x86-64: Add vector atanh/atanhf implementation to libmvec

Implement vectorized atanh/atanhf containing SSE, AVX, AVX2 and
AVX512 versions for libmvec as per vector ABI.  It also contains
accuracy and ABI tests for vector atanh/atanhf with regenerated ulps.

Reviewed-by: H.J. Lu <hjl.tools@gmail.com>

50 files changed, 5060 insertions, 1 deletions

diff --git a/bits/libm-simd-decl-stubs.h b/bits/libm-simd-decl-stubs.h
index 845246fab9..bb7380a446 100644
--- a/bits/libm-simd-decl-stubs.h
+++ b/bits/libm-simd-decl-stubs.h
@@ -252,4 +252,15 @@
 #define __DECL_SIMD_log1pf32x
 #define __DECL_SIMD_log1pf64x
 #define __DECL_SIMD_log1pf128x
+
+#define __DECL_SIMD_atanh
+#define __DECL_SIMD_atanhf
+#define __DECL_SIMD_atanhl
+#define __DECL_SIMD_atanhf16
+#define __DECL_SIMD_atanhf32
+#define __DECL_SIMD_atanhf64
+#define __DECL_SIMD_atanhf128
+#define __DECL_SIMD_atanhf32x
+#define __DECL_SIMD_atanhf64x
+#define __DECL_SIMD_atanhf128x
 #endif
diff --git a/math/bits/mathcalls.h b/math/bits/mathcalls.h
index aa4bc61aa4..04dd9c5d1b 100644
--- a/math/bits/mathcalls.h
+++ b/math/bits/mathcalls.h
@@ -86,7 +86,7 @@ __MATHCALL (acosh,, (_Mdouble_ __x));
 /* Hyperbolic arc sine of X.  */
 __MATHCALL (asinh,, (_Mdouble_ __x));
 /* Hyperbolic arc tangent of X.  */
-__MATHCALL (atanh,, (_Mdouble_ __x));
+__MATHCALL_VEC (atanh,, (_Mdouble_ __x));
 #endif
 
 /* Exponential and logarithmic functions.  */
diff --git a/sysdeps/unix/sysv/linux/x86_64/libmvec.abilist b/sysdeps/unix/sysv/linux/x86_64/libmvec.abilist
index 68b940606a..2d389912b1 100644
--- a/sysdeps/unix/sysv/linux/x86_64/libmvec.abilist
+++ b/sysdeps/unix/sysv/linux/x86_64/libmvec.abilist
@@ -49,6 +49,7 @@ GLIBC_2.22 _ZGVeN8vvv_sincos F
 GLIBC_2.35 _ZGVbN2v_acos F
 GLIBC_2.35 _ZGVbN2v_asin F
 GLIBC_2.35 _ZGVbN2v_atan F
+GLIBC_2.35 _ZGVbN2v_atanh F
 GLIBC_2.35 _ZGVbN2v_cbrt F
 GLIBC_2.35 _ZGVbN2v_cosh F
 GLIBC_2.35 _ZGVbN2v_exp10 F
@@ -63,6 +64,7 @@ GLIBC_2.35 _ZGVbN2vv_hypot F
 GLIBC_2.35 _ZGVbN4v_acosf F
 GLIBC_2.35 _ZGVbN4v_asinf F
 GLIBC_2.35 _ZGVbN4v_atanf F
+GLIBC_2.35 _ZGVbN4v_atanhf F
 GLIBC_2.35 _ZGVbN4v_cbrtf F
 GLIBC_2.35 _ZGVbN4v_coshf F
 GLIBC_2.35 _ZGVbN4v_exp10f F
@@ -77,6 +79,7 @@ GLIBC_2.35 _ZGVbN4vv_hypotf F
 GLIBC_2.35 _ZGVcN4v_acos F
 GLIBC_2.35 _ZGVcN4v_asin F
 GLIBC_2.35 _ZGVcN4v_atan F
+GLIBC_2.35 _ZGVcN4v_atanh F
 GLIBC_2.35 _ZGVcN4v_cbrt F
 GLIBC_2.35 _ZGVcN4v_cosh F
 GLIBC_2.35 _ZGVcN4v_exp10 F
@@ -91,6 +94,7 @@ GLIBC_2.35 _ZGVcN4vv_hypot F
 GLIBC_2.35 _ZGVcN8v_acosf F
 GLIBC_2.35 _ZGVcN8v_asinf F
 GLIBC_2.35 _ZGVcN8v_atanf F
+GLIBC_2.35 _ZGVcN8v_atanhf F
 GLIBC_2.35 _ZGVcN8v_cbrtf F
 GLIBC_2.35 _ZGVcN8v_coshf F
 GLIBC_2.35 _ZGVcN8v_exp10f F
@@ -105,6 +109,7 @@ GLIBC_2.35 _ZGVcN8vv_hypotf F
 GLIBC_2.35 _ZGVdN4v_acos F
 GLIBC_2.35 _ZGVdN4v_asin F
 GLIBC_2.35 _ZGVdN4v_atan F
+GLIBC_2.35 _ZGVdN4v_atanh F
 GLIBC_2.35 _ZGVdN4v_cbrt F
 GLIBC_2.35 _ZGVdN4v_cosh F
 GLIBC_2.35 _ZGVdN4v_exp10 F
@@ -119,6 +124,7 @@ GLIBC_2.35 _ZGVdN4vv_hypot F
 GLIBC_2.35 _ZGVdN8v_acosf F
 GLIBC_2.35 _ZGVdN8v_asinf F
 GLIBC_2.35 _ZGVdN8v_atanf F
+GLIBC_2.35 _ZGVdN8v_atanhf F
 GLIBC_2.35 _ZGVdN8v_cbrtf F
 GLIBC_2.35 _ZGVdN8v_coshf F
 GLIBC_2.35 _ZGVdN8v_exp10f F
@@ -133,6 +139,7 @@ GLIBC_2.35 _ZGVdN8vv_hypotf F
 GLIBC_2.35 _ZGVeN16v_acosf F
 GLIBC_2.35 _ZGVeN16v_asinf F
 GLIBC_2.35 _ZGVeN16v_atanf F
+GLIBC_2.35 _ZGVeN16v_atanhf F
 GLIBC_2.35 _ZGVeN16v_cbrtf F
 GLIBC_2.35 _ZGVeN16v_coshf F
 GLIBC_2.35 _ZGVeN16v_exp10f F
@@ -147,6 +154,7 @@ GLIBC_2.35 _ZGVeN16vv_hypotf F
 GLIBC_2.35 _ZGVeN8v_acos F
 GLIBC_2.35 _ZGVeN8v_asin F
 GLIBC_2.35 _ZGVeN8v_atan F
+GLIBC_2.35 _ZGVeN8v_atanh F
 GLIBC_2.35 _ZGVeN8v_cbrt F
 GLIBC_2.35 _ZGVeN8v_cosh F
 GLIBC_2.35 _ZGVeN8v_exp10 F
diff --git a/sysdeps/x86/fpu/bits/math-vector.h b/sysdeps/x86/fpu/bits/math-vector.h
index 14c9db3bb3..4937b6811f 100644
--- a/sysdeps/x86/fpu/bits/math-vector.h
+++ b/sysdeps/x86/fpu/bits/math-vector.h
@@ -114,6 +114,10 @@
 #  define __DECL_SIMD_log1p __DECL_SIMD_x86_64
 #  undef __DECL_SIMD_log1pf
 #  define __DECL_SIMD_log1pf __DECL_SIMD_x86_64
+#  undef __DECL_SIMD_atanh
+#  define __DECL_SIMD_atanh __DECL_SIMD_x86_64
+#  undef __DECL_SIMD_atanhf
+#  define __DECL_SIMD_atanhf __DECL_SIMD_x86_64
 
 # endif
 #endif
diff --git a/sysdeps/x86/fpu/finclude/math-vector-fortran.h b/sysdeps/x86/fpu/finclude/math-vector-fortran.h
index 3dca196432..da39c08ba9 100644
--- a/sysdeps/x86/fpu/finclude/math-vector-fortran.h
+++ b/sysdeps/x86/fpu/finclude/math-vector-fortran.h
@@ -56,6 +56,8 @@
 !GCC$ builtin (log2f) attributes simd (notinbranch) if('x86_64')
 !GCC$ builtin (log1p) attributes simd (notinbranch) if('x86_64')
 !GCC$ builtin (log1pf) attributes simd (notinbranch) if('x86_64')
+!GCC$ builtin (atanh) attributes simd (notinbranch) if('x86_64')
+!GCC$ builtin (atanhf) attributes simd (notinbranch) if('x86_64')
 
 !GCC$ builtin (cos) attributes simd (notinbranch) if('x32')
 !GCC$ builtin (cosf) attributes simd (notinbranch) if('x32')
@@ -97,3 +99,5 @@
 !GCC$ builtin (log2f) attributes simd (notinbranch) if('x32')
 !GCC$ builtin (log1p) attributes simd (notinbranch) if('x32')
 !GCC$ builtin (log1pf) attributes simd (notinbranch) if('x32')
+!GCC$ builtin (atanh) attributes simd (notinbranch) if('x32')
+!GCC$ builtin (atanhf) attributes simd (notinbranch) if('x32')
diff --git a/sysdeps/x86_64/fpu/Makeconfig b/sysdeps/x86_64/fpu/Makeconfig
index 378cb06d37..de87544259 100644
--- a/sysdeps/x86_64/fpu/Makeconfig
+++ b/sysdeps/x86_64/fpu/Makeconfig
@@ -26,6 +26,7 @@ libmvec-funcs = \
   asin \
   atan \
   atan2 \
+  atanh \
   cbrt \
   cos \
   cosh \
diff --git a/sysdeps/x86_64/fpu/Versions b/sysdeps/x86_64/fpu/Versions
index 155fb115f3..df0ea83711 100644
--- a/sysdeps/x86_64/fpu/Versions
+++ b/sysdeps/x86_64/fpu/Versions
@@ -17,6 +17,7 @@ libmvec {
     _ZGVbN2v_acos; _ZGVcN4v_acos; _ZGVdN4v_acos; _ZGVeN8v_acos;
     _ZGVbN2v_asin; _ZGVcN4v_asin; _ZGVdN4v_asin; _ZGVeN8v_asin;
     _ZGVbN2v_atan; _ZGVcN4v_atan; _ZGVdN4v_atan; _ZGVeN8v_atan;
+    _ZGVbN2v_atanh; _ZGVcN4v_atanh; _ZGVdN4v_atanh; _ZGVeN8v_atanh;
     _ZGVbN2v_cbrt; _ZGVcN4v_cbrt; _ZGVdN4v_cbrt; _ZGVeN8v_cbrt;
     _ZGVbN2v_cosh; _ZGVcN4v_cosh; _ZGVdN4v_cosh; _ZGVeN8v_cosh;
     _ZGVbN2v_exp10; _ZGVcN4v_exp10; _ZGVdN4v_exp10; _ZGVeN8v_exp10;
@@ -31,6 +32,7 @@ libmvec {
     _ZGVbN4v_acosf; _ZGVcN8v_acosf; _ZGVdN8v_acosf; _ZGVeN16v_acosf;
     _ZGVbN4v_asinf; _ZGVcN8v_asinf; _ZGVdN8v_asinf; _ZGVeN16v_asinf;
     _ZGVbN4v_atanf; _ZGVcN8v_atanf; _ZGVdN8v_atanf; _ZGVeN16v_atanf;
+    _ZGVbN4v_atanhf; _ZGVcN8v_atanhf; _ZGVdN8v_atanhf; _ZGVeN16v_atanhf;
     _ZGVbN4v_cbrtf; _ZGVcN8v_cbrtf; _ZGVdN8v_cbrtf; _ZGVeN16v_cbrtf;
     _ZGVbN4v_coshf; _ZGVcN8v_coshf; _ZGVdN8v_coshf; _ZGVeN16v_coshf;
     _ZGVbN4v_exp10f; _ZGVcN8v_exp10f; _ZGVdN8v_exp10f; _ZGVeN16v_exp10f;
diff --git a/sysdeps/x86_64/fpu/libm-test-ulps b/sysdeps/x86_64/fpu/libm-test-ulps
index a2b15a795b..09a46190b6 100644
--- a/sysdeps/x86_64/fpu/libm-test-ulps
+++ b/sysdeps/x86_64/fpu/libm-test-ulps
@@ -248,6 +248,26 @@ float: 3
 float128: 4
 ldouble: 5
 
+Function: "atanh_vlen16":
+float: 1
+
+Function: "atanh_vlen2":
+double: 1
+
+Function: "atanh_vlen4":
+double: 1
+float: 1
+
+Function: "atanh_vlen4_avx2":
+double: 1
+
+Function: "atanh_vlen8":
+double: 1
+float: 1
+
+Function: "atanh_vlen8_avx2":
+float: 1
+
 Function: "cabs":
 double: 1
 float128: 1
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_atanh2_core-sse2.S b/sysdeps/x86_64/fpu/multiarch/svml_d_atanh2_core-sse2.S
new file mode 100644
index 0000000000..b154ab8649
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_atanh2_core-sse2.S
@@ -0,0 +1,20 @@
+/* SSE2 version of vectorized atanh, vector length is 2.
+   Copyright (C) 2021 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <https://www.gnu.org/licenses/>.  */
+
+#define _ZGVbN2v_atanh _ZGVbN2v_atanh_sse2
+#include "../svml_d_atanh2_core.S"
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_atanh2_core.c b/sysdeps/x86_64/fpu/multiarch/svml_d_atanh2_core.c
new file mode 100644
index 0000000000..138190e568
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_atanh2_core.c
@@ -0,0 +1,27 @@
+/* Multiple versions of vectorized atanh, vector length is 2.
+   Copyright (C) 2021 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <https://www.gnu.org/licenses/>.  */
+
+#define SYMBOL_NAME _ZGVbN2v_atanh
+#include "ifunc-mathvec-sse4_1.h"
+
+libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ());
+
+#ifdef SHARED
+__hidden_ver1 (_ZGVbN2v_atanh, __GI__ZGVbN2v_atanh, __redirect__ZGVbN2v_atanh)
+  __attribute__ ((visibility ("hidden")));
+#endif
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_atanh2_core_sse4.S b/sysdeps/x86_64/fpu/multiarch/svml_d_atanh2_core_sse4.S
new file mode 100644
index 0000000000..7e70b036f7
--- /dev/null
+++ b/