Re: CVS commit: src/sys/arch/mac68k/mac68k

[Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost>]

martin%duskware.de@localhost wrote:

> On Mon, Jan 19, 2009 at 08:45:21AM +0100, Ignatios Souvatzis wrote:
> > Your support function should be somehwere in arch/m68k/m68k/, and
> > called from every m68k port (well, at least Amiga, too ;-) Can you
> > do this?
> > 
> > [Is there a PR?]
> 
> No PR that I'm aware of - just run "make regress" in
> regress/lib/libc/ieeefp.
> 
> Attached patch does what you suggested - but I can't test on anything
> but mac68k.

With a quick test on on HP 9000/382 (68040), round gets an error:
---
% make -i regress
regress ===> except
./except
regress ===> infinity
./infinity
regress ===> nan
./nan
regress ===> round
./round
assertion "FLT_ROUNDS == 2" failed: file 
"/usr/src/regress/lib/libc/ieeefp/round/round.c", line 29, function "main"
*** Signal 6 (ignored)
regress ===> subnormal
./subnormal
% 
---

But accroding to your patch there is not any bad side effect
other than ieeefp so I think it's okay to commit if it's
confirmed at least on mac68k.
---
Izumi Tsutsui