Question regarding libm

Jan-Benedict Glaw jbglaw at lug-owl.de
Sat Oct 1 09:00:31 UTC 2005


On Sat, 2005-10-01 10:19:49 +0200, Peter S. Mazinger <ps.m at gmx.net> wrote:
> On Fri, 30 Sep 2005, Jan-Benedict Glaw wrote:
> > > +config TARGET_HAS_NEGATIVE_ZERO_FLOATS
> > > +	bool "    Target CPU differs between +0 and -0"
> > > +	default y
> > > +	depends on UCLIBC_HAS_FLOATS
> 
> I would rather do in config UCLIBC_HAS_FLOATS:
> select TARGET_HAS_NEGATIVE_ZERO_FLOATS if some archs (so that it is 
> enforced where appropiate
> 
> or at least one of
> 
> default y if some archs
> default n if some archs

You're very right on this.

But with my yesterday nights findings, I think there's no need to
apply this patch right now. A lot of code in libm is based about IEEE
floating point representation, so basically any function calling one
of the xxx_WORD() macros from math_private.h would produce broken
values on VAX. That is, I think libc is quite useful to have, but libm
is 99% broken :-)  So I either wait for the rewrite (@Manuel: is there
already some code available?), or I'll look into GNU libc, which at
some time has had VAX FP support. But anyways, I'll first continue to
build me a vax-linux-uclibc compiler.

Thanks, JBG

-- 
Jan-Benedict Glaw       jbglaw at lug-owl.de    . +49-172-7608481             _ O _
"Eine Freie Meinung in  einem Freien Kopf    | Gegen Zensur | Gegen Krieg  _ _ O
 für einen Freien Staat voll Freier Bürger"  | im Internet! |   im Irak!   O O O
ret = do_actions((curr | FREE_SPEECH) & ~(NEW_COPYRIGHT_LAW | DRM | TCPA));
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.busybox.net/pipermail/uclibc/attachments/20051001/ea0817eb/attachment-0002.pgp 


More information about the uClibc mailing list