[backport from gcc-4.7/trunk r184627 <mikpe@it.uu.se>]

gcc/

2012-02-28  Richard Guenther  <rguenther@suse.de>

	PR target/52407
	* config/i386/i386.c (ix86_expand_vector_set): Fix element
	ordering for the VEC_CONCAT for two element vectors for
	V2SFmode, V2SImode and V2DImode.

gcc/testsuite/

2012-02-28  Richard Guenther  <rguenther@suse.de>

	PR target/52407
	* gcc.dg/torture/pr52407.c: New testcase.

--- gcc-4.6.3/gcc/config/i386/i386.c.~1~	2012-02-22 20:17:42.000000000 +0100
+++ gcc-4.6.3/gcc/config/i386/i386.c	2012-03-04 17:58:23.000000000 +0100
@@ -31377,9 +31377,9 @@ ix86_expand_vector_set (bool mmx_ok, rtx
 	  tmp = gen_reg_rtx (GET_MODE_INNER (mode));
 	  ix86_expand_vector_extract (true, tmp, target, 1 - elt);
 	  if (elt == 0)
-	    tmp = gen_rtx_VEC_CONCAT (mode, tmp, val);
-	  else
 	    tmp = gen_rtx_VEC_CONCAT (mode, val, tmp);
+	  else
+	    tmp = gen_rtx_VEC_CONCAT (mode, tmp, val);
 	  emit_insn (gen_rtx_SET (VOIDmode, target, tmp));
 	  return;
 	}
@@ -31393,9 +31393,9 @@ ix86_expand_vector_set (bool mmx_ok, rtx
       tmp = gen_reg_rtx (GET_MODE_INNER (mode));
       ix86_expand_vector_extract (false, tmp, target, 1 - elt);
       if (elt == 0)
-	tmp = gen_rtx_VEC_CONCAT (mode, tmp, val);
-      else
 	tmp = gen_rtx_VEC_CONCAT (mode, val, tmp);
+      else
+	tmp = gen_rtx_VEC_CONCAT (mode, tmp, val);
       emit_insn (gen_rtx_SET (VOIDmode, target, tmp));
       return;
 
--- gcc-4.6.3/gcc/testsuite/gcc.dg/torture/pr52407.c.~1~	1970-01-01 01:00:00.000000000 +0100
+++ gcc-4.6.3/gcc/testsuite/gcc.dg/torture/pr52407.c	2012-03-04 17:58:23.000000000 +0100
@@ -0,0 +1,33 @@
+/* { dg-do run } */
+
+extern void abort (void);
+
+typedef long long T;
+typedef T vl_t __attribute__((vector_size(2 * sizeof (T))));
+
+vl_t	ul[4], vl[4] = { { 1, 2 }, { 3, 4 }, { 5, 6 }, { 7, 8 } };
+
+static void
+mul_vl_l(vl_t *u, vl_t *v, T x, int m)
+{
+  vl_t	 w;
+  T *p = (T *)&w;
+  p[0] = p[1] = x;
+  while (m--)
+    *u++ = *v++ * w;
+}
+
+int
+main(int argc, char *argv[])
+{
+  int i;
+  T *pl;
+
+  pl = (T *) &ul;
+  mul_vl_l(ul, vl, 2, 4);
+  for (i = 0; i < 8; i++)
+    if (pl[i] != 2 * (i + 1))
+      abort ();
+
+  return 0;
+}