Fix ldbl-128ibm lrintl, lroundl missing "invalid" exceptions (bug 22690).

The ldbl-128ibm implementations of lrintl and lroundl are missing
"invalid" exceptions for certain overflow cases when compiled with GCC
8.  The cause of this is after-the-fact integer overflow checks that
fail when the compiler optimizes on the basis of integer overflow
being undefined; GCC 8 must be able to detect new cases of
undefinedness here.

Failure: lrint (-0x80000001p0): Exception "Invalid operation" not set
Failure: lrint_downward (-0x80000001p0): Exception "Invalid operation" not set
Failure: lrint_towardzero (-0x80000001p0): Exception "Invalid operation" not set
Failure: lrint_upward (-0x80000001p0): Exception "Invalid operation" not set

Failure: lround (-0x80000001p0): Exception "Invalid operation" not set
Failure: lround_downward (-0x80000001p0): Exception "Invalid operation" not set
Failure: lround_towardzero (-0x80000001p0): Exception "Invalid operation" not set
Failure: lround_upward (-0x80000001p0): Exception "Invalid operation" not set

(Tested that these failures occur before the patch for powerpc
soft-float, but the issue applies in principle for hard-float as well,
whether or not the particular optimizations in fact occur there at
present.)

This patch fixes the bug by ensuring the additions / subtractions in
question cast arguments to unsigned long int, or use 1UL as a constant
argument, so that the arithmetic occurs in an unsigned type with the
result then converted back to a signed type.

Tested for powerpc (soft-float).

	[BZ #22690]
	* sysdeps/ieee754/ldbl-128ibm/s_lrintl.c (__lrintl): Use unsigned
	long int for arguments of possibly overflowing addition or
	subtraction.
	* sysdeps/ieee754/ldbl-128ibm/s_lroundl.c (__lroundl): Likewise.

1 files changed, 5 insertions, 5 deletions

diff --git a/sysdeps/ieee754/ldbl-128ibm/s_lroundl.c b/sysdeps/ieee754/ldbl-128ibm/s_lroundl.c
index 9cf4a121a7..010e8e28a1 100644
--- a/sysdeps/ieee754/ldbl-128ibm/s_lroundl.c
+++ b/sysdeps/ieee754/ldbl-128ibm/s_lroundl.c
@@ -80,7 +80,7 @@ __lroundl (long double x)
       /* Peg at max/min values, assuming that the above conversions do so.
          Strictly speaking, we can return anything for values that overflow,
          but this is more useful.  */
-      res = hi + lo;
+      res = (long int) ((unsigned long int) hi + (unsigned long int) lo);
 
       /* This is just sign(hi) == sign(lo) && sign(res) != sign(hi).  */
       if (__glibc_unlikely (((~(hi ^ lo) & (res ^ hi)) < 0)))
@@ -92,21 +92,21 @@ __lroundl (long double x)
       hi = res;
       if (xh > 0.5)
 	{
-	  res += 1;
+	  res += 1UL;
 	}
       else if (xh == 0.5)
 	{
 	  if (xl > 0.0 || (xl == 0.0 && res >= 0))
-	    res += 1;
+	    res += 1UL;
 	}
       else if (-xh > 0.5)
 	{
-	  res -= 1;
+	  res -= 1UL;
 	}
       else if (-xh == 0.5)
 	{
 	  if (xl < 0.0 || (xl == 0.0 && res <= 0))
-	    res -= 1;
+	    res -= 1UL;
 	}
 
       if (__glibc_unlikely (((~(hi ^ (res - hi)) & (res ^ hi)) < 0)))