On Oct 1,  6:46pm, Mark Brinicombe wrote:
> Subject: Re: Installing....
> Someone else wrote:
> >How much of a change is the StrongARM going to cause - are we talking
> >just a new kernel or replacing/recompiling most of the programs/utilities.
>
> Definitely a new kernel. Eventually I would like to have a generic kernel
that
> supported all ARM's however the different cache structure and abort behaviour
> mean that StrongARM support is a kernel compile time option.
>
> I do not expect very much to change. The major problem that RiscOS programs
> have been affected by is self modifing code etc. Under un*x the text area of
a
> process is read only so the problem of inconsistancies with the icache does
not
> exist.
>
> There will however by an architecure specific library libarm32.a appearing
soon
> which will contain a function that will allow a user process sync the icache
> but I do not expect any existing software to need to use it.

Right, I've checked a few things. There should be no problem with emacs/xemacs
as long as the address of _DYNAMIC is NULL (ie they are statically linked).
There is a compilation option for X which allows handling of harvard cache
architectures which changes the way that a lot of the optimisations are done.
As far as I can tell, if you run the old version of Xarm on a SA you just might
need to keep typing xrefresh a lot (provided that you aren't using extensions
in shared libraries - since we don't support shared libraries I can't see this
being a problem).

Cheers

Rob

--