arm: vld1q_types_x4 ACLE intrinsics

This patch is part of a series of patches implementing the _xN
variants of the vld1q intrinsic for the arm port. This patch adds the
_x4 variants of the vld1q intrinsic.

ACLE documents:
https://developer.arm.com/documentation/ihi0053/latest/

ISA documents:
https://developer.arm.com/documentation/ddi0487/latest/

gcc/ChangeLog:
	* config/arm/arm_neon.h
	(vld1q_u8_x4, vld1q_u16_x4, vld1q_u32_x4, vld1q_u64_x4): New.
	(vld1q_s8_x4, vld1q_s16_x4, vld1q_s32_x4, vld1q_s64_x4): New.
	(vld1q_f16_x4, vld1q_f32_x4): New.
	(vld1q_p8_x4, vld1q_p16_x4, vld1q_p64_x4): New.
	(vld1q_bf16_x4): New.
	* config/arm/arm_neon_builtins.def (vld1_x4): New entries.
	* config/arm/neon.md
	(neon_vld1_x4<mode>): New.
	(neon_vld1x4qa<mode>, neon_vld1x4qb<mode>): New
	* config/arm/unspecs.md
	(UNSPEC_VLD1X4A, UNSPEC_VLD1X4B): New.

gcc/testsuite/ChangeLog:
	* gcc.target/arm/simd/vld1q_base_xN_1.c: Updated.
	* gcc.target/arm/simd/vld1q_bf16_xN_1.c: Updated.
	* gcc.target/arm/simd/vld1q_fp16_xN_1.c: Updated.
	* gcc.target/arm/simd/vld1q_p64_xN_1.c: Updated.

8 files changed, 263 insertions, 14 deletions

diff --git a/gcc/config/arm/arm_neon.h b/gcc/config/arm/arm_neon.h
index 8f53248..bf9d2c1 100644
--- a/gcc/config/arm/arm_neon.h
+++ b/gcc/config/arm/arm_neon.h
@@ -10421,6 +10421,15 @@ vld1q_p64_x3 (const poly64_t * __a)
   return __rv.__i;
 }
 
+__extension__ extern __inline poly64x2x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld1q_p64_x4 (const poly64_t * __a)
+{
+  union { poly64x2x4_t __i; __builtin_neon_xi __o; } __rv;
+  __rv.__o = __builtin_neon_vld1_x4v2di ((const __builtin_neon_di *) __a);
+  return __rv.__i;
+}
+
 #pragma GCC pop_options
 __extension__ extern __inline int8x16_t
 __attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
@@ -10522,6 +10531,42 @@ vld1q_s64_x3 (const int64_t * __a)
   return __rv.__i;
 }
 
+__extension__ extern __inline int8x16x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld1q_s8_x4 (const uint8_t * __a)
+{
+  union { int8x16x4_t __i; __builtin_neon_xi __o; } __rv;
+  __rv.__o = __builtin_neon_vld1_x4v16qi ((const __builtin_neon_qi *) __a);
+  return __rv.__i;
+}
+
+__extension__ extern __inline int16x8x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld1q_s16_x4 (const uint16_t * __a)
+{
+  union { int16x8x4_t __i; __builtin_neon_xi __o; } __rv;
+  __rv.__o = __builtin_neon_vld1_x4v8hi ((const __builtin_neon_hi *) __a);
+  return __rv.__i;
+}
+
+__extension__ extern __inline int32x4x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld1q_s32_x4 (const int32_t * __a)
+{
+  union { int32x4x4_t __i; __builtin_neon_xi __o; } __rv;
+  __rv.__o = __builtin_neon_vld1_x4v4si ((const __builtin_neon_si *) __a);
+  return __rv.__i;
+}
+
+__extension__ extern __inline int64x2x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld1q_s64_x4 (const int64_t * __a)
+{
+  union { int64x2x4_t __i; __builtin_neon_xi __o; } __rv;
+  __rv.__o = __builtin_neon_vld1_x4v2di ((const __builtin_neon_di *) __a);
+  return __rv.__i;
+}
+
 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
 __extension__ extern __inline float16x8_t
 __attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
@@ -10578,6 +10623,26 @@ vld1q_f32_x3 (const float32_t * __a)
   return __rv.__i;
 }
 
+#if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
+__extension__ extern __inline float16x8x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld1q_f16_x4 (const float16_t * __a)
+{
+  union { float16x8x4_t __i; __builtin_neon_xi __o; } __rv;
+  __rv.__o = __builtin_neon_vld1_x4v8hf (__a);
+  return __rv.__i;
+}
+#endif
+
+__extension__ extern __inline float32x4x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld1q_f32_x4 (const float32_t * __a)
+{
+  union { float32x4x4_t __i; __builtin_neon_xi __o; } __rv;
+  __rv.__o = __builtin_neon_vld1_x4v4sf ((const __builtin_neon_sf *) __a);
+  return __rv.__i;
+}
+
 __extension__ extern __inline uint8x16_t
 __attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
 vld1q_u8 (const uint8_t * __a)
@@ -10678,6 +10743,42 @@ vld1q_u64_x3 (const uint64_t * __a)
   return __rv.__i;
 }
 
+__extension__ extern __inline uint8x16x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld1q_u8_x4 (const uint8_t * __a)
+{
+  union { uint8x16x4_t __i; __builtin_neon_xi __o; } __rv;
+  __rv.__o = __builtin_neon_vld1_x4v16qi ((const __builtin_neon_qi *) __a);
+  return __rv.__i;
+}
+
+__extension__ extern __inline uint16x8x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld1q_u16_x4 (const uint16_t * __a)
+{
+  union { uint16x8x4_t __i; __builtin_neon_xi __o; } __rv;
+  __rv.__o = __builtin_neon_vld1_x4v8hi ((const __builtin_neon_hi *) __a);
+  return __rv.__i;
+}
+
+__extension__ extern __inline uint32x4x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld1q_u32_x4 (const uint32_t * __a)
+{
+  union { uint32x4x4_t __i; __builtin_neon_xi __o; } __rv;
+  __rv.__o = __builtin_neon_vld1_x4v4si ((const __builtin_neon_si *) __a);
+  return __rv.__i;
+}
+
+__extension__ extern __inline uint64x2x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld1q_u64_x4 (const uint64_t * __a)
+{
+  union { uint64x2x4_t __i; __builtin_neon_xi __o; } __rv;
+  __rv.__o = __builtin_neon_vld1_x4v2di ((const __builtin_neon_di *) __a);
+  return __rv.__i;
+}
+
 __extension__ extern __inline poly8x16_t
 __attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
 vld1q_p8 (const poly8_t * __a)
@@ -10728,6 +10829,24 @@ vld1q_p16_x3 (const poly16_t * __a)
   return __rv.__i;
 }
 
+__extension__ extern __inline poly8x16x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld1q_p8_x4 (const poly8_t * __a)
+{
+  union { poly8x16x4_t __i; __builtin_neon_xi __o; } __rv;
+  __rv.__o = __builtin_neon_vld1_x4v16qi ((const __builtin_neon_qi *) __a);
+  return __rv.__i;
+}
+
+__extension__ extern __inline poly16x8x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld1q_p16_x4 (const poly16_t * __a)
+{
+  union { poly16x8x4_t __i; __builtin_neon_xi __o; } __rv;
+  __rv.__o = __builtin_neon_vld1_x4v8hi ((const __builtin_neon_hi *) __a);
+  return __rv.__i;
+}
+
 __extension__ extern __inline int8x8_t
 __attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
 vld1_lane_s8 (const int8_t * __a, int8x8_t __b, const int __c)
@@ -20038,6 +20157,15 @@ vld1q_bf16_x3 (const bfloat16_t * __ptr)
   return __rv.__i;
 }
 
+__extension__ extern __inline bfloat16x8x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld1q_bf16_x4 (const bfloat16_t * __ptr)
+{
+  union { bfloat16x8x4_t __i; __builtin_neon_xi __o; } __rv;
+  __rv.__o = __builtin_neon_vld1_x4v8bf ((const __builtin_neon_bf *) __ptr);
+  return __rv.__i;
+}
+
 __extension__ extern __inline bfloat16x4x2_t
 __attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
 vld2_bf16 (bfloat16_t const * __ptr)
diff --git a/gcc/config/arm/arm_neon_builtins.def b/gcc/config/arm/arm_neon_builtins.def
index 1782606..83dff8f 100644
--- a/gcc/config/arm/arm_neon_builtins.def
+++ b/gcc/config/arm/arm_neon_builtins.def
@@ -300,6 +300,7 @@ VAR1 (TERNOP, vtbx3, v8qi)
 VAR1 (TERNOP, vtbx4, v8qi)
 VAR7 (LOAD1, vld1_x2, v16qi, v8hi, v4si, v2di, v8hf, v4sf, v8bf)
 VAR7 (LOAD1, vld1_x3, v16qi, v8hi, v4si, v2di, v8hf, v4sf, v8bf)
+VAR7 (LOAD1, vld1_x4, v16qi, v8hi, v4si, v2di, v8hf, v4sf, v8bf)
 VAR13 (LOAD1, vld1,
         v8qi, v4hi, v4hf, v2si, v2sf, v16qi, v8hi, v8hf, v4si, v4sf, v2di,
         v4bf, v8bf)
diff --git a/gcc/config/arm/neon.md b/gcc/config/arm/neon.md
index 6e73737..d2923f7 100644
--- a/gcc/config/arm/neon.md
+++ b/gcc/config/arm/neon.md
@@ -5064,6 +5064,54 @@ if (BYTES_BIG_ENDIAN)
   [(set_attr "type" "neon_load1_3reg<q>")]
 )
 
+(define_expand "neon_vld1_x4<mode>"
+  [(match_operand:XI 0 "s_register_operand")
+   (match_operand:XI 1 "neon_struct_operand")
+   (unspec:VQXBF [(const_int 0)] UNSPEC_VSTRUCTDUMMY)]
+  "TARGET_NEON"
+{
+  rtx mem = adjust_address (operands[1], OImode, 0);
+  emit_insn (gen_neon_vld1x4qa<mode> (operands[0], mem));
+  mem = adjust_address (mem, OImode, GET_MODE_SIZE (OImode));
+  emit_insn (gen_neon_vld1x4qb<mode> (operands[0], mem, operands[0]));
+  DONE;
+})
+
+(define_insn "neon_vld1x4qa<mode>"
+  [(set (match_operand:XI 0 "s_register_operand" "=w")
+        (unspec:XI [(match_operand:OI 1 "neon_struct_operand" "Um")
+                    (unspec:VQXBF [(const_int 0)] UNSPEC_VSTRUCTDUMMY)]
+                   UNSPEC_VLD1X4A))]
+  "TARGET_NEON"
+{
+  rtx ops[2];
+  ops[0] = gen_rtx_REG (OImode, REGNO (operands[0]));
+  ops[1] = operands[1];
+
+  output_asm_insn ("vld1.<V_sz_elem>\t%h0, %A1", ops);
+  return "";
+}
+  [(set_attr "type" "neon_load1_4reg<q>")]
+)
+
+(define_insn "neon_vld1x4qb<mode>"
+  [(set (match_operand:XI 0 "s_register_operand" "=w")
+        (unspec:XI [(match_operand:OI 1 "neon_struct_operand" "Um")
+                    (match_operand:XI 2 "s_register_operand" "0")
+                    (unspec:VQXBF [(const_int 0)] UNSPEC_VSTRUCTDUMMY)]
+                   UNSPEC_VLD1X4B))]
+  "TARGET_NEON"
+{
+  rtx ops[2];
+  ops[0] = gen_rtx_REG (OImode, REGNO (operands[0]) + 8);
+  ops[1] = operands[1];
+
+  output_asm_insn ("vld1.<V_sz_elem>\t%h0, %A1", ops);
+  return "";
+}
+  [(set_attr "type" "neon_load1_4reg<q>")]
+)
+
 ;; The lane numbers in the RTL are in GCC lane order, having been flipped
 ;; in arm_expand_neon_args. The lane numbers are restored to architectural
 ;; lane order here.
diff --git a/gcc/config/arm/unspecs.md b/gcc/config/arm/unspecs.md
index 2f88c07..25609dc 100644
--- a/gcc/config/arm/unspecs.md
+++ b/gcc/config/arm/unspecs.md
@@ -344,6 +344,8 @@
   UNSPEC_VLD1
   UNSPEC_VLD1X3A
   UNSPEC_VLD1X3B
+  UNSPEC_VLD1X4A
+  UNSPEC_VLD1X4B
   UNSPEC_VLD1_LANE
   UNSPEC_VLD2
   UNSPEC_VLD2_DUP
diff --git a/gcc/testsuite/gcc.target/arm/simd/vld1q_base_xN_1.c b/gcc/testsuite/gcc.target/arm/simd/vld1q_base_xN_1.c
index 117bc58..01b29b6 100644
--- a/gcc/testsuite/gcc.target/arm/simd/vld1q_base_xN_1.c
+++ b/gcc/testsuite/gcc.target/arm/simd/vld1q_base_xN_1.c
@@ -115,14 +115,69 @@ poly16x8x3_t test_vld1q_p16_x3 (poly16_t * a)
     return vld1q_p16_x3 (a);
 }
 
-/* { dg-final { scan-assembler-times {vld1.8\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]\n} 6 } }  */
-/* { dg-final { scan-assembler-times {vld1.8\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]!\n} 3 } }  */
+uint8x16x4_t test_vld1q_u8_x4 (uint8_t * a)
+{
+    return vld1q_u8_x4 (a);
+}
+
+uint16x8x4_t test_vld1q_u16_x4 (uint16_t * a)
+{
+    return vld1q_u16_x4 (a);
+}
+
+uint32x4x4_t test_vld1q_u32_x4 (uint32_t * a)
+{
+    return vld1q_u32_x4 (a);
+}
+
+uint64x2x4_t test_vld1q_u64_x4 (uint64_t * a)
+{
+    return vld1q_u64_x4 (a);
+}
+
+int8x16x4_t test_vld1q_s8_x4 (int8_t * a)
+{
+    return vld1q_s8_x4 (a);
+}
+
+int16x8x4_t test_vld1q_s16_x4 (int16_t * a)
+{
+    return vld1q_s16_x4 (a);
+}
+
+int32x4x4_t test_vld1q_s32_x4 (int32_t * a)
+{
+    return vld1q_s32_x4 (a);
+}
+
+int64x2x4_t test_vld1q_s64_x4 (int64_t * a)
+{
+    return vld1q_s64_x4 (a);
+}
+
+float32x4x4_t test_vld1q_f32_x4 (float32_t * a)
+{
+    return vld1q_f32_x4 (a);
+}
+
+poly8x16x4_t test_vld1q_p8_x4 (poly8_t * a)
+{
+    return vld1q_p8_x4 (a);
+}
+
+poly16x8x4_t test_vld1q_p16_x4 (poly16_t * a)
+{
+    return vld1q_p16_x4 (a);
+}
+
+/* { dg-final { scan-assembler-times {vld1.8\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]\n} 9 } }  */
+/* { dg-final { scan-assembler-times {vld1.8\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]!\n} 6 } }  */
 
-/* { dg-final { scan-assembler-times {vld1.16\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]\n} 6 } }  */
-/* { dg-final { scan-assembler-times {vld1.16\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]!\n} 3 } }  */
+/* { dg-final { scan-assembler-times {vld1.16\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]\n} 9 } }  */
+/* { dg-final { scan-assembler-times {vld1.16\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]!\n} 6 } }  */
 
-/* { dg-final { scan-assembler-times {vld1.32\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]\n} 6 } }  */
-/* { dg-final { scan-assembler-times {vld1.32\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]!\n} 3 } }  */
+/* { dg-final { scan-assembler-times {vld1.32\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]\n} 9 } }  */
+/* { dg-final { scan-assembler-times {vld1.32\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]!\n} 6 } }  */
 
-/* { dg-final { scan-assembler-times {vld1.64\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+:64\]\n} 4 } }  */
-/* { dg-final { scan-assembler-times {vld1.64\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+:64\]!\n} 2 } }  */
+/* { dg-final { scan-assembler-times {vld1.64\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+:64\]\n} 6 } }  */
+/* { dg-final { scan-assembler-times {vld1.64\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+:64\]!\n} 4 } }  */
diff --git a/gcc/testsuite/gcc.target/arm/simd/vld1q_bf16_xN_1.c b/gcc/testsuite/gcc.target/arm/simd/vld1q_bf16_xN_1.c
index 75b61f1..21db352 100644
--- a/gcc/testsuite/gcc.target/arm/simd/vld1q_bf16_xN_1.c
+++ b/gcc/testsuite/gcc.target/arm/simd/vld1q_bf16_xN_1.c
@@ -15,5 +15,10 @@ bfloat16x8x3_t test_vld1q_bf16_x3 (bfloat16_t * a)
     return vld1q_bf16_x3 (a);
 }
 
-/* { dg-final { scan-assembler-times {vld1.16\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]\n} 2 } }  */
-/* { dg-final { scan-assembler-times {vld1.16\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]!\n} 1 } }  */
+bfloat16x8x4_t test_vld1q_bf16_x4 (bfloat16_t * a)
+{
+    return vld1q_bf16_x4 (a);
+}
+
+/* { dg-final { scan-assembler-times {vld1.16\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]\n} 3 } }  */
+/* { dg-final { scan-assembler-times {vld1.16\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]!\n} 2 } }  */
diff --git a/gcc/testsuite/gcc.target/arm/simd/vld1q_fp16_xN_1.c b/gcc/testsuite/gcc.target/arm/simd/vld1q_fp16_xN_1.c
index 9032048..3838cd0 100644
--- a/gcc/testsuite/gcc.target/arm/simd/vld1q_fp16_xN_1.c
+++ b/gcc/testsuite/gcc.target/arm/simd/vld1q_fp16_xN_1.c
@@ -15,5 +15,10 @@ float16x8x3_t test_vld1q_f16_x3 (float16_t * a)
     return vld1q_f16_x3 (a);
 }
 
-/* { dg-final { scan-assembler-times {vld1.16\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]\n} 2 } }  */
-/* { dg-final { scan-assembler-times {vld1.16\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]!\n} 1 } }  */
+float16x8x4_t test_vld1q_f16_x4 (float16_t * a)
+{
+    return vld1q_f16_x4 (a);
+}
+
+/* { dg-final { scan-assembler-times {vld1.16\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]\n} 3 } }  */
+/* { dg-final { scan-assembler-times {vld1.16\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+\]!\n} 2 } }  */
diff --git a/gcc/testsuite/gcc.target/arm/simd/vld1q_p64_xN_1.c b/gcc/testsuite/gcc.target/arm/simd/vld1q_p64_xN_1.c
index aabc31b..d359a5a 100644
--- a/gcc/testsuite/gcc.target/arm/simd/vld1q_p64_xN_1.c
+++ b/gcc/testsuite/gcc.target/arm/simd/vld1q_p64_xN_1.c
@@ -15,5 +15,10 @@ poly64x2x3_t test_vld1q_p64_x3 (poly64_t * a)
     return vld1q_p64_x3 (a);
 }
 
-/* { dg-final { scan-assembler-times {vld1.64\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+:64\]\n} 2 } }  */
-/* { dg-final { scan-assembler-times {vld1.64\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+:64\]!\n} 1 } }  */
+poly64x2x4_t test_vld1q_p64_x4 (poly64_t * a)
+{
+    return vld1q_p64_x4 (a);
+}
+
+/* { dg-final { scan-assembler-times {vld1.64\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+:64\]\n} 3 } }  */
+/* { dg-final { scan-assembler-times {vld1.64\t\{d[0-9]+-d[0-9]+\}, \[r[0-9]+:64\]!\n} 2 } }  */