Why 2.95.4 ? (and libscl)

Nick Burrett nick at dsvr.net
Thu Feb 7 05:22:33 PST 2002


Ian Jeffray <ian at paradise-uk.net> writes:

> Nick Burrett wrote:
> 
> [porting 2.96]
> 
> > This would probably mean that we change the local variable references
> > to be relative to `fp' rather than `sp'.  Though my suspicion is that
> > if we do this, then we lose support for non-contiguous stack chunks.
> 
> 
> That's pretty interesting.  I'd been discussing (or maybe just listening to
> discussion ;) about support for reentrant code output from gcc which, iirc,
> had a similar issue in that the workspace pointer would need to move.
> 
> It would be nice to get some of these varying issues noted down somewhere
> as we may be able to crack several nuts with one stone.  Maybe. ;-)

Easier said than done.  I think when one starts to go down this path
we soon start to realise how frustrating it can be to get this stuff
working on RISC OS.

Nick.



More information about the gcc mailing list