Implement AArch64 SISD intrinsics for vget_high and vget_low.

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@195073 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/include/clang/Basic/arm_neon.td b/include/clang/Basic/arm_neon.td
index ccca04e7b6a94e9b98239fb9c3f18ae095012b3e..e83978cb130f3e2d91d82ea29cf54da742da00be 100644 (file)
--- a/include/clang/Basic/arm_neon.td
+++ b/include/clang/Basic/arm_neon.td
@@ -979,6 +979,13 @@ def SCALAR_SUB : SInst<"vsub", "sss",  "SlSUl">;
 // Scalar  Saturating Sub
 def SCALAR_QSUB   : SInst<"vqsub", "sss", "ScSsSiSlSUcSUsSUiSUl">;
 
+let InstName = "vmov" in {
+def VGET_HIGH_A64 : NoTestOpInst<"vget_high", "dk", "csilhfdUcUsUiUlPcPsPl",
+                                 OP_HI>;
+def VGET_LOW_A64  : NoTestOpInst<"vget_low", "dk", "csilhfdUcUsUiUlPcPsPl",
+                                 OP_LO>;
+}
+
 ////////////////////////////////////////////////////////////////////////////////
 // Scalar Shift
 // Scalar Shift Left

diff --git a/test/CodeGen/aarch64-neon-vget-hilo.c b/test/CodeGen/aarch64-neon-vget-hilo.c
new file mode 100644 (file)
index 0000000..9edd31a
--- /dev/null
+++ b/test/CodeGen/aarch64-neon-vget-hilo.c
@@ -0,0 +1,176 @@
+// REQUIRES: aarch64-registered-target
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +neon \
+// RUN:   -ffp-contract=fast -S -O3 -o - %s | FileCheck %s
+
+// Test new aarch64 intrinsics and types
+
+#include <arm_neon.h>
+
+int8x8_t test_vget_high_s8(int8x16_t a) {
+   // CHECK: test_vget_high_s8
+  return vget_high_s8(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+int16x4_t test_vget_high_s16(int16x8_t a) {
+   // CHECK: test_vget_high_s16
+  return vget_high_s16(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+int32x2_t test_vget_high_s32(int32x4_t a) {
+   // CHECK: test_vget_high_s32
+  return vget_high_s32(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+int64x1_t test_vget_high_s64(int64x2_t a) {
+   // CHECK: test_vget_high_s64
+  return vget_high_s64(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+uint8x8_t test_vget_high_u8(uint8x16_t a) {
+   // CHECK: test_vget_high_u8
+  return vget_high_u8(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+uint16x4_t test_vget_high_u16(uint16x8_t a) {
+   // CHECK: test_vget_high_u16
+  return vget_high_u16(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+uint32x2_t test_vget_high_u32(uint32x4_t a) {
+   // CHECK: test_vget_high_u32
+  return vget_high_u32(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+uint64x1_t test_vget_high_u64(uint64x2_t a) {
+   // CHECK: test_vget_high_u64
+  return vget_high_u64(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+poly64x1_t test_vget_high_p64(poly64x2_t a) {
+   // CHECK: test_vget_high_p64
+  return vget_high_p64(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+float16x4_t test_vget_high_f16(float16x8_t a) {
+   // CHECK: test_vget_high_f16
+  return vget_high_f16(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+float32x2_t test_vget_high_f32(float32x4_t a) {
+   // CHECK: test_vget_high_f32
+  return vget_high_f32(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+poly8x8_t test_vget_high_p8(poly8x16_t a) {
+   // CHECK: test_vget_high_p8
+  return vget_high_p8(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+poly16x4_t test_vget_high_p16(poly16x8_t a) {
+   // CHECK: test_vget_high_p16
+  return vget_high_p16(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+float64x1_t test_vget_high_f64(float64x2_t a) {
+   // CHECK: test_vget_high_f64
+  return vget_high_f64(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+int8x8_t test_vget_low_s8(int8x16_t a) {
+   // CHECK: test_vget_low_s8
+  return vget_low_s8(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+int16x4_t test_vget_low_s16(int16x8_t a) {
+   // CHECK: test_vget_low_s16
+  return vget_low_s16(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+int32x2_t test_vget_low_s32(int32x4_t a) {
+   // CHECK: test_vget_low_s32
+  return vget_low_s32(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+int64x1_t test_vget_low_s64(int64x2_t a) {
+   // CHECK: test_vget_low_s64
+  return vget_low_s64(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+uint8x8_t test_vget_low_u8(uint8x16_t a) {
+   // CHECK: test_vget_low_u8
+  return vget_low_u8(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+uint16x4_t test_vget_low_u16(uint16x8_t a) {
+   // CHECK: test_vget_low_u16
+  return vget_low_u16(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+uint32x2_t test_vget_low_u32(uint32x4_t a) {
+   // CHECK: test_vget_low_u32
+  return vget_low_u32(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+uint64x1_t test_vget_low_u64(uint64x2_t a) {
+   // CHECK: test_vget_low_u64
+  return vget_low_u64(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+poly64x1_t test_vget_low_p64(poly64x2_t a) {
+   // CHECK: test_vget_low_p64
+  return vget_low_p64(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+float16x4_t test_vget_low_f16(float16x8_t a) {
+   // CHECK: test_vget_low_f16
+  return vget_low_f16(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+float32x2_t test_vget_low_f32(float32x4_t a) {
+   // CHECK: test_vget_low_f32
+  return vget_low_f32(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+poly8x8_t test_vget_low_p8(poly8x16_t a) {
+   // CHECK: test_vget_low_p8
+  return vget_low_p8(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+poly16x4_t test_vget_low_p16(poly16x8_t a) {
+   // CHECK: test_vget_low_p16
+  return vget_low_p16(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+float64x1_t test_vget_low_f64(float64x2_t a) {
+   // CHECK: test_vget_low_f64
+  return vget_low_f64(a);
+  // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+