This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [RFC PATCH, i386]: Pass FP arguments in i387 registers


Ian Lance Taylor <iant@google.com> writes:

> "Richard Guenther" <richard.guenther@gmail.com> writes:
> 
> > On 12 Oct 2006 08:29:31 -0700, Ian Lance Taylor <iant@google.com> wrote:
> > > "Uros Bizjak" <ubizjak@gmail.com> writes:
> > >
> > > >         * config/i386/i386.opt: New target option -m387regparm.
> > >
> > > I know Richard suggested an option along the lines of -mregparm, but I
> > > think that as we add options of this sort it would make more sense to
> > > group them into a single -mabi= option.  The existing options would
> > > then be synonyms for a -mabi= option.
> > 
> > I don't see the advantage of this - we need to support the old switches anyway
> > and it's not very likely that the list of options will grow further
> > significantly.
> 
> That's what we say until we add one more.
> 
> We've already discussed the possibility of adding yet another ABI
> which only uses the low-order 32 bits of the x86_64 registers.

Oh, and the other reason to unify on a -mabi= option is that we then
have a single coherent string to put in a NOTE section which the
linker can use to detect attempts to link together objects compiled
with incompatible ABIs.

Ian


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]