Don't force vget_lane returning a 64-bit result to transfer to core

registers.

Backport from mainline r185603:

gcc/
2012-03-21  Richard Earnshaw  <rearnsha@arm.com>

5 files changed, 29 insertions, 14 deletions

diff --git a/gcc/config/arm/neon-testgen.ml b/gcc/config/arm/neon-testgen.ml
index eb2917da9b1..a69a53917b1 100644
--- a/gcc/config/arm/neon-testgen.ml
+++ b/gcc/config/arm/neon-testgen.ml
@@ -79,9 +79,12 @@ let emit_automatics chan c_types features =
           (* The intrinsic returns a value.  We need to do explict register
              allocation for vget_low tests or they fail because of copy
              elimination.  *)
-          ((if List.mem Fixed_return_reg features then
+          ((if List.mem Fixed_vector_reg features then
               Printf.fprintf chan "  register %s out_%s asm (\"d18\");\n"
                              return_ty return_ty
+            else if List.mem Fixed_core_reg features then
+              Printf.fprintf chan "  register %s out_%s asm (\"r0\");\n"
+                             return_ty return_ty
             else
               Printf.fprintf chan "  %s out_%s;\n" return_ty return_ty);
 	   emit ())
diff --git a/gcc/config/arm/neon.md b/gcc/config/arm/neon.md
index 078a8fd47ee..9a523751843 100644
--- a/gcc/config/arm/neon.md
+++ b/gcc/config/arm/neon.md
@@ -2720,14 +2720,24 @@
 })
 
 (define_expand "neon_vget_lanev2di"
-  [(match_operand:DI 0 "s_register_operand" "=r")
-   (match_operand:V2DI 1 "s_register_operand" "w")
-   (match_operand:SI 2 "immediate_operand" "i")
-   (match_operand:SI 3 "immediate_operand" "i")]
+  [(match_operand:DI 0 "s_register_operand" "")
+   (match_operand:V2DI 1 "s_register_operand" "")
+   (match_operand:SI 2 "immediate_operand" "")
+   (match_operand:SI 3 "immediate_operand" "")]
   "TARGET_NEON"
 {
-  neon_lane_bounds (operands[2], 0, 2);
-  emit_insn (gen_vec_extractv2di (operands[0], operands[1], operands[2]));
+  switch (INTVAL (operands[2]))
+    {
+    case 0:
+      emit_move_insn (operands[0], gen_lowpart (DImode, operands[1]));
+      break;
+    case 1:
+      emit_move_insn (operands[0], gen_highpart (DImode, operands[1]));
+      break;
+    default:
+      neon_lane_bounds (operands[2], 0, 1);
+      FAIL;
+    }
   DONE;
 })
 
diff --git a/gcc/config/arm/neon.ml b/gcc/config/arm/neon.ml
index 363e55c713c..63389a3ecb6 100644
--- a/gcc/config/arm/neon.ml
+++ b/gcc/config/arm/neon.ml
@@ -234,7 +234,8 @@ type features =
        cases.  The function supplied must return the integer to be written
        into the testcase for the argument number (0-based) supplied to it.  *)
   | Const_valuator of (int -> int)
-  | Fixed_return_reg
+  | Fixed_vector_reg
+  | Fixed_core_reg
 
 exception MixedMode of elts * elts
 
@@ -999,7 +1000,8 @@ let ops =
     Vget_lane,
       [InfoWord;
        Disassembles_as [Use_operands [| Corereg; Corereg; Dreg |]];
-       Instruction_name ["vmov"]; Const_valuator (fun _ -> 0)],
+       Instruction_name ["vmov"; "fmrrd"]; Const_valuator (fun _ -> 0);
+       Fixed_core_reg],
       Use_operands [| Corereg; Qreg; Immed |],
       "vgetQ_lane", notype_2, [S64; U64];
 
@@ -1115,7 +1117,7 @@ let ops =
       notype_1, pf_su_8_64;
     Vget_low, [Instruction_name ["vmov"];
                Disassembles_as [Use_operands [| Dreg; Dreg |]];
-	       Fixed_return_reg],
+	       Fixed_vector_reg],
       Use_operands [| Dreg; Qreg |], "vget_low",
       notype_1, pf_su_8_32;
      Vget_low, [No_op],
diff --git a/gcc/testsuite/gcc.target/arm/neon/vgetQ_lanes64.c b/gcc/testsuite/gcc.target/arm/neon/vgetQ_lanes64.c
index b7f7f33502a..e3d3c178e50 100644
--- a/gcc/testsuite/gcc.target/arm/neon/vgetQ_lanes64.c
+++ b/gcc/testsuite/gcc.target/arm/neon/vgetQ_lanes64.c
@@ -10,11 +10,11 @@
 
 void test_vgetQ_lanes64 (void)
 {
-  int64_t out_int64_t;
+  register int64_t out_int64_t asm ("r0");
   int64x2_t arg0_int64x2_t;
 
   out_int64_t = vgetq_lane_s64 (arg0_int64x2_t, 0);
 }
 
-/* { dg-final { scan-assembler "vmov\[ 	\]+\[rR\]\[0-9\]+, \[rR\]\[0-9\]+, \[dD\]\[0-9\]+!?\(\[ 	\]+@\[a-zA-Z0-9 \]+\)?\n" } } */
+/* { dg-final { scan-assembler "((vmov)|(fmrrd))\[ 	\]+\[rR\]\[0-9\]+, \[rR\]\[0-9\]+, \[dD\]\[0-9\]+!?\(\[ 	\]+@\[a-zA-Z0-9 \]+\)?\n" } } */
 /* { dg-final { cleanup-saved-temps } } */
diff --git a/gcc/testsuite/gcc.target/arm/neon/vgetQ_laneu64.c b/gcc/testsuite/gcc.target/arm/neon/vgetQ_laneu64.c
index 33c463e6409..3426e469487 100644
--- a/gcc/testsuite/gcc.target/arm/neon/vgetQ_laneu64.c
+++ b/gcc/testsuite/gcc.target/arm/neon/vgetQ_laneu64.c
@@ -10,11 +10,11 @@
 
 void test_vgetQ_laneu64 (void)
 {
-  uint64_t out_uint64_t;
+  register uint64_t out_uint64_t asm ("r0");
   uint64x2_t arg0_uint64x2_t;
 
   out_uint64_t = vgetq_lane_u64 (arg0_uint64x2_t, 0);
 }
 
-/* { dg-final { scan-assembler "vmov\[ 	\]+\[rR\]\[0-9\]+, \[rR\]\[0-9\]+, \[dD\]\[0-9\]+!?\(\[ 	\]+@\[a-zA-Z0-9 \]+\)?\n" } } */
+/* { dg-final { scan-assembler "((vmov)|(fmrrd))\[ 	\]+\[rR\]\[0-9\]+, \[rR\]\[0-9\]+, \[dD\]\[0-9\]+!?\(\[ 	\]+@\[a-zA-Z0-9 \]+\)?\n" } } */
 /* { dg-final { cleanup-saved-temps } } */