#!/usr/bin/python3 # Update libquadmath code from glibc sources. # Copyright (C) 2018 Free Software Foundation, Inc. # This file is part of the libquadmath library. # # Libquadmath 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. # # Libquadmath 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 libquadmath; if not, see # . # Usage: update-quadmath.py glibc_srcdir quadmath_srcdir import argparse from collections import defaultdict import os.path import re def replace_in_file(repl_map, extra_map, src, dest): """Apply the replacements in repl_map, then those in extra_map, to the file src, producing dest.""" with open(src, 'r') as src_file: text = src_file.read() for re_src, re_repl in sorted(repl_map.items()): text = re.sub(re_src, re_repl, text) for re_src, re_repl in sorted(extra_map.items()): text = re.sub(re_src, re_repl, text) text = text.rstrip() + '\n' with open(dest, 'w') as dest_file: dest_file.write(text) def update_sources(glibc_srcdir, quadmath_srcdir): """Update libquadmath sources.""" glibc_ldbl128 = os.path.join(glibc_srcdir, 'sysdeps/ieee754/ldbl-128') glibc_math = os.path.join(glibc_srcdir, 'math') quadmath_math = os.path.join(quadmath_srcdir, 'math') float128_h = os.path.join(glibc_srcdir, 'sysdeps/ieee754/float128/float128_private.h') repl_map = {} # Use float128_private.h to get an initial list of names to # replace for libquadmath. repl_names = {} with open(float128_h, 'r') as header: for line in header: line = line.strip() if not line.startswith('#define '): continue match = re.fullmatch('^#define[ \t]+([a-zA-Z0-9_]+)' '[ \t]+([a-zA-Z0-9_]+)', line) if not match: continue macro = match.group(1) result = match.group(2) result = result.replace('f128', 'q') result = result.replace('__ieee754_', '') if result not in ('__expq_table', '__sincosq_table', '__builtin_signbit'): result = result.replace('__', '') result = result.replace('_do_not_use', '') if result in ('rem_pio2q', 'kernel_sincosq', 'kernel_sinq', 'kernel_cosq', 'kernel_tanq', 'gammaq_r', 'gamma_productq', 'lgamma_negq', 'lgamma_productq', 'lgammaq_r', 'x2y2m1q'): # Internal function names, for which the above removal # of leading '__' was inappropriate and a leading # '__quadmath_' needs adding instead. In the # libquadmath context, lgammaq_r is an internal name. result = '__quadmath_' + result if result == 'ieee854_float128_shape_type': result = 'ieee854_float128' if result == 'HUGE_VAL_F128': result = 'HUGE_VALQ' repl_names[macro] = result # More such names that aren't simply defined as object-like macros # in float128_private.h. repl_names['_Float128'] = '__float128' repl_names['SET_RESTORE_ROUNDL'] = 'SET_RESTORE_ROUNDF128' repl_names['parts32'] = 'words32' for macro in ('GET_LDOUBLE_LSW64', 'GET_LDOUBLE_MSW64', 'GET_LDOUBLE_WORDS64', 'SET_LDOUBLE_LSW64', 'SET_LDOUBLE_MSW64', 'SET_LDOUBLE_WORDS64'): repl_names[macro] = macro.replace('LDOUBLE', 'FLT128') # The classication macros are replaced. for macro in ('FP_NAN', 'FP_INFINITE', 'FP_ZERO', 'FP_SUBNORMAL', 'FP_NORMAL'): repl_names[macro] = 'QUAD' + macro for macro in ('fpclassify', 'signbit', 'isnan', 'isinf', 'issignaling'): repl_names[macro] = macro + 'q' repl_names['isfinite'] = 'finiteq' # Map comparison macros to the __builtin forms. for macro in ('isgreater', 'isgreaterequal', 'isless', 'islessequal', 'islessgreater', 'isunordered'): repl_names[macro] = '__builtin_' + macro # Replace macros used in type-generic templates in glibc. repl_names['FLOAT'] = '__float128' repl_names['CFLOAT'] = '__complex128' repl_names['M_NAN'] = 'nanq ("")' repl_names['M_HUGE_VAL'] = 'HUGE_VALQ' repl_names['INFINITY'] = '__builtin_inf ()' for macro in ('MIN_EXP', 'MAX_EXP', 'MIN', 'MAX', 'MANT_DIG', 'EPSILON'): repl_names['M_%s' % macro] = 'FLT128_%s' % macro for macro in ('COPYSIGN', 'FABS', 'SINCOS', 'SCALBN', 'LOG1P', 'ATAN2', 'COSH', 'EXP', 'HYPOT', 'LOG', 'SINH', 'SQRT'): repl_names['M_%s' % macro] = macro.lower() + 'q' # Each such name is replaced when it appears as a whole word. for macro in repl_names: repl_map[r'\b%s\b' % macro] = repl_names[macro] # Also replace the L macro for constants; likewise M_LIT and M_MLIT. repl_map[r'\bL *\((.*?)\)'] = r'\1Q' repl_map[r'\bM_LIT *\((.*?)\)'] = r'\1Q' repl_map[r'\bM_MLIT *\((.*?)\)'] = r'\1q' # M_DECL_FUNC and M_SUF need similar replacements. repl_map[r'\bM_DECL_FUNC *\((?:__)?(?:ieee754_)?(.*?)\)'] = r'\1q' repl_map[r'\bM_SUF *\((?:__)?(?:ieee754_)?(.*?)\)'] = r'\1q' # Further adjustments are then needed for certain internal # functions called via M_SUF. repl_map[r'\bx2y2m1q\b'] = '__quadmath_x2y2m1q' repl_map[r'\bkernel_casinhq\b'] = '__quadmath_kernel_casinhq' # Replace calls to __set_errno. repl_map[r'\b__set_errno *\((.*?)\)'] = r'errno = \1' # Eliminate glibc diagnostic macros. repl_map[r' *\bDIAG_PUSH_NEEDS_COMMENT;'] = '' repl_map[r' *\bDIAG_IGNORE_NEEDS_COMMENT *\(.*?\);'] = '' repl_map[r' *\bDIAG_POP_NEEDS_COMMENT;'] = '' # Different names used in union. repl_map[r'\.d\b'] = '.value' repl_map[r'\bunion ieee854_float128\b'] = 'ieee854_float128' # Calls to alias and hidden_def macros are all eliminated. for macro in ('strong_alias', 'weak_alias', 'libm_alias_ldouble', 'declare_mgen_alias', 'declare_mgen_finite_alias', 'libm_hidden_def', 'mathx_hidden_def'): repl_map[r'\b%s *\(.*?\);?' % macro] = '' # Replace all #includes with a single include of quadmath-imp.h. repl_map['(\n+#include[^\n]*)+\n+'] = '\n\n#include "quadmath-imp.h"\n\n' # Omitted from this list because code comes from more than one # glibc source file: rem_pio2. ldbl_files = { 'e_acoshl.c': 'acoshq.c', 'e_acosl.c': 'acosq.c', 's_asinhl.c': 'asinhq.c', 'e_asinl.c': 'asinq.c', 'e_atan2l.c': 'atan2q.c', 'e_atanhl.c': 'atanhq.c', 's_atanl.c': 'atanq.c', 's_cbrtl.c': 'cbrtq.c', 's_ceill.c': 'ceilq.c', 's_copysignl.c': 'copysignq.c', 'e_coshl.c': 'coshq.c', 's_cosl.c': 'cosq.c', 'k_cosl.c': 'cosq_kernel.c', 's_erfl.c': 'erfq.c', 's_expm1l.c': 'expm1q.c', 'e_expl.c': 'expq.c', 't_expl.h': 'expq_table.h', 's_fabsl.c': 'fabsq.c', 's_finitel.c': 'finiteq.c', 's_floorl.c': 'floorq.c', 's_fmal.c': 'fmaq.c', 'e_fmodl.c': 'fmodq.c', 's_frexpl.c': 'frexpq.c', 'e_lgammal_r.c': 'lgammaq.c', 'lgamma_negl.c': 'lgammaq_neg.c', 'lgamma_productl.c': 'lgammaq_product.c', 'e_hypotl.c': 'hypotq.c', 'e_ilogbl.c': 'ilogbq.c', 's_isinfl.c': 'isinfq.c', 's_isnanl.c': 'isnanq.c', 's_issignalingl.c': 'issignalingq.c', 'e_j0l.c': 'j0q.c', 'e_j1l.c': 'j1q.c', 'e_jnl.c': 'jnq.c', 's_llrintl.c': 'llrintq.c', 's_llroundl.c': 'llroundq.c', 'e_log10l.c': 'log10q.c', 's_log1pl.c': 'log1pq.c', 'e_log2l.c': 'log2q.c', 's_logbl.c': 'logbq.c', 'e_logl.c': 'logq.c', 's_lrintl.c': 'lrintq.c', 's_lroundl.c': 'lroundq.c', 's_modfl.c': 'modfq.c', 's_nearbyintl.c': 'nearbyintq.c', 's_nextafterl.c': 'nextafterq.c', 'e_powl.c': 'powq.c', 'e_remainderl.c': 'remainderq.c', 's_remquol.c': 'remquoq.c', 's_rintl.c': 'rintq.c', 's_roundl.c': 'roundq.c', 's_scalblnl.c': 'scalblnq.c', 's_scalbnl.c': 'scalbnq.c', 's_signbitl.c': 'signbitq.c', 't_sincosl.c': 'sincos_table.c', 's_sincosl.c': 'sincosq.c', 'k_sincosl.c': 'sincosq_kernel.c', 'e_sinhl.c': 'sinhq.c', 's_sinl.c': 'sinq.c', 'k_sinl.c': 'sinq_kernel.c', 's_tanhl.c': 'tanhq.c', 's_tanl.c': 'tanq.c', 'k_tanl.c': 'tanq_kernel.c', 'e_gammal_r.c': 'tgammaq.c', 'gamma_productl.c': 'tgammaq_product.c', 's_truncl.c': 'truncq.c', 'x2y2m1l.c': 'x2y2m1q.c' } template_files = { 's_cacosh_template.c': 'cacoshq.c', 's_cacos_template.c': 'cacosq.c', 's_casinh_template.c': 'casinhq.c', 'k_casinh_template.c': 'casinhq_kernel.c', 's_casin_template.c': 'casinq.c', 's_catanh_template.c': 'catanhq.c', 's_catan_template.c': 'catanq.c', 's_ccosh_template.c': 'ccoshq.c', 's_cexp_template.c': 'cexpq.c', 'cimag_template.c': 'cimagq.c', 's_clog10_template.c': 'clog10q.c', 's_clog_template.c': 'clogq.c', 'conj_template.c': 'conjq.c', 's_cproj_template.c': 'cprojq.c', 'creal_template.c': 'crealq.c', 's_csinh_template.c': 'csinhq.c', 's_csin_template.c': 'csinq.c', 's_csqrt_template.c': 'csqrtq.c', 's_ctanh_template.c': 'ctanhq.c', 's_ctan_template.c': 'ctanq.c', 'e_exp2_template.c': 'exp2q.c', 's_fdim_template.c': 'fdimq.c', 's_fmax_template.c': 'fmaxq.c', 's_fmin_template.c': 'fminq.c', 's_ldexp_template.c': 'ldexpq.c' } # Some files have extra substitutions to apply. extra_maps = defaultdict(dict) extra_maps['expq.c'] = {r'#include "quadmath-imp\.h"\n': '#include "quadmath-imp.h"\n' '#include "expq_table.h"\n'} extra_maps['ilogbq.c'] = {r'#include "quadmath-imp\.h"\n': '#include \n' '#include "quadmath-imp.h"\n' '#ifndef FP_ILOGB0\n' '# define FP_ILOGB0 INT_MIN\n' '#endif\n' '#ifndef FP_ILOGBNAN\n' '# define FP_ILOGBNAN INT_MAX\n' '#endif\n', r'return ([A-Z0-9_]+);': r'{ errno = EDOM; feraiseexcept (FE_INVALID); ' r'return \1; }'} extra_maps['lgammaq.c'] = {r'#include "quadmath-imp\.h"\n': '#include "quadmath-imp.h"\n' '#ifdef HAVE_MATH_H_SIGNGAM\n' '# include \n' '#endif\n' '__float128\n' 'lgammaq (__float128 x)\n' '{\n' '#ifndef HAVE_MATH_H_SIGNGAM\n' ' int signgam;\n' '#endif\n' ' return __quadmath_lgammaq_r (x, &signgam);\n' '}\n'} extra_maps['tgammaq.c'] = {r'#include "quadmath-imp\.h"\n': '#include "quadmath-imp.h"\n' '__float128\n' 'tgammaq (__float128 x)\n' '{\n' ' int sign;\n' ' __float128 ret;\n' ' ret = __quadmath_gammaq_r (x, &sign);\n' ' return sign < 0 ? -ret : ret;\n' '}\n'} for src, dest in ldbl_files.items(): replace_in_file(repl_map, extra_maps[dest], os.path.join(glibc_ldbl128, src), os.path.join(quadmath_math, dest)) for src, dest in template_files.items(): replace_in_file(repl_map, extra_maps[dest], os.path.join(glibc_math, src), os.path.join(quadmath_math, dest)) def main(): parser = argparse.ArgumentParser(description='Update libquadmath code.') parser.add_argument('glibc_srcdir', help='glibc source directory') parser.add_argument('quadmath_srcdir', help='libquadmath source directory') args = parser.parse_args() update_sources(args.glibc_srcdir, args.quadmath_srcdir) if __name__ == '__main__': main()