|
Message
From: Matjaz Breskvar <phoenix@o...>
Date: Wed, 12 Nov 2003 13:28:55 +0100
Subject: Re: [openrisc] -O2 problem
* Matjaz Breskvar (phoenix@o...) wrote:
> * Damjan Lampret (lampret@o...) wrote:
> > I strictly use GCC 3.1 from opencores cvs. I don't use any other derivatives
> > that other folks have been making.
> >
> > I don't use any special flags just -O2 together with flags that don't have
> > anything to do how the code gets compiled (like -c and -o for example)
> >
> > I didn't have any problems with conditional branches.
>
> opencores gcc-3.2.3 should be fine too (i think it even solved a few
> problems for some people).
>
> i can say that when porting linux i started
> with -O2 optimizations and never had problems with it. then i've switched
> to -O just to be safe. i did try to turn it back at some milestones and it
> worked ok. (i'm using gcc-3.2.3 from opencores cvs)
>
> i'm going to recompile linux with -O2 and report what happens.
ok i did this and linux compiles & runs just as good as with -O.
this seams (by now) to be triggered only by eCos.
i'll look into all the stuff submited here so we can close
this issue ASAP. if someone has a working fix that would be
even better ;).
regards,
p.
|
Browse
