From: Roger Sayle Date: Tue, 19 Dec 2006 04:17:11 +0000 (+0000) Subject: re PR target/29302 (isfinite returns wrong result at -O1) X-Git-Tag: releases/gcc-4.3.0~7911 X-Git-Url: http://git.ipfire.org/gitweb.cgi?a=commitdiff_plain;h=f839548fc7279f813bebeca759f0a04dae5b4f62;p=thirdparty%2Fgcc.git re PR target/29302 (isfinite returns wrong result at -O1) 2006-12-18 Roger Sayle Eric Christopher PR target/29302 * real.c (real_maxval): Correctly handle IBM extended double format. Co-Authored-By: Eric Christopher From-SVN: r120040 --- diff --git a/gcc/ChangeLog b/gcc/ChangeLog index 5e7bd5ce94b5..a76a75a470d8 100644 --- a/gcc/ChangeLog +++ b/gcc/ChangeLog @@ -1,3 +1,9 @@ +2006-12-18 Roger Sayle + Eric Christopher + + PR target/29302 + * real.c (real_maxval): Correctly handle IBM extended double format. + 2006-12-19 Jan Hubicka * expmed.c (expand_mult_const): Force operand to constant. diff --git a/gcc/real.c b/gcc/real.c index 0d239b3379d0..b6e5ed6ee3e9 100644 --- a/gcc/real.c +++ b/gcc/real.c @@ -2282,6 +2282,15 @@ real_maxval (REAL_VALUE_TYPE *r, int sign, enum machine_mode mode) np2 = SIGNIFICAND_BITS - fmt->p * fmt->log2_b; memset (r->sig, -1, SIGSZ * sizeof (unsigned long)); clear_significand_below (r, np2); + + if (fmt->pnan < fmt->p) + /* This is an IBM extended double format made up of two IEEE + doubles. The value of the long double is the sum of the + values of the two parts. The most significant part is + required to be the value of the long double rounded to the + nearest double. Rounding means we need a slightly smaller + value for LDBL_MAX. */ + clear_significand_bit (r, SIGNIFICAND_BITS - fmt->pnan); } }