, I thought we'd long since got past the claim that 
deriving headers from the XML files was verboten.  There is nothing 
standing in the way of fixing up the script and just getting on with it.

There is *absolutely* no reason we have to "wait" for Atmel to do this, 
especially since many of these Xmega chips have been out for 2 to 3 
*years* with no support from the *open-source* toolchain!  Especially 
since my patches have been sitting there for over a year, after being 
told to put them there and they'd get processed.

*Every* time I have to set up a machine to do development for my 
*product* I have to carefully root around for my own local patches and 
do a full toolchain build, then hope it still works.  Getting very old.

I'll gladly file the assignment paperwork with the the FSF to be able to 
get this stuff done myself, it somebody will let me.