Re: [avr-gcc-list] What GCC bug currently effect the AVR port of

[E. Weddington]

[I've CCed interested parties.]

Andy Hutchinson wrote:

> I got FSF copyright papers today and sent them in.
>
> That will mean I will be able to contribute code for gcc to help with 
> avr port. (Once I figure out who I can rely on to help comit them)

Marek Michalkiewicz: AVR port maintainer. Can approve patches to the AVR 
port.
Denis Chertykov: AVR port maintainter. Can approve patches to the AVR port.
Bernardo (Bernie) Innocenti: Has write after approval. He's done some 
work with the AVR port among many other things.
Giovanni Bajo: Has write after approval. He has been nice enough to 
occasionally help keep track of some AVR specific bugs. Perhaps he might 
be willing to help with commits?

> I dont mind fixing a few of the bugs listed in Eric list but it is 
> probably worth somebody (Eric?) giving them a priority.
Anything that has to do with wrong-code or ICE.

From the AVR target bugs listed here:
<http://rtems.org/phpwiki/index.php/GCCAVRBugs>

These come to mind:
#10733: Need to check to see if there is a bug, or can it be closed.
#16563
#18551
#19636
#19885
#20082

But, if you wanted to start out with some easy bugs, then there are these:
#1078
#17822
#18553
#20016

And if any of the maintainers or people with Bugzilla permissions are 
reading this far ;-), then there are some bugs that need some 
administration:
#12017: Could be closed as Objective-C is not supported yet for the AVR.
#18887: See comment #21, and related bug #19331
#19331

> I wont be able to fix middle or front end issues - hell I have a few 
> bugs listed myself against that! Similary the will be bugs that get 
> put in the "too difficult" tray.
>
> Nevertheless, I feel pretty confident around the avr backend code and 
> there are a few others folk around that can provide ideas and help 
> check out solutions.

Obviously Björn Haase has done a lot of good work getting the testsuite 
running for the AVR. If you (or someone) else can do this too it would 
be a big help.

> It would seem logical that Winavr should have a experimental build to 
> help test such patches and test out other changes to gcc - especially 
> with 4.0
>
> Not everybody can build gcc from scratch - but many can spot a bug in 
> Winavr.


That makes sense. I'll have to put it on my TODO list. I can't guarantee 
that it will get done soon but I'll see what I can do.

HTH
Eric