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: PATCH: PR target/15383: __float128 function name is used for long double


On Thu, May 13, 2004 at 10:58:02AM -0700, Zack Weinberg wrote:
> "H. J. Lu" <hjl@lucon.org> writes:
> 
> > On Thu, May 13, 2004 at 12:35:54AM -0700, Zack Weinberg wrote:
> >> > Like this?
> >> 
> >> Yup.  However (as I said at the beginning of this thread) I think it
> >> makes sense to duplicate the entire set of HP interface names, not
> >> just the one for the divide routine.   Then you only have to have one
> >> hook to adjust the libfunc names, and so on.
> >> 
> >
> > The HP interface has some additional functions. This patch changes
> > the common subset of TFmode functions with the HPUX conventions.
> 
> Looks better, but:
> 
> 1) Shouldn't these routines be in libgcc?

Does libgcc support TFmode? My changes are based on soft-fp in glibc.

> 2) Some of your comments leave something to be desired.
> 3) Does ia64_expand_compare do the right thing?

I didn't implement _U_Qfcmp, which is used by HPUX only. I added
_U_Qfeq, _U_Qfne, _U_Qfgt, _U_Qfge, _U_Qflt, _U_Qfle for non-HPUX.

> 4) Check what Jim Wilson thinks of this.


H.J.


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