Why 2.95.4 ? (and libscl)

Nick Burrett nick at dsvr.net
Fri Feb 8 01:13:21 PST 2002

John Tytgat <John.Tytgat at aaug.net> writes:

> In message <m34rkte5qz.fsf at nick.ws.noc.dsvr.net>
>           Nick Burrett <nick at dsvr.net> wrote:
> > Peter Naulls <peter at chocky.org> writes:
> > >  - This SP vs FP variable access.  Nick, if you can collect your
> > >    thoughts on this one, I'd be interested to hear why this change is
> > >    needed.
> > 
> > Simply because it reduces the ARM backend code changes required for
> > RISC OS down to a few lines rather than about 150Kb of diffs.  This
> > is not necessarily my laziness, but I think that by sticking to sp-based
> > local variables, we will lose out in the long run.
> > 
> > I would like to get an initial port of GCC 3.1 working based on fp-based
> > variables then consider the issues that arise after.
> I'm probably showing my ignorance here but would this change have any
> influence on APCS-R/32 compliancy ? Or is this just a discussion how
> local variables are addressed/managed in the GCC's ARM backend and doesn't
> have any influence on interprocedural calling convention ?

It does change compliancy, but only within the function. Interprocedural
calling conventions remain the same.


More information about the gcc mailing list