Hi all,
This is the second part of James's patch from:
https://gcc.gnu.org/ml/gcc-patches/2013-09/msg01068.html
separated out. It reimplements the vcopyq_lane* intrinsics in C and
adds implementations of the other missing vcopy<q>_lane_<q> intrinsics.
The differences from that patch are in the use of __aarch64_vset_lane_any and
__aarch64_vget_lane_any rather than the typed variants of these that were used
back in 2013 (and don't exist anymore).
The testcase is also adjusted for the ABI change in GCC 5 where integer x1
vectors
are now passed and returned in SIMD registers.
The vcopy_laneq_f64 test in the testcase is currently XFAILed because it
currently
doesn't generate the optimal DUP instruction but instead emits a UMOV to a
scalar
register and then an FMOV. This is a GCC 7 regression tracked by PR 71307 and I
think
unrelated to this patch.
Bootstrapped and tested on aarch64-none-linux-gnu.
Also tested on aarch64_be-none-elf.
Ok for trunk?
Thanks,
Kyrill
2016-06-07 Kyrylo Tkachov <kyrylo.tkac...@arm.com>
James Greenhalgh <james.greenha...@arm.com>
* config/aarch64/arm_neon.h (vcopyq_lane_f32, vcopyq_lane_f64,
vcopyq_lane_p8, vcopyq_lane_p16, vcopyq_lane_s8, vcopyq_lane_s16,
vcopyq_lane_s32, vcopyq_lane_s64, vcopyq_lane_u8, vcopyq_lane_u16,
vcopyq_lane_u32, vcopyq_lane_u64): Reimplement in C.
(vcopy_lane_f32, vcopy_lane_f64, vcopy_lane_p8, vcopy_lane_p16,
vcopy_lane_s8, vcopy_lane_s16, vcopy_lane_s32, vcopy_lane_s64,
vcopy_lane_u8, vcopy_lane_u16, vcopy_lane_u32, vcopy_lane_u64,
vcopy_laneq_f32, vcopy_laneq_f64, vcopy_laneq_p8, vcopy_laneq_p16,
vcopy_laneq_s8, vcopy_laneq_s16, vcopy_laneq_s32, vcopy_laneq_s64,
vcopy_laneq_u8, vcopy_laneq_u16, vcopy_laneq_u32, vcopy_laneq_u64,
vcopyq_laneq_f32, vcopyq_laneq_f64, vcopyq_laneq_p8, vcopyq_laneq_p16,
vcopyq_laneq_s8, vcopyq_laneq_s16, vcopyq_laneq_s32, vcopyq_laneq_s64,
vcopyq_laneq_u8, vcopyq_laneq_u16, vcopyq_laneq_u32, vcopyq_laneq_u64):
New intrinsics.
2016-06-07 Kyrylo Tkachov <kyrylo.tkac...@arm.com>
James Greenhalgh <james.greenha...@arm.com>
* gcc.target/aarch64/vect_copy_lane_1.c: New test.
diff --git a/gcc/config/aarch64/arm_neon.h b/gcc/config/aarch64/arm_neon.h
index d20caf0919356eb7a87e7c7a9cd336d8408db35b..e33ecdbb72c8e8a483f82adc6f0e28edb3e5f4e6 100644
--- a/gcc/config/aarch64/arm_neon.h
+++ b/gcc/config/aarch64/arm_neon.h
@@ -5869,162 +5869,6 @@ vaddlvq_u32 (uint32x4_t a)
return result;
}
-#define vcopyq_lane_f32(a, b, c, d) \
- __extension__ \
- ({ \
- float32x4_t c_ = (c); \
- float32x4_t a_ = (a); \
- float32x4_t result; \
- __asm__ ("ins %0.s[%2], %3.s[%4]" \
- : "=w"(result) \
- : "0"(a_), "i"(b), "w"(c_), "i"(d) \
- : /* No clobbers */); \
- result; \
- })
-
-#define vcopyq_lane_f64(a, b, c, d) \
- __extension__ \
- ({ \
- float64x2_t c_ = (c); \
- float64x2_t a_ = (a); \
- float64x2_t result; \
- __asm__ ("ins %0.d[%2], %3.d[%4]" \
- : "=w"(result) \
- : "0"(a_), "i"(b), "w"(c_), "i"(d) \
- : /* No clobbers */); \
- result; \
- })
-
-#define vcopyq_lane_p8(a, b, c, d) \
- __extension__ \
- ({ \
- poly8x16_t c_ = (c); \
- poly8x16_t a_ = (a); \
- poly8x16_t result; \
- __asm__ ("ins %0.b[%2], %3.b[%4]" \
- : "=w"(result) \
- : "0"(a_), "i"(b), "w"(c_), "i"(d) \
- : /* No clobbers */); \
- result; \
- })
-
-#define vcopyq_lane_p16(a, b, c, d) \
- __extension__ \
- ({ \
- poly16x8_t c_ = (c); \
- poly16x8_t a_ = (a); \
- poly16x8_t result; \
- __asm__ ("ins %0.h[%2], %3.h[%4]" \
- : "=w"(result) \
- : "0"(a_), "i"(b), "w"(c_), "i"(d) \
- : /* No clobbers */); \
- result; \
- })
-
-#define vcopyq_lane_s8(a, b, c, d) \
- __extension__ \
- ({ \
- int8x16_t c_ = (c); \
- int8x16_t a_ = (a); \
- int8x16_t result; \
- __asm__ ("ins %0.b[%2], %3.b[%4]" \
- : "=w"(result) \
- : "0"(a_), "i"(b), "w"(c_), "i"(d) \
- : /* No clobbers */); \
- result; \
- })
-
-#define vcopyq_lane_s16(a, b, c, d) \
- __extension__ \
- ({ \
- int16x8_t c_ = (c); \
- int16x8_t a_ = (a); \
- int16x8_t result; \
- __asm__ ("ins %0.h[%2], %3.h[%4]" \
- : "=w"(result) \
- : "0"(a_), "i"(b), "w"(c_), "i"(d) \
- : /* No clobbers */); \
- result; \
- })
-
-#define vcopyq_lane_s32(a, b, c, d) \
- __extension__ \
- ({ \
- int32x4_t c_ = (c); \
- int32x4_t a_ = (a); \
- int32x4_t result; \
- __asm__ ("ins %0.s[%2], %3.s[%4]" \
- : "=w"(result) \
- : "0"(a_), "i"(b), "w"(c_), "i"(d) \
- : /* No clobbers */); \
- result; \
- })
-
-#define vcopyq_lane_s64(a, b, c, d) \
- __extension__ \
- ({ \
- int64x2_t c_ = (c); \
- int64x2_t a_ = (a); \
- int64x2_t result; \
- __asm__ ("ins %0.d[%2], %3.d[%4]" \
- : "=w"(result) \
- : "0"(a_), "i"(b), "w"(c_), "i"(d) \
- : /* No clobbers */); \
- result; \
- })
-
-#define vcopyq_lane_u8(a, b, c, d) \
- __extension__ \
- ({ \
- uint8x16_t c_ = (c); \
- uint8x16_t a_ = (a); \
- uint8x16_t result; \
- __asm__ ("ins %0.b[%2], %3.b[%4]" \
- : "=w"(result) \
- : "0"(a_), "i"(b), "w"(c_), "i"(d) \
- : /* No clobbers */); \
- result; \
- })
-
-#define vcopyq_lane_u16(a, b, c, d) \
- __extension__ \
- ({ \
- uint16x8_t c_ = (c); \
- uint16x8_t a_ = (a); \
- uint16x8_t result; \
- __asm__ ("ins %0.h[%2], %3.h[%4]" \
- : "=w"(result) \
- : "0"(a_), "i"(b), "w"(c_), "i"(d) \
- : /* No clobbers */); \
- result; \
- })
-
-#define vcopyq_lane_u32(a, b, c, d) \
- __extension__ \
- ({ \
- uint32x4_t c_ = (c); \
- uint32x4_t a_ = (a); \
- uint32x4_t result; \
- __asm__ ("ins %0.s[%2], %3.s[%4]" \
- : "=w"(result) \
- : "0"(a_), "i"(b), "w"(c_), "i"(d) \
- : /* No clobbers */); \
- result; \
- })
-
-#define vcopyq_lane_u64(a, b, c, d) \
- __extension__ \
- ({ \
- uint64x2_t c_ = (c); \
- uint64x2_t a_ = (a); \
- uint64x2_t result; \
- __asm__ ("ins %0.d[%2], %3.d[%4]" \
- : "=w"(result) \
- : "0"(a_), "i"(b), "w"(c_), "i"(d) \
- : /* No clobbers */); \
- result; \
- })
-
#define vcvt_n_f32_s32(a, b) \
__extension__ \
({ \
@@ -12777,6 +12621,398 @@ vcntq_u8 (uint8x16_t __a)
return (uint8x16_t) __builtin_aarch64_popcountv16qi ((int8x16_t) __a);
}
+/* vcopy_lane. */
+
+__extension__ static __inline float32x2_t __attribute__ ((__always_inline__))
+vcopy_lane_f32 (float32x2_t __a, const int __lane1,
+ float32x2_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline float64x1_t __attribute__ ((__always_inline__))
+vcopy_lane_f64 (float64x1_t __a, const int __lane1,
+ float64x1_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline poly8x8_t __attribute__ ((__always_inline__))
+vcopy_lane_p8 (poly8x8_t __a, const int __lane1,
+ poly8x8_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline poly16x4_t __attribute__ ((__always_inline__))
+vcopy_lane_p16 (poly16x4_t __a, const int __lane1,
+ poly16x4_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int8x8_t __attribute__ ((__always_inline__))
+vcopy_lane_s8 (int8x8_t __a, const int __lane1,
+ int8x8_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int16x4_t __attribute__ ((__always_inline__))
+vcopy_lane_s16 (int16x4_t __a, const int __lane1,
+ int16x4_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int32x2_t __attribute__ ((__always_inline__))
+vcopy_lane_s32 (int32x2_t __a, const int __lane1,
+ int32x2_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int64x1_t __attribute__ ((__always_inline__))
+vcopy_lane_s64 (int64x1_t __a, const int __lane1,
+ int64x1_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint8x8_t __attribute__ ((__always_inline__))
+vcopy_lane_u8 (uint8x8_t __a, const int __lane1,
+ uint8x8_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint16x4_t __attribute__ ((__always_inline__))
+vcopy_lane_u16 (uint16x4_t __a, const int __lane1,
+ uint16x4_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint32x2_t __attribute__ ((__always_inline__))
+vcopy_lane_u32 (uint32x2_t __a, const int __lane1,
+ uint32x2_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint64x1_t __attribute__ ((__always_inline__))
+vcopy_lane_u64 (uint64x1_t __a, const int __lane1,
+ uint64x1_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+/* vcopy_laneq. */
+
+__extension__ static __inline float32x2_t __attribute__ ((__always_inline__))
+vcopy_laneq_f32 (float32x2_t __a, const int __lane1,
+ float32x4_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline float64x1_t __attribute__ ((__always_inline__))
+vcopy_laneq_f64 (float64x1_t __a, const int __lane1,
+ float64x2_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline poly8x8_t __attribute__ ((__always_inline__))
+vcopy_laneq_p8 (poly8x8_t __a, const int __lane1,
+ poly8x16_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline poly16x4_t __attribute__ ((__always_inline__))
+vcopy_laneq_p16 (poly16x4_t __a, const int __lane1,
+ poly16x8_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int8x8_t __attribute__ ((__always_inline__))
+vcopy_laneq_s8 (int8x8_t __a, const int __lane1,
+ int8x16_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int16x4_t __attribute__ ((__always_inline__))
+vcopy_laneq_s16 (int16x4_t __a, const int __lane1,
+ int16x8_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int32x2_t __attribute__ ((__always_inline__))
+vcopy_laneq_s32 (int32x2_t __a, const int __lane1,
+ int32x4_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int64x1_t __attribute__ ((__always_inline__))
+vcopy_laneq_s64 (int64x1_t __a, const int __lane1,
+ int64x2_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint8x8_t __attribute__ ((__always_inline__))
+vcopy_laneq_u8 (uint8x8_t __a, const int __lane1,
+ uint8x16_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint16x4_t __attribute__ ((__always_inline__))
+vcopy_laneq_u16 (uint16x4_t __a, const int __lane1,
+ uint16x8_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint32x2_t __attribute__ ((__always_inline__))
+vcopy_laneq_u32 (uint32x2_t __a, const int __lane1,
+ uint32x4_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint64x1_t __attribute__ ((__always_inline__))
+vcopy_laneq_u64 (uint64x1_t __a, const int __lane1,
+ uint64x2_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+/* vcopyq_lane. */
+
+__extension__ static __inline float32x4_t __attribute__ ((__always_inline__))
+vcopyq_lane_f32 (float32x4_t __a, const int __lane1,
+ float32x2_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline float64x2_t __attribute__ ((__always_inline__))
+vcopyq_lane_f64 (float64x2_t __a, const int __lane1,
+ float64x1_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline poly8x16_t __attribute__ ((__always_inline__))
+vcopyq_lane_p8 (poly8x16_t __a, const int __lane1,
+ poly8x8_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline poly16x8_t __attribute__ ((__always_inline__))
+vcopyq_lane_p16 (poly16x8_t __a, const int __lane1,
+ poly16x4_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int8x16_t __attribute__ ((__always_inline__))
+vcopyq_lane_s8 (int8x16_t __a, const int __lane1,
+ int8x8_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int16x8_t __attribute__ ((__always_inline__))
+vcopyq_lane_s16 (int16x8_t __a, const int __lane1,
+ int16x4_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int32x4_t __attribute__ ((__always_inline__))
+vcopyq_lane_s32 (int32x4_t __a, const int __lane1,
+ int32x2_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int64x2_t __attribute__ ((__always_inline__))
+vcopyq_lane_s64 (int64x2_t __a, const int __lane1,
+ int64x1_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint8x16_t __attribute__ ((__always_inline__))
+vcopyq_lane_u8 (uint8x16_t __a, const int __lane1,
+ uint8x8_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint16x8_t __attribute__ ((__always_inline__))
+vcopyq_lane_u16 (uint16x8_t __a, const int __lane1,
+ uint16x4_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint32x4_t __attribute__ ((__always_inline__))
+vcopyq_lane_u32 (uint32x4_t __a, const int __lane1,
+ uint32x2_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint64x2_t __attribute__ ((__always_inline__))
+vcopyq_lane_u64 (uint64x2_t __a, const int __lane1,
+ uint64x1_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+/* vcopyq_laneq. */
+
+__extension__ static __inline float32x4_t __attribute__ ((__always_inline__))
+vcopyq_laneq_f32 (float32x4_t __a, const int __lane1,
+ float32x4_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline float64x2_t __attribute__ ((__always_inline__))
+vcopyq_laneq_f64 (float64x2_t __a, const int __lane1,
+ float64x2_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline poly8x16_t __attribute__ ((__always_inline__))
+vcopyq_laneq_p8 (poly8x16_t __a, const int __lane1,
+ poly8x16_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline poly16x8_t __attribute__ ((__always_inline__))
+vcopyq_laneq_p16 (poly16x8_t __a, const int __lane1,
+ poly16x8_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int8x16_t __attribute__ ((__always_inline__))
+vcopyq_laneq_s8 (int8x16_t __a, const int __lane1,
+ int8x16_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int16x8_t __attribute__ ((__always_inline__))
+vcopyq_laneq_s16 (int16x8_t __a, const int __lane1,
+ int16x8_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int32x4_t __attribute__ ((__always_inline__))
+vcopyq_laneq_s32 (int32x4_t __a, const int __lane1,
+ int32x4_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline int64x2_t __attribute__ ((__always_inline__))
+vcopyq_laneq_s64 (int64x2_t __a, const int __lane1,
+ int64x2_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint8x16_t __attribute__ ((__always_inline__))
+vcopyq_laneq_u8 (uint8x16_t __a, const int __lane1,
+ uint8x16_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint16x8_t __attribute__ ((__always_inline__))
+vcopyq_laneq_u16 (uint16x8_t __a, const int __lane1,
+ uint16x8_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint32x4_t __attribute__ ((__always_inline__))
+vcopyq_laneq_u32 (uint32x4_t __a, const int __lane1,
+ uint32x4_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
+__extension__ static __inline uint64x2_t __attribute__ ((__always_inline__))
+vcopyq_laneq_u64 (uint64x2_t __a, const int __lane1,
+ uint64x2_t __b, const int __lane2)
+{
+ return __aarch64_vset_lane_any (__aarch64_vget_lane_any (__b, __lane2),
+ __a, __lane1);
+}
+
/* vcvt (double -> float). */
__extension__ static __inline float16x4_t __attribute__ ((__always_inline__))
diff --git a/gcc/testsuite/gcc.target/aarch64/vect_copy_lane_1.c b/gcc/testsuite/gcc.target/aarch64/vect_copy_lane_1.c
new file mode 100644
index 0000000000000000000000000000000000000000..e144def83862fb9b895533bd48a80be4bc643bf8
--- /dev/null
+++ b/gcc/testsuite/gcc.target/aarch64/vect_copy_lane_1.c
@@ -0,0 +1,86 @@
+/* { dg-do compile } */
+/* { dg-options "-O3" } */
+
+#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 vcopy##Q1##_lane##Q2##_##SUFFIX (a, INDEX1, b, INDEX2); \
+}
+
+/* vcopy_lane. */
+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 (int64x1_t, int64x1_t, , , s64, 0, 0)
+BUILD_TEST (uint64x1_t, uint64x1_t, , , u64, 0, 0)
+BUILD_TEST (float64x1_t, float64x1_t, , , f64, 0, 0)
+/* { dg-final { scan-assembler-times "fmov\\td0, d1" 3 } } */
+
+/* vcopy_laneq. */
+
+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 (float64x1_t, float64x2_t, , q, f64, 0, 1)
+BUILD_TEST (int64x1_t, int64x2_t, , q, s64, 0, 1)
+BUILD_TEST (uint64x1_t, uint64x2_t, , q, u64, 0, 1)
+/* XFAIL due to PR 71307. */
+/* { dg-final { scan-assembler-times "dup\\td0, v1.d\\\[1\\\]" 3 { xfail *-*-* } } } */
+
+/* vcopyq_lane. */
+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 } } */
+
+/* vcopyq_laneq. */
+
+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 } } */
