aarch64: Use RTL builtins for vpaddq intrinsics

Rewrite vpaddq Neon intrinsics to use RTL builtins rather than inline
assembly code, allowing for better scheduling and optimization.

gcc/ChangeLog:

2021-02-08  Jonathan Wright  <jonathan.wright@arm.com>

	* config/aarch64/aarch64-simd-builtins.def: Use VDQ_I iterator
	for aarch64_addp<mode> builtin macro generator.
	* config/aarch64/aarch64-simd.md: Use VDQ_I iterator in
	aarch64_addp<mode> RTL pattern.
	* config/aarch64/arm_neon.h (vpaddq_s8): Use RTL builtin
	instead of inline asm.
	(vpaddq_s16): Likewise.
	(vpaddq_s32): Likewise.
	(vpaddq_s64): Likewise.
	(vpaddq_u8): Likewise.
	(vpaddq_u16): Likewise.
	(vpaddq_u32): Likewise.
	(vpaddq_u64): Likewise.

3 files changed, 17 insertions, 53 deletions

diff --git a/gcc/config/aarch64/aarch64-simd-builtins.def b/gcc/config/aarch64/aarch64-simd-builtins.def
index f79e716..92804e0 100644
--- a/gcc/config/aarch64/aarch64-simd-builtins.def
+++ b/gcc/config/aarch64/aarch64-simd-builtins.def
@@ -48,7 +48,7 @@
   BUILTIN_VB (BINOP, pmul, 0, NONE)
   BUILTIN_VHSDF_HSDF (BINOP, fmulx, 0, FP)
   BUILTIN_VHSDF_DF (UNOP, sqrt, 2, FP)
-  BUILTIN_VD_BHSI (BINOP, addp, 0, NONE)
+  BUILTIN_VDQ_I (BINOP, addp, 0, NONE)
   VAR1 (UNOP, addp, 0, NONE, di)
   BUILTIN_VDQ_BHSI (UNOP, clrsb, 2, NONE)
   BUILTIN_VDQ_BHSI (UNOP, clz, 2, NONE)
diff --git a/gcc/config/aarch64/aarch64-simd.md b/gcc/config/aarch64/aarch64-simd.md
index 5245cf0..60e11c6 100644
--- a/gcc/config/aarch64/aarch64-simd.md
+++ b/gcc/config/aarch64/aarch64-simd.md
@@ -6004,10 +6004,10 @@
 ;; addp
 
 (define_insn "aarch64_addp<mode>"
-  [(set (match_operand:VD_BHSI 0 "register_operand" "=w")
-        (unspec:VD_BHSI
-          [(match_operand:VD_BHSI 1 "register_operand" "w")
-	   (match_operand:VD_BHSI 2 "register_operand" "w")]
+  [(set (match_operand:VDQ_I 0 "register_operand" "=w")
+        (unspec:VDQ_I
+          [(match_operand:VDQ_I 1 "register_operand" "w")
+	   (match_operand:VDQ_I 2 "register_operand" "w")]
           UNSPEC_ADDP))]
   "TARGET_SIMD"
   "addp\t%<v>0<Vmtype>, %<v>1<Vmtype>, %<v>2<Vmtype>"
diff --git a/gcc/config/aarch64/arm_neon.h b/gcc/config/aarch64/arm_neon.h
index 5fb2b3d..52f3714 100644
--- a/gcc/config/aarch64/arm_neon.h
+++ b/gcc/config/aarch64/arm_neon.h
@@ -8673,96 +8673,60 @@ __extension__ extern __inline int8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 vpaddq_s8 (int8x16_t __a, int8x16_t __b)
 {
-  int8x16_t __result;
-  __asm__ ("addp %0.16b,%1.16b,%2.16b"
-           : "=w"(__result)
-           : "w"(__a), "w"(__b)
-           : /* No clobbers */);
-  return __result;
+  return __builtin_aarch64_addpv16qi (__a, __b);
 }
 
 __extension__ extern __inline int16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 vpaddq_s16 (int16x8_t __a, int16x8_t __b)
 {
-  int16x8_t __result;
-  __asm__ ("addp %0.8h,%1.8h,%2.8h"
-           : "=w"(__result)
-           : "w"(__a), "w"(__b)
-           : /* No clobbers */);
-  return __result;
+  return __builtin_aarch64_addpv8hi (__a, __b);
 }
 
 __extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 vpaddq_s32 (int32x4_t __a, int32x4_t __b)
 {
-  int32x4_t __result;
-  __asm__ ("addp %0.4s,%1.4s,%2.4s"
-           : "=w"(__result)
-           : "w"(__a), "w"(__b)
-           : /* No clobbers */);
-  return __result;
+  return __builtin_aarch64_addpv4si (__a, __b);
 }
 
 __extension__ extern __inline int64x2_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 vpaddq_s64 (int64x2_t __a, int64x2_t __b)
 {
-  int64x2_t __result;
-  __asm__ ("addp %0.2d,%1.2d,%2.2d"
-           : "=w"(__result)
-           : "w"(__a), "w"(__b)
-           : /* No clobbers */);
-  return __result;
+  return __builtin_aarch64_addpv2di (__a, __b);
 }
 
 __extension__ extern __inline uint8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 vpaddq_u8 (uint8x16_t __a, uint8x16_t __b)
 {
-  uint8x16_t __result;
-  __asm__ ("addp %0.16b,%1.16b,%2.16b"
-           : "=w"(__result)
-           : "w"(__a), "w"(__b)
-           : /* No clobbers */);
-  return __result;
+  return (uint8x16_t) __builtin_aarch64_addpv16qi ((int8x16_t) __a,
+						   (int8x16_t) __b);
 }
 
 __extension__ extern __inline uint16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 vpaddq_u16 (uint16x8_t __a, uint16x8_t __b)
 {
-  uint16x8_t __result;
-  __asm__ ("addp %0.8h,%1.8h,%2.8h"
-           : "=w"(__result)
-           : "w"(__a), "w"(__b)
-           : /* No clobbers */);
-  return __result;
+  return (uint16x8_t) __builtin_aarch64_addpv8hi ((int16x8_t) __a,
+						  (int16x8_t) __b);
 }
 
 __extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 vpaddq_u32 (uint32x4_t __a, uint32x4_t __b)
 {
-  uint32x4_t __result;
-  __asm__ ("addp %0.4s,%1.4s,%2.4s"
-           : "=w"(__result)
-           : "w"(__a), "w"(__b)
-           : /* No clobbers */);
-  return __result;
+  return (uint32x4_t) __builtin_aarch64_addpv4si ((int32x4_t) __a,
+						  (int32x4_t) __b);
 }
 
 __extension__ extern __inline uint64x2_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 vpaddq_u64 (uint64x2_t __a, uint64x2_t __b)
 {
-  uint64x2_t __result;
-  __asm__ ("addp %0.2d,%1.2d,%2.2d"
-           : "=w"(__result)
-           : "w"(__a), "w"(__b)
-           : /* No clobbers */);
-  return __result;
+  return (uint64x2_t) __builtin_aarch64_addpv2di ((int64x2_t) __a,
+						  (int64x2_t) __b);
 }
 
 __extension__ extern __inline int16x4_t