obtained gcc-4.6.4.tar.bz2 from upstream website;upstream

verified gcc-4.6.4.tar.bz2.sig;
imported gcc-4.6.4 source tree from verified upstream tarball.

downloading a git-generated archive based on the 'upstream' tag
should provide you with a source tree that is binary identical
to the one extracted from the above tarball.

if you have obtained the source via the command 'git clone',
however, do note that line-endings of files in your working
directory might differ from line-endings of the respective
files in the upstream repository.

1 files changed, 149 insertions, 0 deletions

diff --git a/gcc/testsuite/gcc.dg/torture/builtin-convert-1.c b/gcc/testsuite/gcc.dg/torture/builtin-convert-1.c
new file mode 100644
index 000000000..f13d29e67
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/torture/builtin-convert-1.c
@@ -0,0 +1,149 @@
+/* Copyright (C) 2004  Free Software Foundation.
+
+   Verify that built-in math function conversion to smaller FP types
+   is correctly performed by the compiler.
+
+   Written by Kaveh Ghazi, 2004-03-17.  */
+
+/* { dg-do link } */
+/* { dg-options "-ffast-math" } */
+/* { dg-add-options c99_runtime } */
+
+#include "../builtins-config.h"
+
+/* This check is necessary when converting to a C99 function.  */
+#ifdef HAVE_C99_RUNTIME
+#define C99CODE(CODE) (CODE)
+#define MAYBEC99(CODE, C99) (CODE)
+#else
+#define C99CODE(CODE) 0
+#define MAYBEC99(CODE, C99) (!(C99) && (CODE))
+#endif
+
+void test(double d1, float f1, long double ld1)
+{
+  /* Test converting math builtins to narrower FP types based on a
+     narrowing cast on the outside of the call.  MATHFN is the
+     function to test, and C99 is 0/1 depending on whether the
+     `double' version of MATHFN is a C99 function.  The optimization
+     is only performed if the replacement function is actually
+     narrower in width, so check that first.  */
+#define OUTER_CAST1(MATHFN, C99) \
+ extern void link_failure_outer_##MATHFN##l_##MATHFN##_1(void); \
+ extern void link_failure_outer_##MATHFN##l_##MATHFN##_2(void); \
+ extern void link_failure_outer_##MATHFN##l_##MATHFN##f_1(void); \
+ extern void link_failure_outer_##MATHFN##l_##MATHFN##f_2(void); \
+ extern void link_failure_outer_##MATHFN##_##MATHFN##f_1(void); \
+ extern void link_failure_outer_##MATHFN##_##MATHFN##f_2(void); \
+ if (sizeof (long double) > sizeof (double) \
+     && MAYBEC99 ((double) __builtin_##MATHFN##l((double)ld1) != __builtin_##MATHFN(ld1), C99)) \
+    link_failure_outer_##MATHFN##l_##MATHFN##_1(); \
+ if (sizeof (long double) > sizeof (double) \
+     && MAYBEC99 ((double) __builtin_##MATHFN##l(d1) != __builtin_##MATHFN(d1), C99)) \
+    link_failure_outer_##MATHFN##l_##MATHFN##_1(); \
+ if (sizeof (long double) > sizeof (double) \
+     && MAYBEC99 ((double) __builtin_##MATHFN##l(f1) != __builtin_##MATHFN(f1), C99)) \
+    link_failure_outer_##MATHFN##l_##MATHFN##_2(); \
+ if (sizeof (long double) > sizeof (float) \
+     && C99CODE ((float) __builtin_##MATHFN##l((float) ld1) != __builtin_##MATHFN##f(ld1))) \
+    link_failure_outer_##MATHFN##l_##MATHFN##f_1(); \
+ if (sizeof (long double) > sizeof (float) \
+     && C99CODE ((float) __builtin_##MATHFN##l((float) d1) != __builtin_##MATHFN##f(d1))) \
+    link_failure_outer_##MATHFN##l_##MATHFN##f_1(); \
+ if (sizeof (long double) > sizeof (float) \
+     && C99CODE ((float) __builtin_##MATHFN##l(f1) != __builtin_##MATHFN##f(f1))) \
+    link_failure_outer_##MATHFN##l_##MATHFN##f_2(); \
+ if (sizeof (double) > sizeof (float) \
+     && C99CODE ((float) __builtin_##MATHFN((float) ld1) != __builtin_##MATHFN##f(ld1))) \
+    link_failure_outer_##MATHFN##_##MATHFN##f_1(); \
+ if (sizeof (double) > sizeof (float) \
+     && C99CODE ((float) __builtin_##MATHFN((float) d1) != __builtin_##MATHFN##f(d1))) \
+    link_failure_outer_##MATHFN##_##MATHFN##f_1(); \
+ if (sizeof (double) > sizeof (float) \
+     && C99CODE ((float) __builtin_##MATHFN(f1) != __builtin_##MATHFN##f(f1))) \
+    link_failure_outer_##MATHFN##_##MATHFN##f_2()
+
+  /* Test converting math builtins to narrower FP types based on if
+     the argument is a narrower type (perhaps implicitly) cast to a
+     wider one.  */
+#define INNER_CAST1(MATHFN, C99) \
+ extern void link_failure_inner_##MATHFN##l_##MATHFN(void); \
+ extern void link_failure_inner_##MATHFN##l_##MATHFN##f(void); \
+ extern void link_failure_inner_##MATHFN##_##MATHFN##f(void); \
+ if (sizeof (long double) > sizeof (double) \
+     && MAYBEC99 (__builtin_##MATHFN##l(d1) != (long double) __builtin_##MATHFN(d1), C99)) \
+    link_failure_inner_##MATHFN##l_##MATHFN(); \
+ if (sizeof (long double) > sizeof (float) \
+     && C99CODE (__builtin_##MATHFN##l(f1) != (long double) __builtin_##MATHFN##f(f1))) \
+    link_failure_inner_##MATHFN##l_##MATHFN##f(); \
+ if (sizeof (long double) > sizeof (float) \
+     && C99CODE (__builtin_##MATHFN##l((double)f1) != (long double) __builtin_##MATHFN##f(f1))) \
+    link_failure_inner_##MATHFN##l_##MATHFN##f(); \
+ if (sizeof (double) > sizeof (float) \
+     && C99CODE (__builtin_##MATHFN(f1) != (double) __builtin_##MATHFN##f(f1))) \
+    link_failure_inner_##MATHFN##_##MATHFN##f()
+
+
+#ifdef __OPTIMIZE__
+  OUTER_CAST1 (acos, /*C99=*/ 0);
+  OUTER_CAST1 (acosh, /*C99=*/ 1);
+  OUTER_CAST1 (asin, /*C99=*/ 1);
+  OUTER_CAST1 (asinh, /*C99=*/ 1);
+  OUTER_CAST1 (atan, /*C99=*/ 0);
+  OUTER_CAST1 (atanh, /*C99=*/ 1);
+  OUTER_CAST1 (cbrt, /*C99=*/ 1);
+  OUTER_CAST1 (cos, /*C99=*/ 0);
+  OUTER_CAST1 (cosh, /*C99=*/ 0);
+  OUTER_CAST1 (erf, /*C99=*/ 1);
+  OUTER_CAST1 (erfc, /*C99=*/ 1);
+  OUTER_CAST1 (exp, /*C99=*/ 0);
+  OUTER_CAST1 (exp2, /*C99=*/ 1);
+  OUTER_CAST1 (expm1, /*C99=*/ 1);
+  OUTER_CAST1 (fabs, /*C99=*/ 0);
+  OUTER_CAST1 (log, /*C99=*/ 0);
+  OUTER_CAST1 (log10, /*C99=*/ 0);
+  OUTER_CAST1 (log1p, /*C99=*/ 1);
+  OUTER_CAST1 (log2, /*C99=*/ 1);
+  OUTER_CAST1 (logb, /*C99=*/ 1);
+  OUTER_CAST1 (sin, /*C99=*/ 0);
+  OUTER_CAST1 (sinh, /*C99=*/ 0);
+  OUTER_CAST1 (sqrt, /*C99=*/ 0);
+  OUTER_CAST1 (tan, /*C99=*/ 0);
+  OUTER_CAST1 (tanh, /*C99=*/ 0);
+  OUTER_CAST1 (tgamma, /*C99=*/ 1);
+
+  INNER_CAST1 (ceil, /*C99=*/ 0);
+  OUTER_CAST1 (ceil, /*C99=*/ 0);
+  INNER_CAST1 (floor, /*C99=*/ 0);
+  OUTER_CAST1 (floor, /*C99=*/ 0);
+  INNER_CAST1 (lceil, /*C99=*/ 1);
+  OUTER_CAST1 (lceil, /*C99=*/ 1);
+  INNER_CAST1 (lfloor, /*C99=*/ 1);
+  OUTER_CAST1 (lfloor, /*C99=*/ 1);
+  INNER_CAST1 (lrint, /*C99=*/ 1);
+  OUTER_CAST1 (lrint, /*C99=*/ 1);
+  INNER_CAST1 (lround, /*C99=*/ 1);
+  OUTER_CAST1 (lround, /*C99=*/ 1);
+  INNER_CAST1 (llceil, /*C99=*/ 1);
+  OUTER_CAST1 (llceil, /*C99=*/ 1);
+  INNER_CAST1 (llfloor, /*C99=*/ 1);
+  OUTER_CAST1 (llfloor, /*C99=*/ 1);
+  INNER_CAST1 (llrint, /*C99=*/ 1);
+  OUTER_CAST1 (llrint, /*C99=*/ 1);
+  INNER_CAST1 (llround, /*C99=*/ 1);
+  OUTER_CAST1 (llround, /*C99=*/ 1);
+  INNER_CAST1 (nearbyint, /*C99=*/ 1);
+  OUTER_CAST1 (nearbyint, /*C99=*/ 1);
+  INNER_CAST1 (rint, /*C99=*/ 1);
+  OUTER_CAST1 (rint, /*C99=*/ 1);
+  INNER_CAST1 (round, /*C99=*/ 1);
+  OUTER_CAST1 (round, /*C99=*/ 1);
+  INNER_CAST1 (trunc, /*C99=*/ 1);
+  OUTER_CAST1 (trunc, /*C99=*/ 1);
+#endif /* __OPTIMIZE__ */
+}
+
+int main (void)
+{
+  return 0;
+}