attribute(mode) ignored by Sun cc

Jay K jay.krell at cornell.edu
Sat Jun 16 23:16:14 CEST 2012


What are the requirements of these types, for gmp and mpfr?


Can they be chosen based on #if against values in limits.h?


In the #else for attribute(mode), does that version definitely work?
Even if long is 32 bits? I suspect not, but don't know.
If not, should there be autoconfery to find a 64bit type, e.g. check if long is 64
bits, or try __int64?


If attribute(mode) is ignored, does that work? I doubt that.
i.e. what is required? What is just nice to have as an optimization?
Add comments there in the code?


I only build gmp/mpfr/mpc as part of gcc, so optimization isn't needed.



Thank you,
 - Jay


----------------------------------------
> From: jay.krell at cornell.edu
> To: tg at gmplib.org
> CC: gmp-bugs at gmplib.org
> Subject: RE: attribute(mode) ignored by Sun cc
> Date: Sat, 16 Jun 2012 17:51:07 +0000
>
>
> mode is ignored, so:
> typedef unsigned int UQItype    __attribute__ ((mode (QI)));
> typedef         int SItype    __attribute__ ((mode (SI)));
> typedef unsigned int USItype    __attribute__ ((mode (SI)));
> typedef         int DItype    __attribute__ ((mode (DI)));
> typedef unsigned int UDItype    __attribute__ ((mode (DI)));
>
>
> ends up as
>
> typedef unsigned int UQItype;
> typedef         int SItype;
> typedef unsigned int USItype;
> typedef         int DItype;
> typedef unsigned int UDItype;
>
>
> isn't that wrong and won't work? Or is it ok?
>
>
> Thank you,
>   - Jay
>
> ----------------------------------------
> > To: jay.krell at cornell.edu
> > CC: gmp-bugs at gmplib.org
> > Subject: Re: attribute(mode) ignored by Sun cc
> > From: tg at gmplib.org
> > Date: Sat, 16 Jun 2012 15:40:20 +0200
> >
> > Jay K <jay.krell at cornell.edu> writes:
> >
> > I reported this years ago.
> > It is still a problem in 5.0.2.
> >
> > Please spell out what the problem is.
> >
> > If it is only a warning, then it doesn't seem very important to fix.
> >
> > --
> > Torbjörn
>
 		 	   		  


More information about the gmp-bugs mailing list