aarch64: Use memcpy to copy vector tables in vst1[q]_x4 intrinsics

Use __builtin_memcpy to copy vector structures instead of using a
union in each of the vst1[q]_x4 Neon intrinsics in arm_neon.h.

Add new code generation tests to verify that superfluous move
instructions are not generated for the vst1q_x4 intrinsics.

gcc/ChangeLog:

2021-07-21  Jonathan Wright  <jonathan.wright@arm.com>

	* config/aarch64/arm_neon.h (vst1_s8_x4): Use
	__builtin_memcpy instead of using a union.
	(vst1q_s8_x4): Likewise.
	(vst1_s16_x4): Likewise.
	(vst1q_s16_x4): Likewise.
	(vst1_s32_x4): Likewise.
	(vst1q_s32_x4): Likewise.
	(vst1_u8_x4): Likewise.
	(vst1q_u8_x4): Likewise.
	(vst1_u16_x4): Likewise.
	(vst1q_u16_x4): Likewise.
	(vst1_u32_x4): Likewise.
	(vst1q_u32_x4): Likewise.
	(vst1_f16_x4): Likewise.
	(vst1q_f16_x4): Likewise.
	(vst1_f32_x4): Likewise.
	(vst1q_f32_x4): Likewise.
	(vst1_p8_x4): Likewise.
	(vst1q_p8_x4): Likewise.
	(vst1_p16_x4): Likewise.
	(vst1q_p16_x4): Likewise.
	(vst1_s64_x4): Likewise.
	(vst1_u64_x4): Likewise.
	(vst1_p64_x4): Likewise.
	(vst1q_s64_x4): Likewise.
	(vst1q_u64_x4): Likewise.
	(vst1q_p64_x4): Likewise.
	(vst1_f64_x4): Likewise.
	(vst1q_f64_x4): Likewise.

gcc/testsuite/ChangeLog:

	* gcc.target/aarch64/vector_structure_intrinsics.c: Add new
	tests.

2 files changed, 204 insertions, 84 deletions

diff --git a/gcc/config/aarch64/arm_neon.h b/gcc/config/aarch64/arm_neon.h
index 0e4ab35..9cf16a8 100644
--- a/gcc/config/aarch64/arm_neon.h
+++ b/gcc/config/aarch64/arm_neon.h
@@ -26984,226 +26984,324 @@ vst1q_p64_x3 (poly64_t * __a, poly64x2x3_t __val)
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1_s8_x4 (int8_t * __a, int8x8x4_t val)
+vst1_s8_x4 (int8_t * __a, int8x8x4_t __val)
 {
-  union { int8x8x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v8qi ((__builtin_aarch64_simd_qi *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  int8x16x4_t __temp;
+  __temp.val[0] = vcombine_s8 (__val.val[0], vcreate_s8 (__AARCH64_INT64_C (0)));
+  __temp.val[1] = vcombine_s8 (__val.val[1], vcreate_s8 (__AARCH64_INT64_C (0)));
+  __temp.val[2] = vcombine_s8 (__val.val[2], vcreate_s8 (__AARCH64_INT64_C (0)));
+  __temp.val[3] = vcombine_s8 (__val.val[3], vcreate_s8 (__AARCH64_INT64_C (0)));
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
+  __builtin_aarch64_st1x4v8qi ((__builtin_aarch64_simd_qi *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1q_s8_x4 (int8_t * __a, int8x16x4_t val)
+vst1q_s8_x4 (int8_t * __a, int8x16x4_t __val)
 {
-  union { int8x16x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v16qi ((__builtin_aarch64_simd_qi *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  __builtin_memcpy (&__o, &__val, sizeof (__val));
+  __builtin_aarch64_st1x4v16qi ((__builtin_aarch64_simd_qi *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1_s16_x4 (int16_t * __a, int16x4x4_t val)
+vst1_s16_x4 (int16_t * __a, int16x4x4_t __val)
 {
-  union { int16x4x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v4hi ((__builtin_aarch64_simd_hi *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  int16x8x4_t __temp;
+  __temp.val[0] = vcombine_s16 (__val.val[0], vcreate_s16 (__AARCH64_INT64_C (0)));
+  __temp.val[1] = vcombine_s16 (__val.val[1], vcreate_s16 (__AARCH64_INT64_C (0)));
+  __temp.val[2] = vcombine_s16 (__val.val[2], vcreate_s16 (__AARCH64_INT64_C (0)));
+  __temp.val[3] = vcombine_s16 (__val.val[3], vcreate_s16 (__AARCH64_INT64_C (0)));
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
+  __builtin_aarch64_st1x4v4hi ((__builtin_aarch64_simd_hi *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1q_s16_x4 (int16_t * __a, int16x8x4_t val)
+vst1q_s16_x4 (int16_t * __a, int16x8x4_t __val)
 {
-  union { int16x8x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v8hi ((__builtin_aarch64_simd_hi *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  __builtin_memcpy (&__o, &__val, sizeof (__val));
+  __builtin_aarch64_st1x4v8hi ((__builtin_aarch64_simd_hi *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1_s32_x4 (int32_t * __a, int32x2x4_t val)
+vst1_s32_x4 (int32_t * __a, int32x2x4_t __val)
 {
-  union { int32x2x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v2si ((__builtin_aarch64_simd_si *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  int32x4x4_t __temp;
+  __temp.val[0] = vcombine_s32 (__val.val[0], vcreate_s32 (__AARCH64_INT64_C (0)));
+  __temp.val[1] = vcombine_s32 (__val.val[1], vcreate_s32 (__AARCH64_INT64_C (0)));
+  __temp.val[2] = vcombine_s32 (__val.val[2], vcreate_s32 (__AARCH64_INT64_C (0)));
+  __temp.val[3] = vcombine_s32 (__val.val[3], vcreate_s32 (__AARCH64_INT64_C (0)));
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
+  __builtin_aarch64_st1x4v2si ((__builtin_aarch64_simd_si *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1q_s32_x4 (int32_t * __a, int32x4x4_t val)
+vst1q_s32_x4 (int32_t * __a, int32x4x4_t __val)
 {
-  union { int32x4x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v4si ((__builtin_aarch64_simd_si *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  __builtin_memcpy (&__o, &__val, sizeof (__val));
+  __builtin_aarch64_st1x4v4si ((__builtin_aarch64_simd_si *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1_u8_x4 (uint8_t * __a, uint8x8x4_t val)
+vst1_u8_x4 (uint8_t * __a, uint8x8x4_t __val)
 {
-  union { uint8x8x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v8qi ((__builtin_aarch64_simd_qi *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  uint8x16x4_t __temp;
+  __temp.val[0] = vcombine_u8 (__val.val[0], vcreate_u8 (__AARCH64_UINT64_C (0)));
+  __temp.val[1] = vcombine_u8 (__val.val[1], vcreate_u8 (__AARCH64_UINT64_C (0)));
+  __temp.val[2] = vcombine_u8 (__val.val[2], vcreate_u8 (__AARCH64_UINT64_C (0)));
+  __temp.val[3] = vcombine_u8 (__val.val[3], vcreate_u8 (__AARCH64_UINT64_C (0)));
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
+  __builtin_aarch64_st1x4v8qi ((__builtin_aarch64_simd_qi *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1q_u8_x4 (uint8_t * __a, uint8x16x4_t val)
+vst1q_u8_x4 (uint8_t * __a, uint8x16x4_t __val)
 {
-  union { uint8x16x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v16qi ((__builtin_aarch64_simd_qi *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  __builtin_memcpy (&__o, &__val, sizeof (__val));
+  __builtin_aarch64_st1x4v16qi ((__builtin_aarch64_simd_qi *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1_u16_x4 (uint16_t * __a, uint16x4x4_t val)
+vst1_u16_x4 (uint16_t * __a, uint16x4x4_t __val)
 {
-  union { uint16x4x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v4hi ((__builtin_aarch64_simd_hi *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  uint16x8x4_t __temp;
+  __temp.val[0] = vcombine_u16 (__val.val[0], vcreate_u16 (__AARCH64_UINT64_C (0)));
+  __temp.val[1] = vcombine_u16 (__val.val[1], vcreate_u16 (__AARCH64_UINT64_C (0)));
+  __temp.val[2] = vcombine_u16 (__val.val[2], vcreate_u16 (__AARCH64_UINT64_C (0)));
+  __temp.val[3] = vcombine_u16 (__val.val[3], vcreate_u16 (__AARCH64_UINT64_C (0)));
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
+  __builtin_aarch64_st1x4v4hi ((__builtin_aarch64_simd_hi *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1q_u16_x4 (uint16_t * __a, uint16x8x4_t val)
+vst1q_u16_x4 (uint16_t * __a, uint16x8x4_t __val)
 {
-  union { uint16x8x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v8hi ((__builtin_aarch64_simd_hi *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  __builtin_memcpy (&__o, &__val, sizeof (__val));
+  __builtin_aarch64_st1x4v8hi ((__builtin_aarch64_simd_hi *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1_u32_x4 (uint32_t * __a, uint32x2x4_t val)
+vst1_u32_x4 (uint32_t * __a, uint32x2x4_t __val)
 {
-  union { uint32x2x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v2si ((__builtin_aarch64_simd_si *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  uint32x4x4_t __temp;
+  __temp.val[0] = vcombine_u32 (__val.val[0], vcreate_u32 (__AARCH64_UINT64_C (0)));
+  __temp.val[1] = vcombine_u32 (__val.val[1], vcreate_u32 (__AARCH64_UINT64_C (0)));
+  __temp.val[2] = vcombine_u32 (__val.val[2], vcreate_u32 (__AARCH64_UINT64_C (0)));
+  __temp.val[3] = vcombine_u32 (__val.val[3], vcreate_u32 (__AARCH64_UINT64_C (0)));
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
+  __builtin_aarch64_st1x4v2si ((__builtin_aarch64_simd_si *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1q_u32_x4 (uint32_t * __a, uint32x4x4_t val)
+vst1q_u32_x4 (uint32_t * __a, uint32x4x4_t __val)
 {
-  union { uint32x4x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v4si ((__builtin_aarch64_simd_si *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  __builtin_memcpy (&__o, &__val, sizeof (__val));
+  __builtin_aarch64_st1x4v4si ((__builtin_aarch64_simd_si *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1_f16_x4 (float16_t * __a, float16x4x4_t val)
+vst1_f16_x4 (float16_t * __a, float16x4x4_t __val)
 {
-  union { float16x4x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v4hf ((__builtin_aarch64_simd_hf *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  float16x8x4_t __temp;
+  __temp.val[0] = vcombine_f16 (__val.val[0], vcreate_f16 (__AARCH64_UINT64_C (0)));
+  __temp.val[1] = vcombine_f16 (__val.val[1], vcreate_f16 (__AARCH64_UINT64_C (0)));
+  __temp.val[2] = vcombine_f16 (__val.val[2], vcreate_f16 (__AARCH64_UINT64_C (0)));
+  __temp.val[3] = vcombine_f16 (__val.val[3], vcreate_f16 (__AARCH64_UINT64_C (0)));
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
+  __builtin_aarch64_st1x4v4hf ((__builtin_aarch64_simd_hf *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1q_f16_x4 (float16_t * __a, float16x8x4_t val)
+vst1q_f16_x4 (float16_t * __a, float16x8x4_t __val)
 {
-  union { float16x8x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v8hf ((__builtin_aarch64_simd_hf *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  __builtin_memcpy (&__o, &__val, sizeof (__val));
+  __builtin_aarch64_st1x4v8hf ((__builtin_aarch64_simd_hf *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1_f32_x4 (float32_t * __a, float32x2x4_t val)
+vst1_f32_x4 (float32_t * __a, float32x2x4_t __val)
 {
-  union { float32x2x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v2sf ((__builtin_aarch64_simd_sf *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  float32x4x4_t __temp;
+  __temp.val[0] = vcombine_f32 (__val.val[0], vcreate_f32 (__AARCH64_UINT64_C (0)));
+  __temp.val[1] = vcombine_f32 (__val.val[1], vcreate_f32 (__AARCH64_UINT64_C (0)));
+  __temp.val[2] = vcombine_f32 (__val.val[2], vcreate_f32 (__AARCH64_UINT64_C (0)));
+  __temp.val[3] = vcombine_f32 (__val.val[3], vcreate_f32 (__AARCH64_UINT64_C (0)));
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
+  __builtin_aarch64_st1x4v2sf ((__builtin_aarch64_simd_sf *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1q_f32_x4 (float32_t * __a, float32x4x4_t val)
+vst1q_f32_x4 (float32_t * __a, float32x4x4_t __val)
 {
-  union { float32x4x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v4sf ((__builtin_aarch64_simd_sf *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  __builtin_memcpy (&__o, &__val, sizeof (__val));
+  __builtin_aarch64_st1x4v4sf ((__builtin_aarch64_simd_sf *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1_p8_x4 (poly8_t * __a, poly8x8x4_t val)
+vst1_p8_x4 (poly8_t * __a, poly8x8x4_t __val)
 {
-  union { poly8x8x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v8qi ((__builtin_aarch64_simd_qi *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  poly8x16x4_t __temp;
+  __temp.val[0] = vcombine_p8 (__val.val[0], vcreate_p8 (__AARCH64_UINT64_C (0)));
+  __temp.val[1] = vcombine_p8 (__val.val[1], vcreate_p8 (__AARCH64_UINT64_C (0)));
+  __temp.val[2] = vcombine_p8 (__val.val[2], vcreate_p8 (__AARCH64_UINT64_C (0)));
+  __temp.val[3] = vcombine_p8 (__val.val[3], vcreate_p8 (__AARCH64_UINT64_C (0)));
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
+  __builtin_aarch64_st1x4v8qi ((__builtin_aarch64_simd_qi *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1q_p8_x4 (poly8_t * __a, poly8x16x4_t val)
+vst1q_p8_x4 (poly8_t * __a, poly8x16x4_t __val)
 {
-  union { poly8x16x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v16qi ((__builtin_aarch64_simd_qi *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  __builtin_memcpy (&__o, &__val, sizeof (__val));
+  __builtin_aarch64_st1x4v16qi ((__builtin_aarch64_simd_qi *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1_p16_x4 (poly16_t * __a, poly16x4x4_t val)
+vst1_p16_x4 (poly16_t * __a, poly16x4x4_t __val)
 {
-  union { poly16x4x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v4hi ((__builtin_aarch64_simd_hi *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  poly16x8x4_t __temp;
+  __temp.val[0] = vcombine_p16 (__val.val[0], vcreate_p16 (__AARCH64_UINT64_C (0)));
+  __temp.val[1] = vcombine_p16 (__val.val[1], vcreate_p16 (__AARCH64_UINT64_C (0)));
+  __temp.val[2] = vcombine_p16 (__val.val[2], vcreate_p16 (__AARCH64_UINT64_C (0)));
+  __temp.val[3] = vcombine_p16 (__val.val[3], vcreate_p16 (__AARCH64_UINT64_C (0)));
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
+  __builtin_aarch64_st1x4v4hi ((__builtin_aarch64_simd_hi *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1q_p16_x4 (poly16_t * __a, poly16x8x4_t val)
+vst1q_p16_x4 (poly16_t * __a, poly16x8x4_t __val)
 {
-  union { poly16x8x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v8hi ((__builtin_aarch64_simd_hi *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  __builtin_memcpy (&__o, &__val, sizeof (__val));
+  __builtin_aarch64_st1x4v8hi ((__builtin_aarch64_simd_hi *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1_s64_x4 (int64_t * __a, int64x1x4_t val)
+vst1_s64_x4 (int64_t * __a, int64x1x4_t __val)
 {
-  union { int64x1x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4di ((__builtin_aarch64_simd_di *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  int64x2x4_t __temp;
+  __temp.val[0] = vcombine_s64 (__val.val[0], vcreate_s64 (__AARCH64_INT64_C (0)));
+  __temp.val[1] = vcombine_s64 (__val.val[1], vcreate_s64 (__AARCH64_INT64_C (0)));
+  __temp.val[2] = vcombine_s64 (__val.val[2], vcreate_s64 (__AARCH64_INT64_C (0)));
+  __temp.val[3] = vcombine_s64 (__val.val[3], vcreate_s64 (__AARCH64_INT64_C (0)));
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
+  __builtin_aarch64_st1x4di ((__builtin_aarch64_simd_di *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1_u64_x4 (uint64_t * __a, uint64x1x4_t val)
+vst1_u64_x4 (uint64_t * __a, uint64x1x4_t __val)
 {
-  union { uint64x1x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4di ((__builtin_aarch64_simd_di *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  uint64x2x4_t __temp;
+  __temp.val[0] = vcombine_u64 (__val.val[0], vcreate_u64 (__AARCH64_UINT64_C (0)));
+  __temp.val[1] = vcombine_u64 (__val.val[1], vcreate_u64 (__AARCH64_UINT64_C (0)));
+  __temp.val[2] = vcombine_u64 (__val.val[2], vcreate_u64 (__AARCH64_UINT64_C (0)));
+  __temp.val[3] = vcombine_u64 (__val.val[3], vcreate_u64 (__AARCH64_UINT64_C (0)));
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
+  __builtin_aarch64_st1x4di ((__builtin_aarch64_simd_di *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1_p64_x4 (poly64_t * __a, poly64x1x4_t val)
+vst1_p64_x4 (poly64_t * __a, poly64x1x4_t __val)
 {
-  union { poly64x1x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4di ((__builtin_aarch64_simd_di *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  poly64x2x4_t __temp;
+  __temp.val[0] = vcombine_p64 (__val.val[0], vcreate_p64 (__AARCH64_UINT64_C (0)));
+  __temp.val[1] = vcombine_p64 (__val.val[1], vcreate_p64 (__AARCH64_UINT64_C (0)));
+  __temp.val[2] = vcombine_p64 (__val.val[2], vcreate_p64 (__AARCH64_UINT64_C (0)));
+  __temp.val[3] = vcombine_p64 (__val.val[3], vcreate_p64 (__AARCH64_UINT64_C (0)));
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
+  __builtin_aarch64_st1x4di ((__builtin_aarch64_simd_di *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1q_s64_x4 (int64_t * __a, int64x2x4_t val)
+vst1q_s64_x4 (int64_t * __a, int64x2x4_t __val)
 {
-  union { int64x2x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v2di ((__builtin_aarch64_simd_di *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  __builtin_memcpy (&__o, &__val, sizeof (__val));
+  __builtin_aarch64_st1x4v2di ((__builtin_aarch64_simd_di *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1q_u64_x4 (uint64_t * __a, uint64x2x4_t val)
+vst1q_u64_x4 (uint64_t * __a, uint64x2x4_t __val)
 {
-  union { uint64x2x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v2di ((__builtin_aarch64_simd_di *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  __builtin_memcpy (&__o, &__val, sizeof (__val));
+  __builtin_aarch64_st1x4v2di ((__builtin_aarch64_simd_di *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1q_p64_x4 (poly64_t * __a, poly64x2x4_t val)
+vst1q_p64_x4 (poly64_t * __a, poly64x2x4_t __val)
 {
-  union { poly64x2x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v2di ((__builtin_aarch64_simd_di *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  __builtin_memcpy (&__o, &__val, sizeof (__val));
+  __builtin_aarch64_st1x4v2di ((__builtin_aarch64_simd_di *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1_f64_x4 (float64_t * __a, float64x1x4_t val)
+vst1_f64_x4 (float64_t * __a, float64x1x4_t __val)
 {
-  union { float64x1x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4df ((__builtin_aarch64_simd_df *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  float64x2x4_t __temp;
+  __temp.val[0] = vcombine_f64 (__val.val[0], vcreate_f64 (__AARCH64_UINT64_C (0)));
+  __temp.val[1] = vcombine_f64 (__val.val[1], vcreate_f64 (__AARCH64_UINT64_C (0)));
+  __temp.val[2] = vcombine_f64 (__val.val[2], vcreate_f64 (__AARCH64_UINT64_C (0)));
+  __temp.val[3] = vcombine_f64 (__val.val[3], vcreate_f64 (__AARCH64_UINT64_C (0)));
+  __builtin_memcpy (&__o, &__temp, sizeof (__temp));
+  __builtin_aarch64_st1x4df ((__builtin_aarch64_simd_df *) __a, __o);
 }
 
 __extension__ extern __inline void
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-vst1q_f64_x4 (float64_t * __a, float64x2x4_t val)
+vst1q_f64_x4 (float64_t * __a, float64x2x4_t __val)
 {
-  union { float64x2x4_t __i; __builtin_aarch64_simd_xi __o; } __u = { val };
-  __builtin_aarch64_st1x4v2df ((__builtin_aarch64_simd_df *) __a, __u.__o);
+  __builtin_aarch64_simd_xi __o;
+  __builtin_memcpy (&__o, &__val, sizeof (__val));
+  __builtin_aarch64_st1x4v2df ((__builtin_aarch64_simd_df *) __a, __o);
 }
 
 /* vstn */
diff --git a/gcc/testsuite/gcc.target/aarch64/vector_structure_intrinsics.c b/gcc/testsuite/gcc.target/aarch64/vector_structure_intrinsics.c
index 5a6663a..6537f68 100644
--- a/gcc/testsuite/gcc.target/aarch64/vector_structure_intrinsics.c
+++ b/gcc/testsuite/gcc.target/aarch64/vector_structure_intrinsics.c
@@ -139,6 +139,27 @@ TEST_ST3 (vst3q, uint64x2x3_t, uint64_t*, u64);
 TEST_ST3 (vst3q, float64x2x3_t, float64_t*, f64);
 TEST_ST3 (vst3q, poly64x2x3_t, poly64_t*, p64);
 
+#define TEST_ST1xN(name, tbltype, ptrtype, ts, xn) \
+  void test_ ## name ## _ ## ts ## _ ## xn (ptrtype a, tbltype b) \
+	{ \
+		name ## _ ## ts ## _ ## xn (a, b); \
+	}
+
+TEST_ST1xN (vst1q, int8x16x4_t, int8_t*, s8, x4);
+TEST_ST1xN (vst1q, uint8x16x4_t, uint8_t*, u8, x4);
+TEST_ST1xN (vst1q, poly8x16x4_t, poly8_t*, p8, x4);
+TEST_ST1xN (vst1q, int16x8x4_t, int16_t*, s16, x4);
+TEST_ST1xN (vst1q, uint16x8x4_t, uint16_t*, u16, x4);
+TEST_ST1xN (vst1q, poly16x8x4_t, poly16_t*, p16, x4);
+TEST_ST1xN (vst1q, float16x8x4_t, float16_t*, f16, x4);
+TEST_ST1xN (vst1q, int32x4x4_t, int32_t*, s32, x4);
+TEST_ST1xN (vst1q, uint32x4x4_t, uint32_t*, u32, x4);
+TEST_ST1xN (vst1q, float32x4x4_t, float32_t*, f32, x4);
+TEST_ST1xN (vst1q, int64x2x4_t, int64_t*, s64, x4);
+TEST_ST1xN (vst1q, uint64x2x4_t, uint64_t*, u64, x4);
+TEST_ST1xN (vst1q, poly64x2x4_t, poly64_t*, p64, x4);
+TEST_ST1xN (vst1q, float64x2x4_t, float64_t*, f64, x4);
+
 /* { dg-final { scan-assembler-not "mov\\t" } } */
 
 /* { dg-final { scan-assembler-times "tbl\\t" 18} }  */
@@ -146,3 +167,4 @@ TEST_ST3 (vst3q, poly64x2x3_t, poly64_t*, p64);
 /* { dg-final { scan-assembler-times "st4\\t" 14} }  */
 /* { dg-final { scan-assembler-times "st3\\t" 14} }  */
 /* { dg-final { scan-assembler-times "st2\\t" 14} }  */
+/* { dg-final { scan-assembler-times "st1\\t" 14} }  */