Re: GDB/libiberty support for IBM long double

[Luis Machado <luisgpm@linux.vnet.ibm.com>]

[-- Attachment #1: Type: text/plain, Size: 568 bytes --]

> However, could you test also replacing the two floatformat_from_double
> calls in convert_doublest_to_floatformat?

Sure. Follows the patch.

The results still show up correctly, casting doubles to long doubles and
long doubles to doubles.

===> Output <===
(gdb) set $ld=(long double)1.2
(gdb) p $ld
$8 = 1.1999999999999999555910790149937384
(gdb) p (double) $ld
$9 = 1.2

(gdb) set $ld=(double)1.2
(gdb) p $ld
$12 = 1.2
(gdb) p (long double) $ld
$13 = 1.1999999999999999555910790149937384


Regards,
-- 
Luis Machado
Software Engineer 
IBM Linux Technology Center

[-- Attachment #2: fix_long_double.diff --]
[-- Type: text/x-patch, Size: 1990 bytes --]

2008-01-02  Luis Machado  <luisgpm@br.ibm.com>

        * doublest.c (convert_floatformat_to_doublest): Call 
	floatformat_to_doublest instead of floatformat_to_double and use
	DOUBLEST variables.
	(convert_doublest_to_floatformat): Call floatformat_from_doublest
	instead of floatformat_from_double and use DOUBLEST variables.

Index: gdb/doublest.c
===================================================================
--- gdb.orig/doublest.c	2008-01-02 08:24:41.000000000 -0800
+++ gdb/doublest.c	2008-01-02 09:53:55.000000000 -0800
@@ -202,19 +202,19 @@
 
   if (fmt->split_half)
     {
-      double dtop, dbot;
-      floatformat_to_double (fmt->split_half, ufrom, &dtop);
+      DOUBLEST dtop, dbot;
+      floatformat_to_doublest (fmt->split_half, ufrom, &dtop);
       /* Preserve the sign of 0, which is the sign of the top
 	 half.  */
       if (dtop == 0.0)
 	{
-	  *to = (DOUBLEST) dtop;
+	  *to = dtop;
 	  return;
 	}
-      floatformat_to_double (fmt->split_half,
+      floatformat_to_doublest (fmt->split_half,
 			     ufrom + fmt->totalsize / FLOATFORMAT_CHAR_BIT / 2,
 			     &dbot);
-      *to = (DOUBLEST) dtop + (DOUBLEST) dbot;
+      *to = dtop + dbot;
       return;
     }
 
@@ -417,7 +417,7 @@
 	 removed via storing in memory, and so the top half really is
 	 the result of converting to double.  */
       static volatile double dtop, dbot;
-      double dtopnv, dbotnv;
+      DOUBLEST dtopnv, dbotnv;
       dtop = (double) dfrom;
       /* If the rounded top half is Inf, the bottom must be 0 not NaN
 	 or Inf.  */
@@ -427,8 +427,8 @@
 	dbot = (double) (dfrom - (DOUBLEST) dtop);
       dtopnv = dtop;
       dbotnv = dbot;
-      floatformat_from_double (fmt->split_half, &dtopnv, uto);
-      floatformat_from_double (fmt->split_half, &dbotnv,
+      floatformat_from_doublest (fmt->split_half, &dtopnv, uto);
+      floatformat_from_doublest (fmt->split_half, &dbotnv,
 			       (uto
 				+ fmt->totalsize / FLOATFORMAT_CHAR_BIT / 2));
       return;