[AArch64][1/2] Add support INS (element) instruction to copy lanes between vectors

2016-06-30  James Greenhalgh  <james.greenhalgh@arm.com>
            Kyrylo Tkachov  <kyrylo.tkachov@arm.com>

	* config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
	New define_insn.
	(*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.

	* gcc.target/aarch64/vget_set_lane_1.c: New test.

Co-Authored-By: Kyrylo Tkachov <kyrylo.tkachov@arm.com>
From-SVN: r237882

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 28b938707318fb7ba225bcd349a10f0f9fa3eb87..1d807ce5625b7a5f57d40d74752f101f3f4938d6 100644 (file)
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,10 @@
+2016-06-30  James Greenhalgh  <james.greenhalgh@arm.com>
+            Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
+
+       * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
+       New define_insn.
+       (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
+
 2016-06-30  David Malcolm  <dmalcolm@redhat.com>
 
        PR driver/71651

diff --git a/gcc/config/aarch64/aarch64-simd.md b/gcc/config/aarch64/aarch64-simd.md
index 3f8289cf7dc398faf656587c2b8f25507e616be9..a19d1711b5bcb516e4aca6a22d1b79df4f32923f 100644 (file)
--- a/gcc/config/aarch64/aarch64-simd.md
+++ b/gcc/config/aarch64/aarch64-simd.md
@@ -546,6 +546,49 @@
   [(set_attr "type" "neon_from_gp<q>, neon_ins<q>, neon_load1_1reg<q>")]
 )
 
+(define_insn "*aarch64_simd_vec_copy_lane<mode>"
+  [(set (match_operand:VALL 0 "register_operand" "=w")
+       (vec_merge:VALL
+           (vec_duplicate:VALL
+             (vec_select:<VEL>
+               (match_operand:VALL 3 "register_operand" "w")
+               (parallel
+                 [(match_operand:SI 4 "immediate_operand" "i")])))
+           (match_operand:VALL 1 "register_operand" "0")
+           (match_operand:SI 2 "immediate_operand" "i")))]
+  "TARGET_SIMD"
+  {
+    int elt = ENDIAN_LANE_N (<MODE>mode, exact_log2 (INTVAL (operands[2])));
+    operands[2] = GEN_INT (HOST_WIDE_INT_1 << elt);
+    operands[4] = GEN_INT (ENDIAN_LANE_N (<MODE>mode, INTVAL (operands[4])));
+
+    return "ins\t%0.<Vetype>[%p2], %3.<Vetype>[%4]";
+  }
+  [(set_attr "type" "neon_ins<q>")]
+)
+
+(define_insn "*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>"
+  [(set (match_operand:VALL 0 "register_operand" "=w")
+       (vec_merge:VALL
+           (vec_duplicate:VALL
+             (vec_select:<VEL>
+               (match_operand:<VSWAP_WIDTH> 3 "register_operand" "w")
+               (parallel
+                 [(match_operand:SI 4 "immediate_operand" "i")])))
+           (match_operand:VALL 1 "register_operand" "0")
+           (match_operand:SI 2 "immediate_operand" "i")))]
+  "TARGET_SIMD"
+  {
+    int elt = ENDIAN_LANE_N (<MODE>mode, exact_log2 (INTVAL (operands[2])));
+    operands[2] = GEN_INT (HOST_WIDE_INT_1 << elt);
+    operands[4] = GEN_INT (ENDIAN_LANE_N (<VSWAP_WIDTH>mode,
+                          INTVAL (operands[4])));
+
+    return "ins\t%0.<Vetype>[%p2], %3.<Vetype>[%4]";
+  }
+  [(set_attr "type" "neon_ins<q>")]
+)
+
 (define_insn "aarch64_simd_lshr<mode>"
  [(set (match_operand:VDQ_I 0 "register_operand" "=w")
        (lshiftrt:VDQ_I (match_operand:VDQ_I 1 "register_operand" "w")

diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog
index 8527a2fed0142a4cdbb1d63fb0f227b118dd8306..1bd8ef6688cf8c3ebcc05a161d2694867c556e92 100644 (file)
--- a/gcc/testsuite/ChangeLog
+++ b/gcc/testsuite/ChangeLog
@@ -1,3 +1,8 @@
+2016-06-30  James Greenhalgh  <james.greenhalgh@arm.com>
+            Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
+
+       * gcc.target/aarch64/vget_set_lane_1.c: New test.
+
 2016-06-30  David Malcolm  <dmalcolm@redhat.com>
 
        PR driver/71651

diff --git a/gcc/testsuite/gcc.target/aarch64/vget_set_lane_1.c b/gcc/testsuite/gcc.target/aarch64/vget_set_lane_1.c
new file mode 100644 (file)
index 0000000..07a77de
--- /dev/null
+++ b/gcc/testsuite/gcc.target/aarch64/vget_set_lane_1.c
@@ -0,0 +1,72 @@
+/* { dg-do compile } */
+/* { dg-options "-O2" } */
+
+#include "arm_neon.h"
+
+#define BUILD_TEST(TYPE1, TYPE2, Q1, Q2, SUFFIX, INDEX1, INDEX2)       \
+TYPE1 __attribute__((noinline,noclone))                                \
+test_copy##Q1##_lane##Q2##_##SUFFIX (TYPE1 a, TYPE2 b)                 \
+{                                                                      \
+  return vset##Q1##_lane_##SUFFIX (vget##Q2##_lane_##SUFFIX (b, INDEX2),\
+                                   a, INDEX1);                         \
+}
+
+BUILD_TEST (poly8x8_t, poly8x8_t, , , p8, 7, 6)
+BUILD_TEST (int8x8_t,  int8x8_t,  , , s8, 7, 6)
+BUILD_TEST (uint8x8_t, uint8x8_t, , , u8, 7, 6)
+/* { dg-final { scan-assembler-times "ins\\tv0.b\\\[7\\\], v1.b\\\[6\\\]" 3 } } */
+BUILD_TEST (poly16x4_t, poly16x4_t, , , p16, 3, 2)
+BUILD_TEST (int16x4_t,  int16x4_t,  , , s16, 3, 2)
+BUILD_TEST (uint16x4_t, uint16x4_t, , , u16, 3, 2)
+/* { dg-final { scan-assembler-times "ins\\tv0.h\\\[3\\\], v1.h\\\[2\\\]" 3 } } */
+BUILD_TEST (float32x2_t, float32x2_t, , , f32, 1, 0)
+BUILD_TEST (int32x2_t,   int32x2_t,   , , s32, 1, 0)
+BUILD_TEST (uint32x2_t,  uint32x2_t,  , , u32, 1, 0)
+/* { dg-final { scan-assembler-times "ins\\tv0.s\\\[1\\\], v1.s\\\[0\\\]" 3 } } */
+
+BUILD_TEST (poly8x8_t, poly8x16_t, , q, p8, 7, 15)
+BUILD_TEST (int8x8_t,  int8x16_t,  , q, s8, 7, 15)
+BUILD_TEST (uint8x8_t, uint8x16_t, , q, u8, 7, 15)
+/* { dg-final { scan-assembler-times "ins\\tv0.b\\\[7\\\], v1.b\\\[15\\\]" 3 } } */
+BUILD_TEST (poly16x4_t, poly16x8_t, , q, p16, 3, 7)
+BUILD_TEST (int16x4_t,  int16x8_t,  , q, s16, 3, 7)
+BUILD_TEST (uint16x4_t, uint16x8_t, , q, u16, 3, 7)
+/* { dg-final { scan-assembler-times "ins\\tv0.h\\\[3\\\], v1.h\\\[7\\\]" 3 } } */
+BUILD_TEST (float32x2_t, float32x4_t, , q, f32, 1, 3)
+BUILD_TEST (int32x2_t,   int32x4_t,   , q, s32, 1, 3)
+BUILD_TEST (uint32x2_t,  uint32x4_t,  , q, u32, 1, 3)
+/* { dg-final { scan-assembler-times "ins\\tv0.s\\\[1\\\], v1.s\\\[3\\\]" 3 } } */
+
+BUILD_TEST (poly8x16_t, poly8x8_t, q, , p8, 15, 7)
+BUILD_TEST (int8x16_t,  int8x8_t,  q, , s8, 15, 7)
+BUILD_TEST (uint8x16_t, uint8x8_t, q, , u8, 15, 7)
+/* { dg-final { scan-assembler-times "ins\\tv0.b\\\[15\\\], v1.b\\\[7\\\]" 3 } } */
+BUILD_TEST (poly16x8_t, poly16x4_t, q, , p16, 7, 3)
+BUILD_TEST (int16x8_t,  int16x4_t,  q, , s16, 7, 3)
+BUILD_TEST (uint16x8_t, uint16x4_t, q, , u16, 7, 3)
+/* { dg-final { scan-assembler-times "ins\\tv0.h\\\[7\\\], v1.h\\\[3\\\]" 3 } } */
+BUILD_TEST (float32x4_t, float32x2_t, q, , f32, 3, 1)
+BUILD_TEST (int32x4_t,   int32x2_t,   q, , s32, 3, 1)
+BUILD_TEST (uint32x4_t,  uint32x2_t,  q, , u32, 3, 1)
+/* { dg-final { scan-assembler-times "ins\\tv0.s\\\[3\\\], v1.s\\\[1\\\]" 3 } } */
+BUILD_TEST (float64x2_t, float64x1_t, q, , f64, 1, 0)
+BUILD_TEST (int64x2_t,   int64x1_t,   q, , s64, 1, 0)
+BUILD_TEST (uint64x2_t,  uint64x1_t,  q, , u64, 1, 0)
+/* { dg-final { scan-assembler-times "ins\\tv0.d\\\[1\\\], v1.d\\\[0\\\]" 3 } } */
+
+BUILD_TEST (poly8x16_t, poly8x16_t, q, q, p8, 14, 15)
+BUILD_TEST (int8x16_t,  int8x16_t,  q, q, s8, 14, 15)
+BUILD_TEST (uint8x16_t, uint8x16_t, q, q, u8, 14, 15)
+/* { dg-final { scan-assembler-times "ins\\tv0.b\\\[14\\\], v1.b\\\[15\\\]" 3 } } */
+BUILD_TEST (poly16x8_t, poly16x8_t, q, q, p16, 6, 7)
+BUILD_TEST (int16x8_t,  int16x8_t,  q, q, s16, 6, 7)
+BUILD_TEST (uint16x8_t, uint16x8_t, q, q, u16, 6, 7)
+/* { dg-final { scan-assembler-times "ins\\tv0.h\\\[6\\\], v1.h\\\[7\\\]" 3 } } */
+BUILD_TEST (float32x4_t, float32x4_t, q, q, f32, 2, 3)
+BUILD_TEST (int32x4_t,   int32x4_t,   q, q, s32, 2, 3)
+BUILD_TEST (uint32x4_t,  uint32x4_t,  q, q, u32, 2, 3)
+/* { dg-final { scan-assembler-times "ins\\tv0.s\\\[2\\\], v1.s\\\[3\\\]" 3 } } */
+BUILD_TEST (float64x2_t, float64x2_t, q, q, f64, 1, 1)
+BUILD_TEST (int64x2_t,   int64x2_t,   q,  q, s64, 1, 1)
+BUILD_TEST (uint64x2_t,  uint64x2_t,  q, q, u64, 1, 1)
+/* { dg-final { scan-assembler-times "ins\\tv0.d\\\[1\\\], v1.d\\\[1\\\]" 3 } } */