This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: ppc64 floating point usage
- From: "Zack Weinberg" <zack at codesourcery dot com>
- To: David Edelsohn <dje at watson dot ibm dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Fri, 08 Aug 2003 18:55:12 -0700
- Subject: Re: ppc64 floating point usage
- References: <200308081401.KAA27398@makai.watson.ibm.com>
David Edelsohn <dje@watson.ibm.com> writes:
>>>>>> Alan Modra writes:
>
> Alan> Going back over gcc-patches archives, it appears that this was to fix
> Alan> a reload problem. (See [RFC PATCH] Fix middle-end/6963 from June and
> Alan> July 2002.) The good news is that the underlying reload problem is
> Alan> fixed, I think. I'm testing the following patch, which compiles the
> Alan> testcase in http://gcc.gnu.org/ml/gcc-patches/2002-06/msg02214.html
> Alan> without a problem, and things are looking good so far with regression
> Alan> tests.
>
> Alan> * config/rs6000/rs6000.md (fctiwz): Don't ignore fpr preference.
> Alan> (floatdidf2, fix_truncdfdi2, floatdisf2_internal1): Likewise.
> Alan> (floatditf2, fix_trunctfdi2): Likewise.
>
> Removing the ignore flag on all of the patterns is incorrect.
> This activity continues to fix symptoms instead of problems.
Would you please explain this in more detail? What do you think the
problem is, and the correct fix?
zw