arm: Add vldN_lane_bf16 + vldNq_lane_bf16 intrisics

gcc/ChangeLog

2020-10-29  Andrea Corallo  <andrea.corallo@arm.com>

	* config/arm/arm_neon.h (vld2_lane_bf16, vld2q_lane_bf16)
	(vld3_lane_bf16, vld3q_lane_bf16, vld4_lane_bf16)
	(vld4q_lane_bf16): Add intrinsics.
	* config/arm/arm_neon_builtins.def: Touch for:
	__builtin_neon_vld2_lanev4bf, __builtin_neon_vld2_lanev8bf,
	__builtin_neon_vld3_lanev4bf, __builtin_neon_vld3_lanev8bf,
	__builtin_neon_vld4_lanev4bf, __builtin_neon_vld4_lanev8bf.
	* config/arm/iterators.md (VQ_HS): Add V8BF to the iterator.

gcc/testsuite/ChangeLog

2020-10-29  Andrea Corallo  <andrea.corallo@arm.com>

	* gcc.target/aarch64/advsimd-intrinsics/vld2_lane_bf16_indices_1.c:
	Run it also for the arm backend.
	* gcc.target/aarch64/advsimd-intrinsics/vld2q_lane_bf16_indices_1.c:
	Likewise.
	* gcc.target/aarch64/advsimd-intrinsics/vld3_lane_bf16_indices_1.c:
	Likewise.
	* gcc.target/aarch64/advsimd-intrinsics/vld3q_lane_bf16_indices_1.c:
	Likewise.
	* gcc.target/aarch64/advsimd-intrinsics/vld4_lane_bf16_indices_1.c:
	Likewise.
	* gcc.target/aarch64/advsimd-intrinsics/vld4q_lane_bf16_indices_1.c:
	Likewise.
	* gcc.target/arm/simd/vldn_lane_bf16_1.c: New test.

10 files changed, 148 insertions, 13 deletions

diff --git a/gcc/config/arm/arm_neon.h b/gcc/config/arm/arm_neon.h
index 24aad3370f6..4fee128ce8d 100644
--- a/gcc/config/arm/arm_neon.h
+++ b/gcc/config/arm/arm_neon.h
@@ -19721,6 +19721,68 @@ vst1q_lane_bf16 (bfloat16_t * __a, bfloat16x8_t __b, const int __c)
   __builtin_neon_vst1_lanev8bf (__a, __b, __c);
 }
 
+__extension__ extern __inline bfloat16x4x2_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld2_lane_bf16 (const bfloat16_t * __a, bfloat16x4x2_t __b, const int __c)
+{
+  union { bfloat16x4x2_t __i; __builtin_neon_ti __o; } __bu = { __b };
+  union { bfloat16x4x2_t __i; __builtin_neon_ti __o; } __rv;
+  __rv.__o = __builtin_neon_vld2_lanev4bf ( __a, __bu.__o, __c);
+  return __rv.__i;
+}
+
+__extension__ extern __inline bfloat16x8x2_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld2q_lane_bf16 (const bfloat16_t * __a, bfloat16x8x2_t __b, const int __c)
+{
+  union { bfloat16x8x2_t __i; __builtin_neon_oi __o; } __bu = { __b };
+  union { bfloat16x8x2_t __i; __builtin_neon_oi __o; } __rv;
+  __rv.__o = __builtin_neon_vld2_lanev8bf (__a, __bu.__o, __c);
+  return __rv.__i;
+}
+
+__extension__ extern __inline bfloat16x4x3_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld3_lane_bf16 (const bfloat16_t * __a, bfloat16x4x3_t __b, const int __c)
+{
+  union { bfloat16x4x3_t __i; __builtin_neon_ei __o; } __bu = { __b };
+  union { bfloat16x4x3_t __i; __builtin_neon_ei __o; } __rv;
+  __rv.__o = __builtin_neon_vld3_lanev4bf (__a, __bu.__o, __c);
+  return __rv.__i;
+}
+
+__extension__ extern __inline bfloat16x8x3_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld3q_lane_bf16 (const bfloat16_t * __a, bfloat16x8x3_t __b, const int __c)
+{
+  union { bfloat16x8x3_t __i; __builtin_neon_ci __o; } __bu = { __b };
+  union { bfloat16x8x3_t __i; __builtin_neon_ci __o; } __rv;
+  __rv.__o = __builtin_neon_vld3_lanev8bf (__a, __bu.__o, __c);
+  return __rv.__i;
+}
+
+__extension__ extern __inline bfloat16x4x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld4_lane_bf16 (const bfloat16_t * __a, bfloat16x4x4_t __b, const int __c)
+{
+  union { bfloat16x4x4_t __i; __builtin_neon_oi __o; } __bu = { __b };
+  union { bfloat16x4x4_t __i; __builtin_neon_oi __o; } __rv;
+  __rv.__o = __builtin_neon_vld4_lanev4bf (__a,
+					   __bu.__o, __c);
+  return __rv.__i;
+}
+
+__extension__ extern __inline bfloat16x8x4_t
+__attribute__  ((__always_inline__, __gnu_inline__, __artificial__))
+vld4q_lane_bf16 (const bfloat16_t * __a, bfloat16x8x4_t __b, const int __c)
+{
+  union { bfloat16x8x4_t __i; __builtin_neon_xi __o; } __bu = { __b };
+  union { bfloat16x8x4_t __i; __builtin_neon_xi __o; } __rv;
+  __rv.__o = __builtin_neon_vld4_lanev8bf (__a,
+					   __bu.__o, __c);
+  return __rv.__i;
+}
+
 #pragma GCC pop_options
 
 #ifdef __cplusplus
diff --git a/gcc/config/arm/arm_neon_builtins.def b/gcc/config/arm/arm_neon_builtins.def
index e3ab6281497..1cb8c8c23b4 100644
--- a/gcc/config/arm/arm_neon_builtins.def
+++ b/gcc/config/arm/arm_neon_builtins.def
@@ -324,8 +324,8 @@ VAR14 (STORE1LANE, vst1_lane,
        v8qi, v4hi, v4hf, v2si, v2sf, di, v16qi, v8hi, v8hf, v4si, v4sf, v2di, v4bf, v8bf)
 VAR13 (LOAD1, vld2,
 	v8qi, v4hi, v4hf, v2si, v2sf, di, v16qi, v8hi, v8hf, v4si, v4sf, v4bf, v8bf)
-VAR9 (LOAD1LANE, vld2_lane,
-	v8qi, v4hi, v4hf, v2si, v2sf, v8hi, v8hf, v4si, v4sf)
+VAR11 (LOAD1LANE, vld2_lane,
+       v8qi, v4hi, v4hf, v2si, v2sf, v8hi, v8hf, v4si, v4sf, v4bf, v8bf)
 VAR8 (LOAD1, vld2_dup, v8qi, v4hi, v4hf, v2si, v2sf, di, v4bf, v8bf)
 VAR13 (STORE1, vst2,
 	v8qi, v4hi, v4hf, v4bf, v2si, v2sf, di, v16qi, v8hi, v8hf, v8bf, v4si, v4sf)
@@ -333,8 +333,8 @@ VAR9 (STORE1LANE, vst2_lane,
 	v8qi, v4hi, v4hf, v2si, v2sf, v8hi, v8hf, v4si, v4sf)
 VAR13 (LOAD1, vld3,
 	v8qi, v4hi, v4hf, v2si, v2sf, di, v16qi, v8hi, v8hf, v4si, v4sf, v4bf, v8bf)
-VAR9 (LOAD1LANE, vld3_lane,
-	v8qi, v4hi, v4hf, v2si, v2sf, v8hi, v8hf, v4si, v4sf)
+VAR11 (LOAD1LANE, vld3_lane,
+        v8qi, v4hi, v4hf, v2si, v2sf, v8hi, v8hf, v4si, v4sf, v4bf, v8bf)
 VAR8 (LOAD1, vld3_dup, v8qi, v4hi, v4hf, v2si, v2sf, di, v4bf, v8bf)
 VAR13 (STORE1, vst3,
 	v8qi, v4hi, v4hf, v4bf, v2si, v2sf, di, v16qi, v8hi, v8hf, v8bf, v4si, v4sf)
@@ -342,8 +342,8 @@ VAR9 (STORE1LANE, vst3_lane,
 	v8qi, v4hi, v4hf, v2si, v2sf, v8hi, v8hf, v4si, v4sf)
 VAR13 (LOAD1, vld4,
 	v8qi, v4hi, v4hf, v2si, v2sf, di, v16qi, v8hi, v8hf, v4si, v4sf, v4bf, v8bf)
-VAR9 (LOAD1LANE, vld4_lane,
-	v8qi, v4hi, v4hf, v2si, v2sf, v8hi, v8hf, v4si, v4sf)
+VAR11 (LOAD1LANE, vld4_lane,
+        v8qi, v4hi, v4hf, v2si, v2sf, v8hi, v8hf, v4si, v4sf, v4bf, v8bf)
 VAR8 (LOAD1, vld4_dup, v8qi, v4hi, v4hf, v2si, v2sf, di, v4bf, v8bf)
 VAR13 (STORE1, vst4,
 	v8qi, v4hi, v4hf, v4bf, v2si, v2sf, di, v16qi, v8hi, v8hf, v8bf, v4si, v4sf)
diff --git a/gcc/config/arm/iterators.md b/gcc/config/arm/iterators.md
index f93487293a7..592af35f038 100644
--- a/gcc/config/arm/iterators.md
+++ b/gcc/config/arm/iterators.md
@@ -116,7 +116,7 @@
 (define_mode_iterator VQ2BF [V16QI V8HI V8HF (V8BF "TARGET_BF16_SIMD") V4SI V4SF])
 
 ;; Quad-width vector modes with 16- or 32-bit elements
-(define_mode_iterator VQ_HS [V8HI V8HF V4SI V4SF])
+(define_mode_iterator VQ_HS [V8HI V8HF V4SI V4SF (V8BF "TARGET_BF16_SIMD")])
 
 ;; Quad-width vector modes plus 64-bit elements.
 (define_mode_iterator VQX [V16QI V8HI V8HF V8BF V4SI V4SF V2DI])
diff --git a/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld2_lane_bf16_indices_1.c b/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld2_lane_bf16_indices_1.c
index 99c979393ff..d568a26ba8a 100644
--- a/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld2_lane_bf16_indices_1.c
+++ b/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld2_lane_bf16_indices_1.c
@@ -1,4 +1,4 @@
-/* { dg-do compile { target { aarch64*-*-* } } } */
+/* { dg-do compile } */
 /* { dg-skip-if "" { *-*-* } { "-fno-fat-lto-objects" } } */
 /* { dg-require-effective-target arm_v8_2a_bf16_neon_ok } */
 /* { dg-add-options arm_v8_2a_bf16_neon }  */
diff --git a/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld2q_lane_bf16_indices_1.c b/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld2q_lane_bf16_indices_1.c
index 86d778a07b1..b91f14a7d01 100644
--- a/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld2q_lane_bf16_indices_1.c
+++ b/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld2q_lane_bf16_indices_1.c
@@ -1,4 +1,4 @@
-/* { dg-do compile { target { aarch64*-*-* } } } */
+/* { dg-do compile } */
 /* { dg-skip-if "" { *-*-* } { "-fno-fat-lto-objects" } } */
 /* { dg-require-effective-target arm_v8_2a_bf16_neon_ok } */
 /* { dg-add-options arm_v8_2a_bf16_neon }  */
diff --git a/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld3_lane_bf16_indices_1.c b/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld3_lane_bf16_indices_1.c
index e91a2bea1ad..331abf42b3c 100644
--- a/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld3_lane_bf16_indices_1.c
+++ b/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld3_lane_bf16_indices_1.c
@@ -1,4 +1,4 @@
-/* { dg-do compile { target { aarch64*-*-* } } } */
+/* { dg-do compile } */
 /* { dg-skip-if "" { *-*-* } { "-fno-fat-lto-objects" } } */
 /* { dg-require-effective-target arm_v8_2a_bf16_neon_ok } */
 /* { dg-add-options arm_v8_2a_bf16_neon }  */
diff --git a/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld3q_lane_bf16_indices_1.c b/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld3q_lane_bf16_indices_1.c
index 95421befd9f..1c52887aa67 100644
--- a/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld3q_lane_bf16_indices_1.c
+++ b/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld3q_lane_bf16_indices_1.c
@@ -1,4 +1,4 @@
-/* { dg-do compile { target { aarch64*-*-* } } } */
+/* { dg-do compile } */
 /* { dg-skip-if "" { *-*-* } { "-fno-fat-lto-objects" } } */
 /* { dg-require-effective-target arm_v8_2a_bf16_neon_ok } */
 /* { dg-add-options arm_v8_2a_bf16_neon }  */
diff --git a/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld4_lane_bf16_indices_1.c b/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld4_lane_bf16_indices_1.c
index 1c819aa8aad..3f486f94c25 100644
--- a/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld4_lane_bf16_indices_1.c
+++ b/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld4_lane_bf16_indices_1.c
@@ -1,4 +1,4 @@
-/* { dg-do compile { target { aarch64*-*-* } } } */
+/* { dg-do compile } */
 /* { dg-skip-if "" { *-*-* } { "-fno-fat-lto-objects" } } */
 /* { dg-require-effective-target arm_v8_2a_bf16_neon_ok } */
 /* { dg-add-options arm_v8_2a_bf16_neon }  */
diff --git a/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld4q_lane_bf16_indices_1.c b/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld4q_lane_bf16_indices_1.c
index f7c76fa0cde..7159cd86651 100644
--- a/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld4q_lane_bf16_indices_1.c
+++ b/gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vld4q_lane_bf16_indices_1.c
@@ -1,4 +1,4 @@
-/* { dg-do compile { target { aarch64*-*-* } } } */
+/* { dg-do compile } */
 /* { dg-skip-if "" { *-*-* } { "-fno-fat-lto-objects" } } */
 /* { dg-require-effective-target arm_v8_2a_bf16_neon_ok } */
 /* { dg-add-options arm_v8_2a_bf16_neon }  */
diff --git a/gcc/testsuite/gcc.target/arm/simd/vldn_lane_bf16_1.c b/gcc/testsuite/gcc.target/arm/simd/vldn_lane_bf16_1.c
new file mode 100644
index 00000000000..58153eddf62
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/simd/vldn_lane_bf16_1.c
@@ -0,0 +1,73 @@
+/* { dg-do assemble } */
+/* { dg-require-effective-target arm_v8_2a_bf16_neon_ok } */
+/* { dg-add-options arm_v8_2a_bf16_neon } */
+/* { dg-additional-options "-save-temps -O2 -mfloat-abi=hard" }  */
+/* { dg-final { check-function-bodies "**" "" } } */
+
+#include "arm_neon.h"
+
+/*
+**test_vld2_lane_bf16:
+**	vld2.16	{d0\[2\], d1\[2\]}, \[r0\]
+**	bx	lr
+*/
+bfloat16x4x2_t
+test_vld2_lane_bf16 (const bfloat16_t *a, bfloat16x4x2_t b)
+{
+  return vld2_lane_bf16 (a, b, 2);
+}
+
+/*
+**test_vld2q_lane_bf16:
+**	vld2.16	{d0\[2\], d2\[2\]}, \[r0\]
+**	bx	lr
+*/
+bfloat16x8x2_t
+test_vld2q_lane_bf16 (const bfloat16_t *a, bfloat16x8x2_t b)
+{
+  return vld2q_lane_bf16 (a, b, 2);
+}
+
+/*
+**test_vld3_lane_bf16:
+**	vld3.16	{d0\[2\], d1\[2\], d2\[2\]}, \[r0\]
+**	bx	lr
+*/
+bfloat16x4x3_t
+test_vld3_lane_bf16 (const bfloat16_t *a, bfloat16x4x3_t b)
+{
+  return vld3_lane_bf16 (a, b, 2);
+}
+
+/*
+**test_vld3q_lane_bf16:
+**	vld3.16	{d0\[2\], d2\[2\], d4\[2\]}, \[r0\]
+**	bx	lr
+*/
+bfloat16x8x3_t
+test_vld3q_lane_bf16 (const bfloat16_t *a, bfloat16x8x3_t b)
+{
+  return vld3q_lane_bf16 (a, b, 2);
+}
+
+/*
+**test_vld4_lane_bf16:
+**	vld4.16	{d0\[2\], d1\[2\], d2\[2\], d3\[2\]}, \[r0\]
+**	bx	lr
+*/
+bfloat16x4x4_t
+test_vld4_lane_bf16 (const bfloat16_t *a, bfloat16x4x4_t b)
+{
+  return vld4_lane_bf16 (a, b, 2);
+}
+
+/*
+**test_vld4q_lane_bf16:
+**	vld4.16	{d0\[2\], d2\[2\], d4\[2\], d6\[2\]}, \[r0\]
+**	bx	lr
+*/
+bfloat16x8x4_t
+test_vld4q_lane_bf16 (const bfloat16_t *a, bfloat16x8x4_t b)
+{
+  return vld4q_lane_bf16 (a, b, 2);
+}