public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
* [PATCH] Fix ldbl-128ibm fma (Inf, Inf, finite) (bug 23272)
@ 2018-06-13 22:01 Tulio Magno Quites Machado Filho
  2018-06-13 23:07 ` Joseph Myers
  0 siblings, 1 reply; 11+ messages in thread
From: Tulio Magno Quites Machado Filho @ 2018-06-13 22:01 UTC (permalink / raw)
  To: libc-alpha

This solution is mainly based on Joseph's fix at commit
ca121b117f2c9c97a4c121334481a96c94fef3a0.

It has extra differences because:
 - part of the non-finite arguments were already being treated;
 - when x and y are +-Inf and z if finite, an overflow can be
   generated.

Tested on powerpc[|64|64le].

2018-06-13  Tulio Magno Quites Machado Filho  <tuliom@linux.ibm.com>

	[BZ 23272]
	* sysdeps/ieee754/ldbl-128ibm/s_fmal.c (__fmal): Handle all
	cases of non-finite arguments.

Signed-off-by: Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
---
 sysdeps/ieee754/ldbl-128ibm/s_fmal.c | 31 ++++++++++++-------------------
 1 file changed, 12 insertions(+), 19 deletions(-)

diff --git a/sysdeps/ieee754/ldbl-128ibm/s_fmal.c b/sysdeps/ieee754/ldbl-128ibm/s_fmal.c
index e72a3e4d59..afba1ca727 100644
--- a/sysdeps/ieee754/ldbl-128ibm/s_fmal.c
+++ b/sysdeps/ieee754/ldbl-128ibm/s_fmal.c
@@ -127,37 +127,30 @@ long double
 __fmal (long double x, long double y, long double z)
 {
   double xhi, xlo, yhi, ylo, zhi, zlo;
-  int64_t hx, hy, hz;
-  int xexp, yexp, zexp;
   double scale_val;
   int scale_exp;
   ldbl_unpack (x, &xhi, &xlo);
-  EXTRACT_WORDS64 (hx, xhi);
-  xexp = (hx & 0x7ff0000000000000LL) >> 52;
   ldbl_unpack (y, &yhi, &ylo);
-  EXTRACT_WORDS64 (hy, yhi);
-  yexp = (hy & 0x7ff0000000000000LL) >> 52;
   ldbl_unpack (z, &zhi, &zlo);
-  EXTRACT_WORDS64 (hz, zhi);
-  zexp = (hz & 0x7ff0000000000000LL) >> 52;
 
-  /* If z is Inf or NaN, but x and y are finite, avoid any exceptions
-     from computing x * y.  */
-  if (zexp == 0x7ff && xexp != 0x7ff && yexp != 0x7ff)
+  if (__glibc_unlikely (!isfinite (x) || !isfinite (y) || x == 0 || y == 0))
+    if (!isfinite (x) && !isfinite (y) && isfinite(z))
+      /* Compute the result as x * y to avoid an overflow.  */
+      return x * y;
+    else
+      /* If x or y is Inf, NaN or zero compute as x * y + z.  */
+      return (x * y) + z;
+  else if (__glibc_unlikely (!isfinite (z)))
+    /* If z is Inf, but x and y are finite, the result should be z
+       rather than NaN.  */
     return (z + x) + y;
 
-  /* If z is zero and x are y are nonzero, compute the result as x * y
+  /* If z is zero and x and y are nonzero, compute the result as x * y
      to avoid the wrong sign of a zero result if x * y underflows to
      0.  */
-  if (z == 0 && x != 0 && y != 0)
+  if (z == 0)
     return x * y;
 
-  /* If x or y or z is Inf/NaN, or if x * y is zero, compute as x * y
-     + z.  */
-  if (xexp == 0x7ff || yexp == 0x7ff || zexp == 0x7ff
-      || x == 0 || y == 0)
-    return (x * y) + z;
-
   {
     SET_RESTORE_ROUND (FE_TONEAREST);
 
-- 
2.14.4

^ permalink raw reply	[flat|nested] 11+ messages in thread
end of thread, other threads:[~2019-01-16 12:44 UTC | newest]

Thread overview: 11+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-06-13 22:01 [PATCH] Fix ldbl-128ibm fma (Inf, Inf, finite) (bug 23272) Tulio Magno Quites Machado Filho
2018-06-13 23:07 ` Joseph Myers
2018-09-21 12:00   ` Florian Weimer
2018-09-21 12:46     ` Joseph Myers
2018-09-21 12:51       ` Florian Weimer
2018-09-21 17:27         ` Tulio Magno Quites Machado Filho
2019-01-14 20:10   ` [PATCH] Add XFAIL_ROUNDING_IBM128_LIBGCC to more fma() tests Tulio Magno Quites Machado Filho
2019-01-14 21:33     ` Joseph Myers
2019-01-15 18:55       ` Tulio Magno Quites Machado Filho
2019-01-15 20:53         ` Joseph Myers
2019-01-16 12:44           ` Tulio Magno Quites Machado Filho

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).