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, 117 insertions, 0 deletions

diff --git a/gcc/testsuite/c-c++-common/dfp/usual-arith-conv.c b/gcc/testsuite/c-c++-common/dfp/usual-arith-conv.c
new file mode 100644
index 000000000..baff06bc1
--- /dev/null
+++ b/gcc/testsuite/c-c++-common/dfp/usual-arith-conv.c
@@ -0,0 +1,117 @@
+/* { dg-options "-O0" } */
+
+/* N1150 5.4: Usual arithmetic conversions.
+   C99 6.3.1.8[1] (New).
+
+   Test arithmetic operators with different decimal float types, and
+   between decimal float types and integer types.  */
+
+#include "dfp-dbg.h"
+
+volatile _Decimal32 d32a, d32b, d32c;
+volatile _Decimal64 d64a, d64b, d64c;
+volatile _Decimal128 d128a, d128b, d128c;
+volatile int i;
+
+void
+init ()
+{
+  d32b = 123.456e94df;
+  d64b = 12.3456789012345e383dd;
+  d128b = 12345.6789012345678901e4000dl;
+
+  d32c = 1.3df;
+  d64c = 1.2dd;
+  d128c = 1.1dl;
+
+  i = 2;
+}
+
+int
+main ()
+{
+  init ();
+
+  /* Usual arithmetic conversions between decimal float types; addition.  */
+  d128a = d128b + d32b;
+  if (d128a < d128b)
+    FAILURE
+  d128a = d32b + d128b;
+  if (d128a < d128b)
+    FAILURE
+  d128a = d128b + d64b;
+  if (d128a < d128b)
+    FAILURE
+  d128a = d64b + d128b;
+  if (d128a < d128b)
+    FAILURE
+  d64a = d64b + d32b;
+  if (d64a < d64b)
+    FAILURE
+  d64a = d32b + d64b;
+  if (d64a < d64b)
+    FAILURE
+
+  /* Usual arithmetic conversions between decimal float types;
+     multiplication.  */
+  d128a = d128b * d32c;
+  if (d128a < d128b)
+    FAILURE
+  d128a = d32c * d128b;
+  if (d128a < d128b)
+    FAILURE
+  d128a = d128b * d64c;
+  if (d128a < d128b)
+    FAILURE
+  d128a = d64c * d128b;
+  if (d128a < d128b)
+    FAILURE
+  d64a = d64b * d32c;
+  if (d64a < d64b)
+    FAILURE
+  d64a = d32c * d64b;
+  if (d64a < d64b)
+    FAILURE
+
+  /* Usual arithmetic conversions between decimal float and integer types.  */
+  d32a = d32c + i;
+  if (d32a != d32c + 2.0df)
+    FAILURE
+  d32a = d32c - i;
+  if (d32a != d32c - 2.0df)
+    FAILURE
+  d32a = i * d32c;
+  if (d32a != d32c + d32c)
+    FAILURE
+  d32a = d32c / i;
+  if (d32a != d32c / 2.0df)
+    FAILURE
+
+  d64a = i + d64c;
+  if (d64a != d64c + 2.0dd)
+    FAILURE
+  d64a = d64c - i;
+  if (d64a != d64c - 2.0dd)
+    FAILURE
+  d64a = d64c * i;
+  if (d64a != d64c + d64c)
+    FAILURE
+  d64a = d64c / i;
+  if (d64a != d64c / 2.0dd)
+    FAILURE
+
+  d128a = d128c + i;
+  if (d128a != d128c + 2.0dl)
+    FAILURE
+  d128a = d128c - i;
+  if (d128a != d128c - 2.0dl)
+    FAILURE
+  d128a = i * d128c;
+  if (d128a != d128c + d128c)
+    FAILURE
+  d128a = d128c / i;
+  if (d128a != d128c / 2.0dl)
+    FAILURE
+
+  FINISH
+}