Seems ot make a lot of sense - with 'arch/xxx' subdirectories as
	appropriate? I think at least sh3 and optionally mips could share
	most of the same softfloat code as arm{26,32}...


		David/absolute
				       -- www.netbsd.org: No hype required --

On Tue, 16 May 2000, Ben Harris wrote:

> On Tue, 16 May 2000, Philip Blundell wrote:
> 
> > In message <Pine.SOL.4.21.0005160007190.25889-100000@taurus.cus.cam.ac.uk>, Ben
> >  Harris writes:
> > >Right, so the way to arrange this is to get the SoftFloat code to return
> > >long longs, since they also get returned in two registers.  It looks as if
> > >the current version of SoftFloat (version 2a) supports doing this.  Is
> > >there any reason not to go over to using this?
> > 
> > This is what Linux does, and it seems to work fine.
> 
> OK.  Given that SoftFloat is (mostly) machine-independent, what do people
> think about a lib/libc/softfloat directory to hold the MI bits?
> 
> -- 
> Ben Harris                                                   <bjh21@netbsd.org>
> Portmaster, NetBSD/arm26               <URL:http://www.netbsd.org/Ports/arm26/>
> 
>