aarch64: Use memcpy to copy vector tables in vtbx4 intrinsics

Use __builtin_memcpy to copy vector structures instead of building
a new opaque structure one vector at a time in each of the vtbx4
Neon intrinsics in arm_neon.h. This simplifies the header file and
also improves code generation - superfluous move instructions were
emitted for every register extraction/set in this additional
structure.

gcc/ChangeLog:

2021-07-19  Jonathan Wright  <jonathan.wright@arm.com>

	* config/aarch64/arm_neon.h (vtbx4_s8): Use __builtin_memcpy
	instead of constructing __builtin_aarch64_simd_oi one vector
	at a time.
	(vtbx4_u8): Likewise.
	(vtbx4_p8): Likewise.

diff --git a/gcc/config/aarch64/arm_neon.h b/gcc/config/aarch64/arm_neon.h
index 0ec46efe7f633bf1a56d8eb09d43105c0d40eabd..d383af30fe701deb0b8a2ac327e9da2cae0bff78 100644 (file)
--- a/gcc/config/aarch64/arm_neon.h
+++ b/gcc/config/aarch64/arm_neon.h
@@ -28417,10 +28417,7 @@ vtbx4_s8 (int8x8_t __r, int8x8x4_t __tab, int8x8_t __idx)
   __builtin_aarch64_simd_oi __o;
   __temp.val[0] = vcombine_s8 (__tab.val[0], __tab.val[1]);
   __temp.val[1] = vcombine_s8 (__tab.val[2], __tab.val[3]);
-  __o = __builtin_aarch64_set_qregoiv16qi (__o,
-                                          (int8x16_t) __temp.val[0], 0);
-  __o = __builtin_aarch64_set_qregoiv16qi (__o,
-                                          (int8x16_t) __temp.val[1], 1);
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
   return __builtin_aarch64_qtbx2v8qi (__r, __o, __idx);
 }
 
@@ -28432,10 +28429,7 @@ vtbx4_u8 (uint8x8_t __r, uint8x8x4_t __tab, uint8x8_t __idx)
   __builtin_aarch64_simd_oi __o;
   __temp.val[0] = vcombine_u8 (__tab.val[0], __tab.val[1]);
   __temp.val[1] = vcombine_u8 (__tab.val[2], __tab.val[3]);
-  __o = __builtin_aarch64_set_qregoiv16qi (__o,
-                                          (int8x16_t) __temp.val[0], 0);
-  __o = __builtin_aarch64_set_qregoiv16qi (__o,
-                                          (int8x16_t) __temp.val[1], 1);
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
   return (uint8x8_t)__builtin_aarch64_qtbx2v8qi ((int8x8_t)__r, __o,
                                                 (int8x8_t)__idx);
 }
@@ -28448,10 +28442,7 @@ vtbx4_p8 (poly8x8_t __r, poly8x8x4_t __tab, uint8x8_t __idx)
   __builtin_aarch64_simd_oi __o;
   __temp.val[0] = vcombine_p8 (__tab.val[0], __tab.val[1]);
   __temp.val[1] = vcombine_p8 (__tab.val[2], __tab.val[3]);
-  __o = __builtin_aarch64_set_qregoiv16qi (__o,
-                                          (int8x16_t) __temp.val[0], 0);
-  __o = __builtin_aarch64_set_qregoiv16qi (__o,
-                                          (int8x16_t) __temp.val[1], 1);
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
   return (poly8x8_t)__builtin_aarch64_qtbx2v8qi ((int8x8_t)__r, __o,
                                                 (int8x8_t)__idx);
 }