[Bug rtl-optimization/28096] fdlibm/strtod.c miscompiled at -O2

[dannysmith at users dot sourceforge dot net]

------- Comment #13 from dannysmith at users dot sourceforge dot net  
2006-09-11 19:47 -------
In my sources for David Gay's gdtoa implemntation it say this:
/* On a machine with IEEE extended-precision registers, it is
 * necessary to specify double-precision (53-bit) rounding precision
 * before invoking strtod or dtoa.  If the machine uses (the equivalent
 * of) Intel 80x87 arithmetic, the call
 *      _control87(PC_53, MCW_PC);
 * does this with many compilers.  Whether this or another call is
 * appropriate depends on the compiler; for this to work, it may be
 * necessary to #include "float.h" or another system-dependent header
 * file.
 */


There is a variant of strtod in the gdtoa file strtodnrp.c:
/* This is a variant of strtod that works on Intel ia32 systems */
/* with the default extended-precision arithmetic -- it does not */
/* require setting the precision control to 53 bits.  */


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28096