This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PATCH: Handle TARGET_96_ROUND_53_LONG_DOUBLE, part I, II, III
- From: "Joseph S. Myers" <joseph at codesourcery dot com>
- To: rittle at labs dot mot dot com
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Fri, 2 Oct 2009 21:20:14 +0000 (UTC)
- Subject: Re: PATCH: Handle TARGET_96_ROUND_53_LONG_DOUBLE, part I, II, III
- References: <200910021856.n92IuspX029629@latour.labs.mot.com>
On Fri, 2 Oct 2009, Loren James Rittle wrote:
> Part IV, newly added. This patch kills another 2 gratitious FAILs.
> (in this case, because sizeof (long long) is greater than LDBL_MANT_DIG)
>
> OK, for trunk?
>
> Still looking at 30 FAILS related to gcc.dg/torture/fp-int-convert*.
> There appears to be multiple problems: (1) we don't have TI mode
> enabled (yet?); (2) similar issue to that of Part IV above (which
> may be hard to narrowly address given the nested macro expansion).
fp-int-convert* are designed to be valid even when dealing with integer
types wider than the floating-point types in question, and even when the
largest value rounds to an infinity (as with TImode-to-float conversions).
They should be perfectly valid for TARGET_96_ROUND_53_LONG_DOUBLE.
--
Joseph S. Myers
joseph@codesourcery.com