[PATCH] AArch64: Implement missing vceq*_p* intrinsics

[PATCH] AArch64: Implement missing vceq*_p* intrinsics

[Kyrylo Tkachov]

[-- Attachment #1: Type: text/plain, Size: 590 bytes --]

Hi all,

This patch implements some missing vceq* intrinsics on poly types.
The behaviour is to produce the appropriate CMEQ instruction as for the unsigned types.

Bootstrapped and tested on aarch64-none-linux-gnu.

Committing to trunk and backporting to the branches after testing.

Thanks,
Kyrill

2020-09-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>

	PR target/71233
	* config/aarch64/arm_neon.h (vceqq_p64, vceqz_p64, vceqzq_p64): Define.

2020-09-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>

	PR target/71233
	* gcc.target/aarch64/simd/vceq_poly_1.c: New test.

[-- Attachment #2: vceq.patch --]
[-- Type: application/octet-stream, Size: 2270 bytes --]

diff --git a/gcc/config/aarch64/arm_neon.h b/gcc/config/aarch64/arm_neon.h
index 577f0771154ed96ce14ecc5cdd4c4ff802b92b69..e2c6a8c5da73851378afad410737e1ca272010ba 100644
--- a/gcc/config/aarch64/arm_neon.h
+++ b/gcc/config/aarch64/arm_neon.h
@@ -12670,6 +12670,13 @@ vceqq_u64 (uint64x2_t __a, uint64x2_t __b)
   return (__a == __b);
 }
 
+__extension__ extern __inline uint64x2_t
+__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
+vceqq_p64 (poly64x2_t __a, poly64x2_t __b)
+{
+  return (__a == __b);
+}
+
 /* vceq - scalar.  */
 
 __extension__ extern __inline uint32_t
@@ -12779,6 +12786,13 @@ vceqz_u64 (uint64x1_t __a)
   return (__a == __AARCH64_UINT64_C (0));
 }
 
+__extension__ extern __inline uint64x1_t
+__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
+vceqz_p64 (poly64x1_t __a)
+{
+  return (__a == __AARCH64_UINT64_C (0));
+}
+
 __extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 vceqzq_f32 (float32x4_t __a)
@@ -12856,6 +12870,13 @@ vceqzq_u64 (uint64x2_t __a)
   return (__a == __AARCH64_UINT64_C (0));
 }
 
+__extension__ extern __inline uint64x2_t
+__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
+vceqzq_p64 (poly64x2_t __a)
+{
+  return (__a == __AARCH64_UINT64_C (0));
+}
+
 /* vceqz - scalar.  */
 
 __extension__ extern __inline uint32_t
diff --git a/gcc/testsuite/gcc.target/aarch64/simd/vceq_poly_1.c b/gcc/testsuite/gcc.target/aarch64/simd/vceq_poly_1.c
new file mode 100644
index 0000000000000000000000000000000000000000..7d43352266c29f2c076dcd0b49b992495f3edb19
--- /dev/null
+++ b/gcc/testsuite/gcc.target/aarch64/simd/vceq_poly_1.c
@@ -0,0 +1,29 @@
+/* { dg-do compile } */
+/* { dg-options "-O" } */
+
+#include <arm_neon.h>
+
+uint64x2_t
+foo (poly64x2_t a, poly64x2_t b)
+{
+  return vceqq_p64 (a, b);
+}
+
+/* { dg-final { scan-assembler-times "cmeq\\tv\[0-9\]+\.2d, v\[0-9\]+\.2d, v\[0-9\]+\.2d" 1 } } */
+
+uint64x1_t
+fooz (poly64x1_t a)
+{
+  return vceqz_p64 (a);
+}
+
+/* { dg-final { scan-assembler-times "cmeq\\td\[0-9\]+, d\[0-9\]+, #0" 1 } } */
+
+uint64x2_t
+fooqz (poly64x2_t a)
+{
+  return vceqzq_p64 (a);
+}
+
+/* { dg-final { scan-assembler-times "cmeq\\tv\[0-9\]+\.2d, v\[0-9\]+\.2d, #0" 1 } } */
+