real.c (emdnorm and etoasc): Disable round to even for c4x target to be compatible with TI compiler.

* real.c (emdnorm and etoasc): Disable round to even for c4x target
        to be compatible with TI compiler.

From-SVN: r23072
This commit is contained in:
Herman A.J. ten Brugge 1998-10-14 02:07:47 +02:00 committed by Jeff Law
parent 2fdc44405f
commit 506b012c6f
2 changed files with 11 additions and 1 deletions

View File

@ -11,6 +11,9 @@ Wed Oct 14 00:38:40 1998 Jeffrey A Law (law@cygnus.com)
1998-10-13 Herman A.J. ten Brugge <Haj.Ten.Brugge@net.HCC.nl>
* real.c (emdnorm and etoasc): Disable round to even for c4x target
to be compatible with TI compiler.
* Makefile.in (USER_H): Add va-c4x.h to definition.
Tue Oct 13 23:03:37 1998 Richard Henderson <rth@cygnus.com>

View File

@ -79,7 +79,10 @@ netlib.att.com: netlib/cephes. */
`C4X' refers specifically to the floating point format used on
Texas Instruments TMS320C3x and TMS320C4x digital signal
processors. This supports QFmode (32-bit float, double) and HFmode
(40-bit long double) where BITS_PER_BYTE is 32.
(40-bit long double) where BITS_PER_BYTE is 32. Unlike IEEE
floats, C4x floats are not rounded to be even. The C4x conversions
were contributed by m.hayes@elec.canterbury.ac.nz (Michael Hayes) and
Haj.Ten.Brugge@net.HCC.nl (Herman ten Brugge).
If LONG_DOUBLE_TYPE_SIZE = 64 (the default, unless tm.h defines it)
then `long double' and `double' are both implemented, but they
@ -2594,6 +2597,7 @@ emdnorm (s, lost, subflg, exp, rcntrl)
s[rw] &= ~rmsk;
if ((r & rmbit) != 0)
{
#ifndef C4X
if (r == rmbit)
{
if (lost == 0)
@ -2607,6 +2611,7 @@ emdnorm (s, lost, subflg, exp, rcntrl)
goto mddone;
}
}
#endif
eaddm (rbit, s);
}
mddone:
@ -4932,8 +4937,10 @@ etoasc (x, string, ndigs)
emovo (y, t);
if (ecmp (t, ezero) != 0)
goto roun; /* round to nearest */
#ifndef C4X
if ((*(s - 1) & 1) == 0)
goto doexp; /* round to even */
#endif
}
/* Round up and propagate carry-outs */
roun: