This is infrastructure update for configure, build and install of vector

math library. Installation of libm.so as linker script only in case of libmvec.so build. 2015-04-28 Andrew Senkevich <andrew.n.senkevich@gmail.com> * Makeconfig (rpath-dirs, all-subdirs): Added mathvec folder. (libmvec): New variable. * configure.ac: Added option for mathvec build. * configure: Regenerated. * mathvec/Depend: New file. * mathvec/Makefile: New file. * shlib-versions: Added libmvec. * math/Makefile: Added rule for libm.so installation.
This patch adds infrastructure for addition of SIMD
2026-09-09 05:59:27 +08:00 · 2015-04-28 13:48:23 +03:00 · 2015-04-28 13:48:22 +03:00 · 2015-04-28 13:48:22 +03:00 · 2015-04-28 13:48:22 +03:00 · 2015-04-28 13:48:22 +03:00
19 changed files with 296 additions and 33 deletions
@@ -498,7 +498,7 @@ link-libc = $(link-libc-rpath-link) $(link-libc-before-gnulib) $(gnulib)
 link-libc-tests = $(link-libc-tests-rpath-link) \
 		  $(link-libc-before-gnulib) $(gnulib-tests)
 # This is how to find at build-time things that will be installed there.
-rpath-dirs = math elf dlfcn nss nis rt resolv crypt
+rpath-dirs = math elf dlfcn nss nis rt resolv crypt mathvec
 rpath-link = \
 $(common-objdir):$(subst $(empty) ,:,$(patsubst ../$(subdir),.,$(rpath-dirs:%=$(common-objpfx)%)))
 else
@@ -1074,8 +1074,10 @@ endif

 ifeq ($(build-shared),yes)
 libm = $(common-objpfx)math/libm.so$(libm.so-version)
+libmvec = $(common-objpfx)mathvec/libmvec.so$(libmvec.so-version)
 else
 libm = $(common-objpfx)math/libm.a
+libmvec = $(common-objpfx)mathvec/libmvec.a
 endif

 # These are the subdirectories containing the library source.  The order
@@ -1085,7 +1087,7 @@ all-subdirs = csu assert ctype locale intl catgets math setjmp signal	    \
 	      stdlib stdio-common libio malloc string wcsmbs time dirent    \
 	      grp pwd posix io termios resource misc socket sysvipc gmon    \
 	      gnulib iconv iconvdata wctype manual shadow gshadow po argp   \
-	      crypt localedata timezone rt conform debug		    \
+	      crypt localedata timezone rt conform debug mathvec	    \
 	      $(add-on-subdirs) dlfcn elf

 ifndef avoid-generated
@@ -0,0 +1,40 @@
+/* Empty definitions required for __MATHCALL_VEC unfolding in mathcalls.h.
+   Copyright (C) 2015 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
+   <http://www.gnu.org/licenses/>.  */
+
+#ifndef _MATH_H
+# error "Never include <bits/libm-simd-decl-stubs.h> directly;\
+ include <math.h> instead."
+#endif
+
+/* Needed definitions could be generated with:
+   for func in $(grep __MATHCALL_VEC math/bits/mathcalls.h |\
+		 sed -r "s|__MATHCALL_VEC.?\(||; s|,.*||"); do
+     echo "#define __DECL_SIMD_${func}";
+     echo "#define __DECL_SIMD_${func}f";
+     echo "#define __DECL_SIMD_${func}l";
+   done
+ */
+
+#ifndef _BITS_LIBM_SIMD_DECL_STUBS_H
+#define _BITS_LIBM_SIMD_DECL_STUBS_H 1
+
+#define __DECL_SIMD_cos
+#define __DECL_SIMD_cosf
+#define __DECL_SIMD_cosl
+
+#endif
@@ -0,0 +1,27 @@
+/* Platform-specific SIMD declarations of math functions.
+   Copyright (C) 2015 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  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
+   <http://www.gnu.org/licenses/>.  */
+
+#ifndef _MATH_H
+# error "Never include <bits/math-vector.h> directly;\
+ include <math.h> instead."
+#endif
+
+/* Get default empty definitions required for __MATHCALL_VEC unfolding.
+   Plaform-specific analogue of this header should redefine them with specific
+   SIMD declarations.  */
+#include <bits/libm-simd-decl-stubs.h>
@@ -777,6 +777,7 @@ enable_systemtap
 enable_build_nscd
 enable_nscd
 enable_pt_chown
+enable_mathvec
 with_cpu
 '
      ac_precious_vars='build_alias
@@ -1441,6 +1442,8 @@ Optional Features:
  --disable-build-nscd    disable building and installing the nscd daemon
  --disable-nscd          library functions will not contact the nscd daemon
  --enable-pt_chown       Enable building and installing pt_chown
+  --enable-mathvec        Enable building and installing mathvec [default
+                          depends on architecture]

 Optional Packages:
  --with-PACKAGE[=ARG]    use PACKAGE [ARG=yes]
@@ -3845,6 +3848,14 @@ gnu*)
  ;;
 esac

+# Check whether --enable-mathvec was given.
+if test "${enable_mathvec+set}" = set; then :
+  enableval=$enable_mathvec; build_mathvec=$enableval
+else
+  build_mathvec=notset
+fi
+
+
 # We keep the original values in `$config_*' and never modify them, so we
 # can write them unchanged into config.make.  Everything else uses
 # $machine, $vendor, and $os, and changes them whenever convenient.
@@ -7215,6 +7226,12 @@ $as_echo "running configure fragment for $dir" >&6; }
  fi
 done

+if test x"$build_mathvec" = xnotset; then
+  build_mathvec=no
+fi
+config_vars="$config_vars
+build-mathvec = $build_mathvec"
+



@@ -381,6 +381,12 @@ gnu*)
  ;;
 esac

+AC_ARG_ENABLE([mathvec],
+	      [AS_HELP_STRING([--enable-mathvec],
+	      [Enable building and installing mathvec @<:@default depends on architecture@:>@])],
+	      [build_mathvec=$enableval],
+	      [build_mathvec=notset])
+
 # We keep the original values in `$config_*' and never modify them, so we
 # can write them unchanged into config.make.  Everything else uses
 # $machine, $vendor, and $os, and changes them whenever convenient.
@@ -1994,6 +2000,11 @@ for dir in $sysnames; do
  fi
 done

+if test x"$build_mathvec" = xnotset; then
+  build_mathvec=no
+fi
+LIBC_CONFIG_VAR([build-mathvec], [$build_mathvec])
+
 AC_SUBST(libc_extra_cflags)
 AC_SUBST(libc_extra_cppflags)

@@ -26,7 +26,8 @@ headers		:= math.h bits/mathcalls.h bits/mathinline.h bits/huge_val.h \
 		   bits/huge_valf.h bits/huge_vall.h bits/inf.h bits/nan.h \
 		   fpu_control.h complex.h bits/cmathcalls.h fenv.h \
 		   bits/fenv.h bits/fenvinline.h bits/mathdef.h tgmath.h \
-		   bits/math-finite.h
+		   bits/math-finite.h bits/math-vector.h \
+		   bits/libm-simd-decl-stubs.h

 # FPU support code.
 aux		:= setfpucw fpu_control
@@ -85,6 +86,22 @@ generated += $(foreach s,.c .S l.c l.S f.c f.S,$(calls:s_%=m_%$s))
 routines = $(calls) $(calls:=f) $(long-c-$(long-double-fcts))
 long-c-yes = $(calls:=l)

+ifeq ($(build-mathvec),yes)
+# We need to install libm.so as linker script
+# for more comfortable use of vector math library.
+install-lib-ldscripts := libm.so
+install_subdir: $(inst_libdir)/libm.so
+$(inst_libdir)/libm.so: $(common-objpfx)format.lds \
+			$(libm) \
+			$(libmvec) \
+			$(+force)
+	(echo '/* GNU ld script'; echo '*/';\
+	 cat $<; \
+	 echo 'GROUP ( $(slibdir)/libm.so$(libm.so-version) ' \
+	      'AS_NEEDED ( $(slibdir)/libmvec.so$(libmvec.so-version) ) )' \
+	) > $@
+endif
+
 # Rules for the test suite.
 tests = test-matherr test-fenv atest-exp atest-sincos atest-exp2 basic-test \
 	test-misc test-fpucw test-fpucw-ieee tst-definitions test-tgmath \
@@ -103,7 +120,7 @@ libm-tests = test-float test-double $(test-longdouble-$(long-double-fcts)) \
 libm-tests.o = $(addsuffix .o,$(libm-tests))

 tests += $(libm-tests)
-libm-tests-generated = libm-test-ulps.h libm-test.c
+libm-tests-generated = libm-test-ulps.h libm-have-vector-test.h libm-test.c
 generated += $(libm-tests-generated) libm-test.stmp

 # This is needed for dependencies
@@ -114,9 +131,10 @@ ulps-file = $(firstword $(wildcard $(sysdirs:%=%/libm-test-ulps)))
 $(addprefix $(objpfx), $(libm-tests-generated)): $(objpfx)libm-test.stmp

 $(objpfx)libm-test.stmp: $(ulps-file) libm-test.inc gen-libm-test.pl \
-			 auto-libm-test-out
+			 gen-libm-have-vector-test.sh auto-libm-test-out
 	$(make-target-directory)
 	$(PERL) gen-libm-test.pl -u $< -o "$(objpfx)"
+	$(SHELL) gen-libm-have-vector-test.sh > $(objpfx)libm-have-vector-test.h
 	@echo > $@

 $(objpfx)test-float.o: $(objpfx)libm-test.stmp
@@ -0,0 +1,47 @@
+#!/bin/sh
+# Copyright (C) 1999-2015 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
+# <http://www.gnu.org/licenses/>.
+
+# Generate series of definitions used for vector math functions tests.
+# TEST_VECTOR_* and WRAPPER_NAME are defined in vector math functions tests.
+# *_VEC_SUFF is used in individual tests, as result of FUNC_TEST unfolding
+# to avoid warnings / errors about undeclared functions.
+print_defs()
+{
+  echo "#if defined TEST_VECTOR_$1 && TEST_VECTOR_$1"
+  echo "# define HAVE_VECTOR_$1 1"
+  echo "# define ${1}_VEC_SUFF WRAPPER_NAME ($1)"
+  echo "#else"
+  echo "# define HAVE_VECTOR_$1 0"
+  echo "# define ${1}_VEC_SUFF $1"
+  echo "#endif"
+  echo
+}
+
+for func in $(grep ALL_RM_TEST libm-test.inc | grep -v define | sed -r "s/.*\(//; s/,.*//"); do
+  print_defs ${func}
+  print_defs ${func}f
+  print_defs ${func}l
+done
+
+# When all functions will use ALL_RM_TEST instead of using START directly,
+# this code can be removed.
+for func in $(grep 'START.*;$' libm-test.inc | sed -r "s/.*\(//; s/,.*//"); do
+  print_defs ${func}
+  print_defs ${func}f
+  print_defs ${func}l
+done
@@ -126,6 +126,7 @@
 #include <argp.h>
 #include <tininess.h>
 #include <math-tests.h>
+#include <math-tests-arch.h>

 /* Structure for ulp data for a function, or the real or imaginary
   part of a function.  */
@@ -685,7 +686,7 @@ static void
 test_single_errno (const char *test_name, int errno_value,
 		   int expected_value, const char *expected_name)
 {
-#ifndef TEST_INLINE
+#if !defined TEST_INLINE && !TEST_MATHVEC
  if (errno_value == expected_value)
    {
      if (print_screen (1))
@@ -1302,8 +1303,8 @@ struct test_fFF_11_data
    if (enable_test (EXCEPTIONS))					\
      {									\
 	COMMON_TEST_SETUP (ARG_STR);					\
-	check_float (test_name, FUNC (FUNC_NAME) (ARG), EXPECTED,	\
-		     EXCEPTIONS);					\
+	check_float (test_name,	FUNC_TEST (FUNC_NAME) (ARG),		\
+		     EXPECTED, EXCEPTIONS);				\
 	COMMON_TEST_CLEANUP;						\
      }									\
  while (0)
@@ -1320,7 +1321,7 @@ struct test_fFF_11_data
    if (enable_test (EXCEPTIONS))				\
      {								\
 	COMMON_TEST_SETUP (ARG_STR);				\
-	check_float (test_name, FUNC (FUNC_NAME) (ARG1, ARG2),	\
+	check_float (test_name, FUNC_TEST (FUNC_NAME) (ARG1, ARG2),	\
 		     EXPECTED, EXCEPTIONS);			\
 	COMMON_TEST_CLEANUP;					\
      }								\
@@ -1347,7 +1348,7 @@ struct test_fFF_11_data
    if (enable_test (EXCEPTIONS))					\
      {									\
 	COMMON_TEST_SETUP (ARG_STR);					\
-	check_float (test_name, FUNC (FUNC_NAME) (ARG1, ARG2, ARG3),	\
+	check_float (test_name, FUNC_TEST (FUNC_NAME) (ARG1, ARG2, ARG3),	\
 		     EXPECTED, EXCEPTIONS);				\
 	COMMON_TEST_CLEANUP;						\
      }									\
@@ -1367,7 +1368,7 @@ struct test_fFF_11_data
      {									\
 	COMMON_TEST_SETUP (ARG_STR);					\
 	check_float (test_name,						\
-		     FUNC (FUNC_NAME) (BUILD_COMPLEX (ARG1, ARG2)),	\
+		     FUNC_TEST (FUNC_NAME) (BUILD_COMPLEX (ARG1, ARG2)),\
 		     EXPECTED, EXCEPTIONS);				\
 	COMMON_TEST_CLEANUP;						\
      }									\
@@ -1388,7 +1389,7 @@ struct test_fFF_11_data
      {									\
 	COMMON_TEST_SETUP (ARG_STR);					\
 	(EXTRA_VAR) = (EXTRA_EXPECTED) == 0 ? 1 : 0;			\
-	check_float (test_name, FUNC (FUNC_NAME) (ARG), EXPECTED,	\
+	check_float (test_name, FUNC_TEST (FUNC_NAME) (ARG), EXPECTED,	\
 		     EXCEPTIONS);					\
 	EXTRA_OUTPUT_TEST_SETUP (ARG_STR, 1);				\
 	if (EXTRA_TEST)							\
@@ -1415,7 +1416,7 @@ struct test_fFF_11_data
      {									\
 	COMMON_TEST_SETUP (ARG_STR);					\
 	(EXTRA_VAR) = (EXTRA_EXPECTED) == 0 ? 1 : 0;			\
-	check_float (test_name, FUNC (FUNC_NAME) (ARG, &(EXTRA_VAR)),	\
+	check_float (test_name, FUNC_TEST (FUNC_NAME) (ARG, &(EXTRA_VAR)),	\
 		     EXPECTED, EXCEPTIONS);				\
 	EXTRA_OUTPUT_TEST_SETUP (ARG_STR, 1);				\
 	if (EXTRA_TEST)							\
@@ -1442,7 +1443,7 @@ struct test_fFF_11_data
      {									\
 	COMMON_TEST_SETUP (ARG_STR);					\
 	(EXTRA_VAR) = (EXTRA_EXPECTED) == 0 ? 1 : 0;			\
-	check_float (test_name, FUNC (FUNC_NAME) (ARG, &(EXTRA_VAR)),	\
+	check_float (test_name, FUNC_TEST (FUNC_NAME) (ARG, &(EXTRA_VAR)),	\
 		     EXPECTED, EXCEPTIONS);				\
 	EXTRA_OUTPUT_TEST_SETUP (ARG_STR, 1);				\
 	if (EXTRA_TEST)							\
@@ -1470,7 +1471,7 @@ struct test_fFF_11_data
 	COMMON_TEST_SETUP (ARG_STR);					\
 	(EXTRA_VAR) = (EXTRA_EXPECTED) == 0 ? 1 : 0;			\
 	check_float (test_name,						\
-		     FUNC (FUNC_NAME) (ARG1, ARG2, &(EXTRA_VAR)),	\
+		     FUNC_TEST (FUNC_NAME) (ARG1, ARG2, &(EXTRA_VAR)),	\
 		     EXPECTED, EXCEPTIONS);				\
 	EXTRA_OUTPUT_TEST_SETUP (ARG_STR, 1);				\
 	if (EXTRA_TEST)							\
@@ -1498,7 +1499,7 @@ struct test_fFF_11_data
      {									\
 	COMMON_TEST_SETUP (ARG_STR);					\
 	check_complex (test_name,					\
-		       FUNC (FUNC_NAME) (BUILD_COMPLEX (ARGR, ARGC)),	\
+		       FUNC_TEST (FUNC_NAME) (BUILD_COMPLEX (ARGR, ARGC)),	\
 		       BUILD_COMPLEX (EXPR, EXPC), EXCEPTIONS);		\
 	COMMON_TEST_CLEANUP;						\
      }									\
@@ -1519,8 +1520,8 @@ struct test_fFF_11_data
      {									\
 	COMMON_TEST_SETUP (ARG_STR);					\
 	check_complex (test_name,					\
-		       FUNC (FUNC_NAME) (BUILD_COMPLEX (ARG1R, ARG1C),	\
-					 BUILD_COMPLEX (ARG2R, ARG2C)),	\
+		       FUNC_TEST (FUNC_NAME) (BUILD_COMPLEX (ARG1R, ARG1C),	\
+					      BUILD_COMPLEX (ARG2R, ARG2C)),	\
 		       BUILD_COMPLEX (EXPR, EXPC), EXCEPTIONS);		\
 	COMMON_TEST_CLEANUP;						\
      }									\
@@ -1540,7 +1541,7 @@ struct test_fFF_11_data
    if (enable_test (EXCEPTIONS))					\
      {									\
 	COMMON_TEST_SETUP (ARG_STR);					\
-	check_int (test_name, FUNC (FUNC_NAME) (ARG), EXPECTED,		\
+	check_int (test_name, FUNC_TEST (FUNC_NAME) (ARG), EXPECTED,	\
 		   EXCEPTIONS);						\
 	COMMON_TEST_CLEANUP;						\
      }									\
@@ -1593,7 +1594,7 @@ struct test_fFF_11_data
    if (enable_test (EXCEPTIONS))					\
      {									\
 	COMMON_TEST_SETUP (ARG_STR);					\
-	check_bool (test_name, FUNC (FUNC_NAME) (ARG), EXPECTED,	\
+	check_bool (test_name, FUNC_TEST (FUNC_NAME) (ARG), EXPECTED,	\
 		    EXCEPTIONS);					\
 	COMMON_TEST_CLEANUP;						\
      }									\
@@ -1627,7 +1628,7 @@ struct test_fFF_11_data
    if (enable_test (EXCEPTIONS))					\
      {									\
 	COMMON_TEST_SETUP (ARG_STR);					\
-	check_long (test_name, FUNC (FUNC_NAME) (ARG), EXPECTED,	\
+	check_long (test_name, FUNC_TEST (FUNC_NAME) (ARG), EXPECTED,	\
 		    EXCEPTIONS);					\
 	COMMON_TEST_CLEANUP;						\
      }									\
@@ -1644,8 +1645,8 @@ struct test_fFF_11_data
    if (enable_test (EXCEPTIONS))					\
      {									\
 	COMMON_TEST_SETUP (ARG_STR);					\
-	check_longlong (test_name, FUNC (FUNC_NAME) (ARG), EXPECTED,	\
-			EXCEPTIONS);					\
+	check_longlong (test_name, FUNC_TEST (FUNC_NAME) (ARG),		\
+			EXPECTED, EXCEPTIONS);				\
 	COMMON_TEST_CLEANUP;						\
      }									\
  while (0)
@@ -1664,7 +1665,7 @@ struct test_fFF_11_data
    if (enable_test (EXCEPTIONS))					\
      {									\
 	COMMON_TEST_SETUP (ARG_STR);					\
-	FUNC (FUNC_NAME) (ARG, &(EXTRA1_VAR), &(EXTRA2_VAR));		\
+	FUNC_TEST (FUNC_NAME) (ARG, &(EXTRA1_VAR), &(EXTRA2_VAR));	\
 	EXTRA_OUTPUT_TEST_SETUP (ARG_STR, 1);				\
 	if (EXTRA1_TEST)						\
 	  check_float (extra1_name, EXTRA1_VAR, EXTRA1_EXPECTED,	\
@@ -1691,9 +1692,23 @@ struct test_fFF_11_data
 		       (ARRAY)[i].RM_##ROUNDING_MODE.extra2_expected);	\
  ROUND_RESTORE_ ## ROUNDING_MODE

+#if !TEST_MATHVEC
+# define VEC_SUFF
+#endif
+
+#define STR_CONCAT(a, b, c) __STRING (a##b##c)
+#define STR_CON3(a, b, c) STR_CONCAT (a, b, c)
+
+/* This generated header defines series of macros started with HAVE_VECTOR_. */
+#include "libm-have-vector-test.h"
+
+#define HAVE_VECTOR(func) __CONCAT (HAVE_VECTOR_, func)
+
 /* Start and end the tests for a given function.  */
-#define START(FUNC, EXACT)			\
-  const char *this_func = #FUNC;		\
+#define START(FUN, SUFF, EXACT)					\
+  CHECK_ARCH_EXT;						\
+  if (TEST_MATHVEC && !HAVE_VECTOR (FUNC (FUN))) return;	\
+  const char *this_func = STR_CON3 (FUN, SUFF, VEC_SUFF);	\
  init_max_error (this_func, EXACT)
 #define END					\
  print_max_error (this_func)
@@ -1706,28 +1721,28 @@ struct test_fFF_11_data
    {									\
      do								\
 	{								\
-	  START (FUNC, EXACT);						\
+	  START (FUNC,, EXACT);						\
 	  LOOP_MACRO (FUNC, ARRAY, , ## __VA_ARGS__);			\
 	  END_MACRO;							\
 	}								\
      while (0);							\
      do								\
 	{								\
-	  START (FUNC ## _downward, EXACT);				\
+	  START (FUNC, _downward, EXACT);				\
 	  LOOP_MACRO (FUNC, ARRAY, FE_DOWNWARD, ## __VA_ARGS__);	\
 	  END_MACRO;							\
 	}								\
      while (0);							\
      do								\
 	{								\
-	  START (FUNC ## _towardzero, EXACT);				\
+	  START (FUNC, _towardzero, EXACT);				\
 	  LOOP_MACRO (FUNC, ARRAY, FE_TOWARDZERO, ## __VA_ARGS__);	\
 	  END_MACRO;							\
 	}								\
      while (0);							\
      do								\
 	{								\
-	  START (FUNC ## _upward, EXACT);				\
+	  START (FUNC, _upward, EXACT);				\
 	  LOOP_MACRO (FUNC, ARRAY, FE_UPWARD, ## __VA_ARGS__);		\
 	  END_MACRO;							\
 	}								\
@@ -6025,7 +6040,7 @@ static const struct test_c_c_data cexp_test_data[] =
 static void
 cexp_test (void)
 {
-  START (cexp, 0);
+  START (cexp,, 0);
  RUN_TEST_LOOP_c_c (cexp, cexp_test_data, );
  END_COMPLEX;
 }
@@ -7544,7 +7559,7 @@ static const struct test_if_f_data jn_test_data[] =
 static void
 jn_test (void)
 {
-  START (jn, 0);
+  START (jn,, 0);
  RUN_TEST_LOOP_if_f (jn, jn_test_data, );
  END;
 }
@@ -9506,7 +9521,7 @@ static const struct test_f_f_data tgamma_test_data[] =
 static void
 tgamma_test (void)
 {
-  START (tgamma, 0);
+  START (tgamma,, 0);
  RUN_TEST_LOOP_f_f (tgamma, tgamma_test_data, );
  END;
 }
@@ -9956,6 +9971,8 @@ main (int argc, char **argv)
  initialize ();
  printf (TEST_MSG);

+  INIT_ARCH_EXT;
+
  check_ulp ();

  /* Keep the tests a wee bit ordered (according to ISO C99).  */
@@ -27,6 +27,9 @@

 __BEGIN_DECLS

+/* Get machine-dependent vector math functions declarations.  */
+#include <bits/math-vector.h>
+
 /* Get machine-dependent HUGE_VAL value (returned on overflow).
   On all IEEE754 machines, this is +Infinity.  */
 #include <bits/huge_val.h>
@@ -49,6 +52,12 @@ __BEGIN_DECLS
   so we can easily declare each function as both `name' and `__name',
   and can declare the float versions `namef' and `__namef'.  */

+#define __SIMD_DECL(function) __CONCAT (__DECL_SIMD_, function)
+
+#define __MATHCALL_VEC(function, suffix, args) 	\
+  __SIMD_DECL (__MATH_PRECNAME (function, suffix)) \
+  __MATHCALL (function, suffix, args)
+
 #define __MATHCALL(function,suffix, args)	\
  __MATHDECL (_Mdouble_,function,suffix, args)
 #define __MATHDECL(type, function,suffix, args) \
@@ -17,6 +17,7 @@
   <http://www.gnu.org/licenses/>.  */

 #define FUNC(function) function
+#define FUNC_TEST(function) FUNC (function)
 #define FLOAT double
 #define TEST_MSG "testing double (without inline functions)\n"
 #define MATHCONST(x) x
@@ -25,6 +26,7 @@
 #define PRINTF_XEXPR "a"
 #define PRINTF_NEXPR "f"
 #define TEST_DOUBLE 1
+#define TEST_MATHVEC 0

 #ifndef __NO_MATH_INLINES
 # define __NO_MATH_INLINES
@@ -17,6 +17,7 @@
   <http://www.gnu.org/licenses/>.  */

 #define FUNC(function) function ## f
+#define FUNC_TEST(function) FUNC (function)
 #define FLOAT float
 #define TEST_MSG "testing float (without inline functions)\n"
 #define MATHCONST(x) x
@@ -25,6 +26,7 @@
 #define PRINTF_XEXPR "a"
 #define PRINTF_NEXPR "f"
 #define TEST_FLOAT 1
+#define TEST_MATHVEC 0

 #ifndef __NO_MATH_INLINES
 # define __NO_MATH_INLINES
@@ -17,6 +17,7 @@
   <http://www.gnu.org/licenses/>.  */

 #define FUNC(function) function
+#define FUNC_TEST(function) FUNC (function)
 #define FLOAT double
 #define TEST_MSG "testing double (inline functions)\n"
 #define MATHCONST(x) x
@@ -25,6 +26,7 @@
 #define PRINTF_XEXPR "a"
 #define PRINTF_NEXPR "f"
 #define TEST_DOUBLE 1
+#define TEST_MATHVEC 0
 #define TEST_INLINE

 #ifdef __NO_MATH_INLINES
@@ -17,6 +17,7 @@
   <http://www.gnu.org/licenses/>.  */

 #define FUNC(function) function ## f
+#define FUNC_TEST(function) FUNC (function)
 #define FLOAT float
 #define TEST_MSG "testing float (inline functions)\n"
 #define MATHCONST(x) x
@@ -25,6 +26,7 @@
 #define PRINTF_XEXPR "a"
 #define PRINTF_NEXPR "f"
 #define TEST_FLOAT 1
+#define TEST_MATHVEC 0
 #define TEST_INLINE 1

 #ifdef __NO_MATH_INLINES
@@ -17,6 +17,7 @@
   <http://www.gnu.org/licenses/>.  */

 #define FUNC(function) function##l
+#define FUNC_TEST(function) FUNC (function)
 #define FLOAT long double
 #define TEST_MSG "testing long double (inline functions)\n"
 #define MATHCONST(x) x##L
@@ -26,6 +27,7 @@
 #define PRINTF_NEXPR "Lf"
 #define TEST_INLINE
 #define TEST_LDOUBLE 1
+#define TEST_MATHVEC 0

 #ifdef __NO_MATH_INLINES
 # undef __NO_MATH_INLINES
@@ -17,6 +17,7 @@
   <http://www.gnu.org/licenses/>.  */

 #define FUNC(function) function##l
+#define FUNC_TEST(function) FUNC (function)
 #define FLOAT long double
 #define TEST_MSG "testing long double (without inline functions)\n"
 #define MATHCONST(x) x##L
@@ -25,6 +26,7 @@
 #define PRINTF_XEXPR "La"
 #define PRINTF_NEXPR "Lf"
 #define TEST_LDOUBLE 1
+#define TEST_MATHVEC 0

 #ifndef __NO_MATH_INLINES
 # define __NO_MATH_INLINES
@@ -0,0 +1 @@
+math
@@ -0,0 +1,35 @@
+# Copyright (C) 2015 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
+# <http://www.gnu.org/licenses/>.
+
+# Makefile for the vector math library.
+
+subdir		:= mathvec
+
+include ../Makeconfig
+
+ifeq ($(build-mathvec),yes)
+extra-libs	:= libmvec
+extra-libs-others = $(extra-libs)
+
+libmvec-routines = $(strip $(libmvec-support))
+
+$(objpfx)libmvec.so: $(libm)
+endif
+
+# Rules for the test suite are in math directory.
+
+include ../Rules
@@ -71,3 +71,6 @@ libanl=1
 # This defines the libgcc soname version this glibc is to load for
 # asynchronous cancellation to work correctly.
 libgcc_s=1
+
+# The vector math library
+libmvec=1
@@ -0,0 +1,24 @@
+/* Runtime architecture check for math tests.
+   Copyright (C) 2015 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
+   <http://www.gnu.org/licenses/>.  */
+
+/* These macros used for architecture check in math tests runtime.
+   INIT_ARCH_EXT should set up for example some global variable which is
+   checked by CHECK_ARCH_EXT which produces return from individual test to
+   prevent run on hardware not supported by tested function implementation. */
+#define INIT_ARCH_EXT
+#define CHECK_ARCH_EXT
Author	SHA1	Message	Date
Andrew Senkevich	ec3a4b85a8	This is infrastructure update for configure, build and install of vector math library. Installation of libm.so as linker script only in case of libmvec.so build. 2015-04-28 Andrew Senkevich <andrew.n.senkevich@gmail.com> * Makeconfig (rpath-dirs, all-subdirs): Added mathvec folder. (libmvec): New variable. * configure.ac: Added option for mathvec build. * configure: Regenerated. * mathvec/Depend: New file. * mathvec/Makefile: New file. * shlib-versions: Added libmvec. * math/Makefile: Added rule for libm.so installation.	2015-04-28 13:48:23 +03:00
Andrew Senkevich	a50ac6ec0a	This patch adds infrastructure for addition of SIMD declarations for math functions in math.h. Added new headers math-vector.h only generic version for now and libm-simd-decl-stubs.h with empty definitions required for proper unfolding of new macros __MATHCALL_VEC which will be used for declaration of vector math functions. 2015-04-28 Andrew Senkevich <andrew.senkevich@intel.com> * bits/math-vector.h: New file. * bits/libm-simd-decl-stubs.h: New header. * math/Makefile (headers): Added new header libm-simd-decl-stubs.h. * math/math.h (__MATHCALL_VEC): New macro.	2015-04-28 13:48:22 +03:00
Andrew Senkevich	93c1dba3b4	Last part of changes regarding to libm-test.inc: addition of method for separation which exactly testing function needed to run with help of generated during make check header with series of conditional definitions. 2015-04-28 Andrew Senkevich <andrew.senkevich@intel.com> * math/gen-libm-have-vector-test.sh: Script generates series of macros for conditions in testing functions. * math/Makefile: Added call of libm-have-vector-test.sh. * math/libm-test.inc (HAVE_VECTOR): New macros.	2015-04-28 13:48:22 +03:00
Andrew Senkevich	a0f4fa7b6e	Refactoring of START for conditions in individual tests and addition of macros used for runtime architecture check. 2015-04-28 Andrew Senkevich <andrew.senkevich@intel.com> * math/libm-test.inc: START refactored. * math/test-double.c (TEST_MATHVEC): Add define. * math/test-float.c: Likewise. * math/test-idouble.c: Likewise. * math/test-ifloat.c: Likewise. * math/test-ildoubl.c: Likewise. * math/test-ldouble.c: Likewise. * sysdeps/generic/math-tests-arch.h (INIT_ARCH_EXT, CHECK_ARCH_EXT): New helper macros for runtime architecture check.	2015-04-28 13:48:22 +03:00
Andrew Senkevich	0d7660a2fc	This is the beginning of series of patches with addition of vector math functions infrastructure. This patch is preparatory change in libm-test.c - splitting of macros which form name of tested functions for ability to use separate name for tested functions and for functions used in test suite infrastructure. 2015-04-28 Andrew Senkevich <andrew.senkevich@intel.com> * math/test-double.c (FUNC_TEST): New macro. * math/test-float.c: Likewise. * math/test-idouble.c: Likewise. * math/test-ifloat.c: Likewise. * math/test-ildoubl.c: Likewise. * math/test-ldouble.c: Likewise. * math/libm-test.inc: Use FUNC_TEST for name of tested functions.	2015-04-28 13:48:22 +03:00