From patchwork Fri Sep 30 00:28:33 2016 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Joseph Myers X-Patchwork-Id: 16154 Received: (qmail 85816 invoked by alias); 30 Sep 2016 00:28:51 -0000 Mailing-List: contact libc-alpha-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Unsubscribe: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: libc-alpha-owner@sourceware.org Delivered-To: mailing list libc-alpha@sourceware.org Received: (qmail 85806 invoked by uid 89); 30 Sep 2016 00:28:50 -0000 Authentication-Results: sourceware.org; auth=none X-Virus-Found: No X-Spam-SWARE-Status: No, score=-1.9 required=5.0 tests=AWL, BAYES_00, RCVD_IN_DNSWL_NONE, SPF_PASS, URIBL_RED autolearn=ham version=3.3.2 spammy=20160930, 2016-09-30, greatest, Never X-HELO: relay1.mentorg.com Date: Fri, 30 Sep 2016 00:28:33 +0000 From: Joseph Myers To: Subject: Add iscanonical [committed] Message-ID: User-Agent: Alpine 2.20 (DEB 67 2015-01-07) MIME-Version: 1.0 X-ClientProxiedBy: svr-ies-mbx-01.mgc.mentorg.com (139.181.222.1) To svr-ies-mbx-01.mgc.mentorg.com (139.181.222.1) TS 18661-1 adds an iscanonical classification macro to . The motivation for this is decimal floating-point, where some values have both canonical and noncanonical encodings. For IEEE binary interchange formats, all encodings are canonical. For x86/m68k ldbl-96, and for ldbl-128ibm, there are encodings that do not represent any valid value of the type; although formally iscanonical does not need to handle trap representations (and so could just always return 1), it seems useful, and in line with the description in the TS of "representations that are extraneous to the floating-point model" as being non-canonical (as well as "redundant representations of some or all of its values"), for it to detect those representations and return 0 for them. This patch adds iscanonical to glibc. It goes in a header , included under appropriate conditions in . The default header version just evaluates the argument (converted to its semantic type, though current GCC will probably discard that conversion and any exceptions resulting from it) and returns 1. ldbl-96 and ldbl-128ibm then have versions of the header that call a function __iscanonicall for long double (the sizeof-based tests will of course need updating for float128 support, like other such type-generic macro implementations). The ldbl-96 version of __iscanonicall has appropriate conditionals to reflect the differences in the m68k version of that format (where the high mantissa bit may be either 0 or 1 when the exponent is 0 or 0x7fff). Corresponding tests for those formats are added as well. Other architectures do not have any new functions added because just returning 1 is correct for all their floating-point formats. Tested for x86_64, x86, mips64 (to test the default macro version) and powerpc. Committed. 2016-09-30 Joseph Myers * math/math.h [__GLIBC_USE (IEC_60559_BFP_EXT)]: Include . * bits/iscanonical.h: New file. * math/s_iscanonicall.c: Likewise. * math/Versions (__iscanonicall): New libm symbol at version GLIBC_2.25. * math/libm-test.inc (iscanonical_test_data): New array. (iscanonical_test): New function. (main): Call iscanonical_test. * math/Makefile (headers): Add bits/iscanonical.h. (type-ldouble-routines): Add s_iscanonicall. * manual/arith.texi (Floating Point Classes): Document iscanonical. * manual/libm-err-tab.pl: Update comment on interfaces without ulps tabulated. * sysdeps/ieee754/ldbl-128ibm/bits/iscanonical.h: New file. * sysdeps/ieee754/ldbl-128ibm/s_iscanonicall.c: Likewise. * sysdeps/ieee754/ldbl-128ibm/test-iscanonical-ldbl-128ibm.c: Likewise. * sysdeps/ieee754/ldbl-128ibm/Makefile (tests): Add test-iscanonical-ldbl-128ibm. * sysdeps/ieee754/ldbl-96/bits/iscanonical.h: New file. * sysdeps/ieee754/ldbl-96/s_iscanonicall.c: Likewise. * sysdeps/ieee754/ldbl-96/test-iscanonical-ldbl-96.c: Likewise. * sysdeps/ieee754/ldbl-96/Makefile: Likewise. * sysdeps/unix/sysv/linux/i386/libm.abilist: Update. * sysdeps/unix/sysv/linux/ia64/libm.abilist: Likewise. * sysdeps/unix/sysv/linux/m68k/m680x0/libm.abilist: Likewise. * sysdeps/unix/sysv/linux/powerpc/powerpc32/fpu/libm.abilist: Likewise. * sysdeps/unix/sysv/linux/powerpc/powerpc32/nofpu/libm.abilist: Likewise. * sysdeps/unix/sysv/linux/powerpc/powerpc64/libm-le.abilist: Likewise. * sysdeps/unix/sysv/linux/powerpc/powerpc64/libm.abilist: Likewise. * sysdeps/unix/sysv/linux/x86_64/64/libm.abilist: Likewise. * sysdeps/unix/sysv/linux/x86_64/x32/libm.abilist: Likewise. diff --git a/NEWS b/NEWS index e295274..0b2ca04 100644 --- a/NEWS +++ b/NEWS @@ -51,7 +51,7 @@ Version 2.25 * New features are added from TS 18661-1:2014: - - Classification macros: issubnormal, iszero. + - Classification macros: iscanonical, issubnormal, iszero. * The header now includes the header. Support for the Linux quota interface which predates kernel version 2.4.22 has diff --git a/bits/iscanonical.h b/bits/iscanonical.h new file mode 100644 index 0000000..97eb736 --- /dev/null +++ b/bits/iscanonical.h @@ -0,0 +1,28 @@ +/* Define iscanonical macro. + Copyright (C) 2016 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 + . */ + +#ifndef _MATH_H +# error "Never use directly; include instead." +#endif + +/* Return nonzero value if X is canonical. By default, we only have + IEEE interchange binary formats, in which all values are canonical, + but the argument must still be converted to its semantic type for + any exceptions arising from the conversion, before being + discarded. */ +#define iscanonical(x) ((void) (__typeof (x)) (x), 1) diff --git a/manual/arith.texi b/manual/arith.texi index 58f3578..a13c46f 100644 --- a/manual/arith.texi +++ b/manual/arith.texi @@ -361,6 +361,23 @@ You should therefore use the specific macros whenever possible. @comment math.h @comment ISO +@deftypefn {Macro} int iscanonical (@emph{float-type} @var{x}) +@safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} +In some floating-point formats, some values have canonical (preferred) +and noncanonical encodings (for IEEE interchange binary formats, all +encodings are canonical). This macro returns a nonzero value if +@var{x} has a canonical encoding. It is from TS 18661-1:2014. + +Note that some formats have multiple encodings of a value which are +all equally canonical; @code{iscanonical} returns a nonzero value for +all such encodings. Also, formats may have encodings that do not +correspond to any valid value of the type. In ISO C terms these are +@dfn{trap representations}; in @theglibc{}, @code{iscanonical} returns +zero for such encodings. +@end deftypefn + +@comment math.h +@comment ISO @deftypefn {Macro} int isfinite (@emph{float-type} @var{x}) @safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}} This macro returns a nonzero value if @var{x} is finite: not plus or diff --git a/manual/libm-err-tab.pl b/manual/libm-err-tab.pl index 2386772..b36ef94 100755 --- a/manual/libm-err-tab.pl +++ b/manual/libm-err-tab.pl @@ -77,7 +77,7 @@ use vars qw (%results @all_floats %suffices @all_functions); "nextup", "pow", "remainder", "remquo", "rint", "round", "scalb", "scalbn", "sin", "sincos", "sinh", "sqrt", "tan", "tanh", "tgamma", "trunc", "y0", "y1", "yn" ); -# fpclassify, isnormal, isfinite, isinf, isnan, issignaling, +# fpclassify, iscanonical, isnormal, isfinite, isinf, isnan, issignaling, # issubnormal, iszero, signbit, isgreater, isgreaterequal, isless, # islessequal, islessgreater, isunordered are not tabulated. diff --git a/math/Makefile b/math/Makefile index 7ccd59a..d2b4fd1 100644 --- a/math/Makefile +++ b/math/Makefile @@ -27,7 +27,7 @@ headers := math.h bits/mathcalls.h bits/mathinline.h bits/huge_val.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-vector.h \ - bits/libm-simd-decl-stubs.h + bits/libm-simd-decl-stubs.h bits/iscanonical.h # FPU support code. aux := setfpucw fpu_control @@ -94,7 +94,7 @@ types = $(type-ldouble-$(long-double-fcts)) double float # long double support type-ldouble-suffix := l -type-ldouble-routines := t_sincosl k_sincosl +type-ldouble-routines := t_sincosl k_sincosl s_iscanonicall type-ldouble-yes := ldouble # double support diff --git a/math/Versions b/math/Versions index a429221..f702051 100644 --- a/math/Versions +++ b/math/Versions @@ -216,5 +216,6 @@ libm { } GLIBC_2.25 { fesetexcept; fetestexceptflag; fegetmode; fesetmode; + __iscanonicall; } } diff --git a/math/libm-test.inc b/math/libm-test.inc index 872bafd..cbc7226 100644 --- a/math/libm-test.inc +++ b/math/libm-test.inc @@ -46,9 +46,9 @@ cbrt, ceil, copysign, cos, cosh, drem, erf, erfc, exp, exp10, exp2, expm1, fabs, fdim, finite, floor, fma, fmax, fmin, fmod, fpclassify, frexp, gamma, hypot, - ilogb, isfinite, isinf, isnan, isnormal, issignaling, issubnormal, iszero, - isless, islessequal, isgreater, isgreaterequal, islessgreater, isunordered, - j0, j1, jn, + ilogb, iscanonical, isfinite, isinf, isnan, isnormal, issignaling, + issubnormal, iszero, isless, islessequal, isgreater, + isgreaterequal, islessgreater, isunordered, j0, j1, jn, ldexp, lgamma, log, log10, log1p, log2, logb, modf, nearbyint, nextafter, nexttoward, pow, pow10, remainder, remquo, rint, lrint, llrint, @@ -8160,6 +8160,31 @@ ilogb_test (void) ALL_RM_TEST (ilogb, 1, ilogb_test_data, RUN_TEST_LOOP_f_i, END); } +static const struct test_f_i_data iscanonical_test_data[] = + { + TEST_f_b (iscanonical, 0, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), + TEST_f_b (iscanonical, minus_zero, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), + TEST_f_b (iscanonical, 10, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), + TEST_f_b (iscanonical, min_subnorm_value, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), + TEST_f_b (iscanonical, -min_subnorm_value, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), + TEST_f_b (iscanonical, min_value, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), + TEST_f_b (iscanonical, -min_value, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), + TEST_f_b (iscanonical, max_value, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), + TEST_f_b (iscanonical, -max_value, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), + TEST_f_b (iscanonical, plus_infty, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), + TEST_f_b (iscanonical, minus_infty, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), + TEST_f_b (iscanonical, qnan_value, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), + TEST_f_b (iscanonical, -qnan_value, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), + TEST_f_b (iscanonical, snan_value, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), + TEST_f_b (iscanonical, -snan_value, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), + }; + +static void +iscanonical_test (void) +{ + ALL_RM_TEST (iscanonical, 1, iscanonical_test_data, RUN_TEST_LOOP_f_b_tg, END); +} + static const struct test_f_i_data isfinite_test_data[] = { TEST_f_b (isfinite, 0, 1, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED), @@ -12713,6 +12738,7 @@ main (int argc, char **argv) /* Classification macros: */ finite_test (); fpclassify_test (); + iscanonical_test (); isfinite_test (); isinf_test (); isnan_test (); diff --git a/math/math.h b/math/math.h index 1382baa..8cd6416 100644 --- a/math/math.h +++ b/math/math.h @@ -317,6 +317,8 @@ enum #endif /* Use ISO C99. */ #if __GLIBC_USE (IEC_60559_BFP_EXT) +# include + /* Return nonzero value if X is a signaling NaN. */ # ifdef __NO_LONG_DOUBLE_MATH # define issignaling(x) \ diff --git a/math/s_iscanonicall.c b/math/s_iscanonicall.c new file mode 100644 index 0000000..b5fd996 --- /dev/null +++ b/math/s_iscanonicall.c @@ -0,0 +1 @@ +/* Not needed by default. */ diff --git a/sysdeps/ieee754/ldbl-128ibm/Makefile b/sysdeps/ieee754/ldbl-128ibm/Makefile index 6242edd..56bebf8 100644 --- a/sysdeps/ieee754/ldbl-128ibm/Makefile +++ b/sysdeps/ieee754/ldbl-128ibm/Makefile @@ -11,5 +11,5 @@ endif ifeq ($(subdir),math) tests += test-fmodl-ldbl-128ibm test-remainderl-ldbl-128ibm \ - test-remquol-ldbl-128ibm + test-remquol-ldbl-128ibm test-iscanonical-ldbl-128ibm endif diff --git a/sysdeps/ieee754/ldbl-128ibm/bits/iscanonical.h b/sysdeps/ieee754/ldbl-128ibm/bits/iscanonical.h new file mode 100644 index 0000000..cbc79ae --- /dev/null +++ b/sysdeps/ieee754/ldbl-128ibm/bits/iscanonical.h @@ -0,0 +1,35 @@ +/* Define iscanonical macro. ldbl-128ibm version. + Copyright (C) 2016 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 + . */ + +#ifndef _MATH_H +# error "Never use directly; include instead." +#endif + +extern int __iscanonicall (long double __x) + __THROW __attribute__ ((__const__)); + +/* Return nonzero value if X is canonical. In IEEE interchange binary + formats, all values are canonical, but the argument must still be + converted to its semantic type for any exceptions arising from the + conversion, before being discarded; in IBM long double, there are + encodings that are not consistently handled as corresponding to any + particular value of the type, and we return 0 for those. */ +#define iscanonical(x) \ + (sizeof (x) == sizeof (long double) \ + ? __iscanonicall (x) \ + : ((void) (__typeof (x)) (x), 1)) diff --git a/sysdeps/ieee754/ldbl-128ibm/s_iscanonicall.c b/sysdeps/ieee754/ldbl-128ibm/s_iscanonicall.c new file mode 100644 index 0000000..100b401 --- /dev/null +++ b/sysdeps/ieee754/ldbl-128ibm/s_iscanonicall.c @@ -0,0 +1,59 @@ +/* Test whether long double value is canonical. ldbl-128ibm version. + Copyright (C) 2016 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 + . */ + +#include +#include +#include + +int +__iscanonicall (long double x) +{ + double xhi, xlo; + uint64_t hx, lx; + + ldbl_unpack (x, &xhi, &xlo); + EXTRACT_WORDS64 (hx, xhi); + EXTRACT_WORDS64 (lx, xlo); + int64_t ix = hx & 0x7fffffffffffffffULL; + int64_t iy = lx & 0x7fffffffffffffffULL; + int hexp = (ix & 0x7ff0000000000000LL) >> 52; + int lexp = (iy & 0x7ff0000000000000LL) >> 52; + + if (iy == 0) + /* Low part 0 is always OK. */ + return 1; + + if (hexp == 0x7ff) + /* If a NaN, the low part does not matter. If an infinity, the + low part must be 0, in which case we have already returned. */ + return ix != 0x7ff0000000000000LL; + + /* The high part is finite and the low part is nonzero. There must + be sufficient difference between the exponents. */ + bool low_p2; + if (lexp == 0) + { + /* Adjust the exponent for subnormal low part. */ + lexp = 12 - __builtin_clzll (iy); + low_p2 = iy == (1LL << (51 + lexp)); + } + else + low_p2 = (iy & 0xfffffffffffffLL) == 0; + int expdiff = hexp - lexp; + return expdiff > 53 || (expdiff == 53 && low_p2 && (ix & 1) == 0); +} diff --git a/sysdeps/ieee754/ldbl-128ibm/test-iscanonical-ldbl-128ibm.c b/sysdeps/ieee754/ldbl-128ibm/test-iscanonical-ldbl-128ibm.c new file mode 100644 index 0000000..92b9c69 --- /dev/null +++ b/sysdeps/ieee754/ldbl-128ibm/test-iscanonical-ldbl-128ibm.c @@ -0,0 +1,203 @@ +/* Test iscanonical for ldbl-128ibm. + Copyright (C) 2016 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 + . */ + +#include +#include +#include +#include +#include + +struct test +{ + double hi, lo; + bool canonical; +}; + +static const struct test tests[] = + { + { __builtin_nan (""), 0.0, true }, + { __builtin_nan (""), DBL_MAX, true }, + { __builtin_nan (""), __builtin_inf (), true }, + { __builtin_nan (""), __builtin_nan (""), true }, + { __builtin_nan (""), __builtin_nans (""), true }, + { __builtin_nans (""), 0.0, true }, + { __builtin_nans (""), DBL_MAX, true }, + { __builtin_nans (""), __builtin_inf (), true }, + { __builtin_nans (""), __builtin_nan (""), true }, + { __builtin_nans (""), __builtin_nans (""), true }, + { __builtin_inf (), 0.0, true }, + { __builtin_inf (), -0.0, true }, + { -__builtin_inf (), 0.0, true }, + { -__builtin_inf (), -0.0, true }, + { __builtin_inf (), DBL_TRUE_MIN, false }, + { __builtin_inf (), -DBL_TRUE_MIN, false }, + { -__builtin_inf (), DBL_TRUE_MIN, false }, + { -__builtin_inf (), -DBL_TRUE_MIN, false }, + { __builtin_inf (), DBL_MIN, false }, + { __builtin_inf (), -DBL_MIN, false }, + { -__builtin_inf (), DBL_MIN, false }, + { -__builtin_inf (), -DBL_MIN, false }, + { __builtin_inf (), __builtin_inf (), false }, + { __builtin_inf (), -__builtin_inf (), false }, + { -__builtin_inf (), __builtin_inf (), false }, + { -__builtin_inf (), -__builtin_inf (), false }, + { __builtin_inf (), __builtin_nan (""), false }, + { __builtin_inf (), -__builtin_nan (""), false }, + { -__builtin_inf (), __builtin_nan (""), false }, + { -__builtin_inf (), -__builtin_nan (""), false }, + { 0.0, 0.0, true }, + { 0.0, -0.0, true }, + { -0.0, 0.0, true }, + { -0.0, -0.0, true }, + { 0.0, DBL_TRUE_MIN, false }, + { 0.0, -DBL_TRUE_MIN, false }, + { -0.0, DBL_TRUE_MIN, false }, + { -0.0, -DBL_TRUE_MIN, false }, + { 0.0, DBL_MAX, false }, + { 0.0, -DBL_MAX, false }, + { -0.0, DBL_MAX, false }, + { -0.0, -DBL_MAX, false }, + { 0.0, __builtin_inf (), false }, + { 0.0, -__builtin_inf (), false }, + { -0.0, __builtin_inf (), false }, + { -0.0, -__builtin_inf (), false }, + { 0.0, __builtin_nan (""), false }, + { 0.0, -__builtin_nan (""), false }, + { -0.0, __builtin_nan (""), false }, + { -0.0, -__builtin_nan (""), false }, + { 1.0, 0.0, true }, + { 1.0, -0.0, true }, + { -1.0, 0.0, true }, + { -1.0, -0.0, true }, + { 1.0, DBL_TRUE_MIN, true }, + { 1.0, -DBL_TRUE_MIN, true }, + { -1.0, DBL_TRUE_MIN, true }, + { -1.0, -DBL_TRUE_MIN, true }, + { 1.0, DBL_MAX, false }, + { 1.0, -DBL_MAX, false }, + { -1.0, DBL_MAX, false }, + { -1.0, -DBL_MAX, false }, + { 1.0, __builtin_inf (), false }, + { 1.0, -__builtin_inf (), false }, + { -1.0, __builtin_inf (), false }, + { -1.0, -__builtin_inf (), false }, + { 1.0, __builtin_nan (""), false }, + { 1.0, -__builtin_nan (""), false }, + { -1.0, __builtin_nan (""), false }, + { -1.0, -__builtin_nan (""), false }, + { 0x1p1023, 0x1.1p969, true }, + { 0x1p1023, -0x1.1p969, true }, + { -0x1p1023, 0x1.1p969, true }, + { -0x1p1023, -0x1.1p969, true }, + { 0x1p1023, 0x1.1p970, false }, + { 0x1p1023, -0x1.1p970, false }, + { -0x1p1023, 0x1.1p970, false }, + { -0x1p1023, -0x1.1p970, false }, + { 0x1p1023, 0x1p970, true }, + { 0x1p1023, -0x1p970, true }, + { -0x1p1023, 0x1p970, true }, + { -0x1p1023, -0x1p970, true }, + { 0x1.0000000000001p1023, 0x1p970, false }, + { 0x1.0000000000001p1023, -0x1p970, false }, + { -0x1.0000000000001p1023, 0x1p970, false }, + { -0x1.0000000000001p1023, -0x1p970, false }, + { 0x1p-969, 0x1.1p-1023, true }, + { 0x1p-969, -0x1.1p-1023, true }, + { -0x1p-969, 0x1.1p-1023, true }, + { -0x1p-969, -0x1.1p-1023, true }, + { 0x1p-969, 0x1.1p-1022, false }, + { 0x1p-969, -0x1.1p-1022, false }, + { -0x1p-969, 0x1.1p-1022, false }, + { -0x1p-969, -0x1.1p-1022, false }, + { 0x1p-969, 0x1p-1022, true }, + { 0x1p-969, -0x1p-1022, true }, + { -0x1p-969, 0x1p-1022, true }, + { -0x1p-969, -0x1p-1022, true }, + { 0x1.0000000000001p-969, 0x1p-1022, false }, + { 0x1.0000000000001p-969, -0x1p-1022, false }, + { -0x1.0000000000001p-969, 0x1p-1022, false }, + { -0x1.0000000000001p-969, -0x1p-1022, false }, + { 0x1p-970, 0x1.1p-1024, true }, + { 0x1p-970, -0x1.1p-1024, true }, + { -0x1p-970, 0x1.1p-1024, true }, + { -0x1p-970, -0x1.1p-1024, true }, + { 0x1p-970, 0x1.1p-1023, false }, + { 0x1p-970, -0x1.1p-1023, false }, + { -0x1p-970, 0x1.1p-1023, false }, + { -0x1p-970, -0x1.1p-1023, false }, + { 0x1p-970, 0x1p-1023, true }, + { 0x1p-970, -0x1p-1023, true }, + { -0x1p-970, 0x1p-1023, true }, + { -0x1p-970, -0x1p-1023, true }, + { 0x1.0000000000001p-970, 0x1p-1023, false }, + { 0x1.0000000000001p-970, -0x1p-1023, false }, + { -0x1.0000000000001p-970, 0x1p-1023, false }, + { -0x1.0000000000001p-970, -0x1p-1023, false }, + { 0x1p-1000, 0x1.1p-1054, true }, + { 0x1p-1000, -0x1.1p-1054, true }, + { -0x1p-1000, 0x1.1p-1054, true }, + { -0x1p-1000, -0x1.1p-1054, true }, + { 0x1p-1000, 0x1.1p-1053, false }, + { 0x1p-1000, -0x1.1p-1053, false }, + { -0x1p-1000, 0x1.1p-1053, false }, + { -0x1p-1000, -0x1.1p-1053, false }, + { 0x1p-1000, 0x1p-1053, true }, + { 0x1p-1000, -0x1p-1053, true }, + { -0x1p-1000, 0x1p-1053, true }, + { -0x1p-1000, -0x1p-1053, true }, + { 0x1.0000000000001p-1000, 0x1p-1053, false }, + { 0x1.0000000000001p-1000, -0x1p-1053, false }, + { -0x1.0000000000001p-1000, 0x1p-1053, false }, + { -0x1.0000000000001p-1000, -0x1p-1053, false }, + { 0x1p-1021, 0x1p-1074, true }, + { 0x1p-1021, -0x1p-1074, true }, + { -0x1p-1021, 0x1p-1074, true }, + { -0x1p-1021, -0x1p-1074, true }, + { 0x1.0000000000001p-1021, 0x1p-1074, false }, + { 0x1.0000000000001p-1021, -0x1p-1074, false }, + { -0x1.0000000000001p-1021, 0x1p-1074, false }, + { -0x1.0000000000001p-1021, -0x1p-1074, false }, + { 0x1p-1022, 0x1p-1074, false }, + { 0x1p-1022, -0x1p-1074, false }, + { -0x1p-1022, 0x1p-1074, false }, + { -0x1p-1022, -0x1p-1074, false }, + }; + +static int +do_test (void) +{ + int result = 0; + + for (size_t i = 0; i < sizeof (tests) / sizeof (tests[0]); i++) + { + long double ld = ldbl_pack (tests[i].hi, tests[i].lo); + bool canonical = iscanonical (ld); + if (canonical == tests[i].canonical) + printf ("PASS: test %zu\n", i); + else + { + printf ("FAIL: test %zu\n", i); + result = 1; + } + } + + return result; +} + +#define TEST_FUNCTION do_test () +#include "../test-skeleton.c" diff --git a/sysdeps/ieee754/ldbl-96/Makefile b/sysdeps/ieee754/ldbl-96/Makefile new file mode 100644 index 0000000..bf9676e --- /dev/null +++ b/sysdeps/ieee754/ldbl-96/Makefile @@ -0,0 +1,21 @@ +# Makefile for sysdeps/ieee754/ldbl-96. +# Copyright (C) 2016 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 +# . + +ifeq ($(subdir),math) +tests += test-iscanonical-ldbl-96 +endif diff --git a/sysdeps/ieee754/ldbl-96/bits/iscanonical.h b/sysdeps/ieee754/ldbl-96/bits/iscanonical.h new file mode 100644 index 0000000..af0c72c --- /dev/null +++ b/sysdeps/ieee754/ldbl-96/bits/iscanonical.h @@ -0,0 +1,35 @@ +/* Define iscanonical macro. ldbl-96 version. + Copyright (C) 2016 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 + . */ + +#ifndef _MATH_H +# error "Never use directly; include instead." +#endif + +extern int __iscanonicall (long double __x) + __THROW __attribute__ ((__const__)); + +/* Return nonzero value if X is canonical. In IEEE interchange binary + formats, all values are canonical, but the argument must still be + converted to its semantic type for any exceptions arising from the + conversion, before being discarded; in extended precision, there + are encodings that are not consistently handled as corresponding to + any particular value of the type, and we return 0 for those. */ +#define iscanonical(x) \ + (sizeof (x) == sizeof (long double) \ + ? __iscanonicall (x) \ + : ((void) (__typeof (x)) (x), 1)) diff --git a/sysdeps/ieee754/ldbl-96/s_iscanonicall.c b/sysdeps/ieee754/ldbl-96/s_iscanonicall.c new file mode 100644 index 0000000..f820030 --- /dev/null +++ b/sysdeps/ieee754/ldbl-96/s_iscanonicall.c @@ -0,0 +1,43 @@ +/* Test whether long double value is canonical. ldbl-96 version. + Copyright (C) 2016 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 + . */ + +#include +#include +#include +#include +#include + +int +__iscanonicall (long double x) +{ + uint32_t se, i0, i1 __attribute__ ((unused)); + + GET_LDOUBLE_WORDS (se, i0, i1, x); + int32_t ix = se & 0x7fff; + bool mant_high = (i0 & 0x80000000) != 0; + + if (LDBL_MIN_EXP == -16381) + /* Intel variant: the high mantissa bit should have a value + determined by the exponent. */ + return ix > 0 ? mant_high : !mant_high; + else + /* M68K variant: both values of the high bit are valid for the + greatest and smallest exponents, while other exponents require + the high bit to be set. */ + return ix == 0 || ix == 0x7fff || mant_high; +} diff --git a/sysdeps/ieee754/ldbl-96/test-iscanonical-ldbl-96.c b/sysdeps/ieee754/ldbl-96/test-iscanonical-ldbl-96.c new file mode 100644 index 0000000..6827aa8 --- /dev/null +++ b/sysdeps/ieee754/ldbl-96/test-iscanonical-ldbl-96.c @@ -0,0 +1,114 @@ +/* Test iscanonical for ldbl-96. + Copyright (C) 2016 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 + . */ + +#include +#include +#include +#include +#include +#include + +struct test +{ + bool sign; + uint16_t exponent; + bool high; + uint64_t mantissa; + bool canonical; +}; + +#define M68K_VARIANT (LDBL_MIN_EXP == -16382) + +static const struct test tests[] = + { + { false, 0, true, 0, M68K_VARIANT }, + { true, 0, true, 0, M68K_VARIANT }, + { false, 0, true, 1, M68K_VARIANT }, + { true, 0, true, 1, M68K_VARIANT }, + { false, 0, true, 0x100000000ULL, M68K_VARIANT }, + { true, 0, true, 0x100000000ULL, M68K_VARIANT }, + { false, 0, false, 0, true }, + { true, 0, false, 0, true }, + { false, 0, false, 1, true }, + { true, 0, false, 1, true }, + { false, 0, false, 0x100000000ULL, true }, + { true, 0, false, 0x100000000ULL, true }, + { false, 1, true, 0, true }, + { true, 1, true, 0, true }, + { false, 1, true, 1, true }, + { true, 1, true, 1, true }, + { false, 1, true, 0x100000000ULL, true }, + { true, 1, true, 0x100000000ULL, true }, + { false, 1, false, 0, false }, + { true, 1, false, 0, false }, + { false, 1, false, 1, false }, + { true, 1, false, 1, false }, + { false, 1, false, 0x100000000ULL, false }, + { true, 1, false, 0x100000000ULL, false }, + { false, 0x7ffe, true, 0, true }, + { true, 0x7ffe, true, 0, true }, + { false, 0x7ffe, true, 1, true }, + { true, 0x7ffe, true, 1, true }, + { false, 0x7ffe, true, 0x100000000ULL, true }, + { true, 0x7ffe, true, 0x100000000ULL, true }, + { false, 0x7ffe, false, 0, false }, + { true, 0x7ffe, false, 0, false }, + { false, 0x7ffe, false, 1, false }, + { true, 0x7ffe, false, 1, false }, + { false, 0x7ffe, false, 0x100000000ULL, false }, + { true, 0x7ffe, false, 0x100000000ULL, false }, + { false, 0x7fff, true, 0, true }, + { true, 0x7fff, true, 0, true }, + { false, 0x7fff, true, 1, true }, + { true, 0x7fff, true, 1, true }, + { false, 0x7fff, true, 0x100000000ULL, true }, + { true, 0x7fff, true, 0x100000000ULL, true }, + { false, 0x7fff, false, 0, M68K_VARIANT }, + { true, 0x7fff, false, 0, M68K_VARIANT }, + { false, 0x7fff, false, 1, M68K_VARIANT }, + { true, 0x7fff, false, 1, M68K_VARIANT }, + { false, 0x7fff, false, 0x100000000ULL, M68K_VARIANT }, + { true, 0x7fff, false, 0x100000000ULL, M68K_VARIANT }, + }; + +static int +do_test (void) +{ + int result = 0; + + for (size_t i = 0; i < sizeof (tests) / sizeof (tests[0]); i++) + { + long double ld; + SET_LDOUBLE_WORDS (ld, tests[i].exponent | (tests[i].sign << 15), + (tests[i].mantissa >> 32) | (tests[i].high << 31), + tests[i].mantissa & 0xffffffffULL); + bool canonical = iscanonical (ld); + if (canonical == tests[i].canonical) + printf ("PASS: test %zu\n", i); + else + { + printf ("FAIL: test %zu\n", i); + result = 1; + } + } + + return result; +} + +#define TEST_FUNCTION do_test () +#include "../test-skeleton.c" diff --git a/sysdeps/unix/sysv/linux/i386/libm.abilist b/sysdeps/unix/sysv/linux/i386/libm.abilist index 63b8da9..19e9792 100644 --- a/sysdeps/unix/sysv/linux/i386/libm.abilist +++ b/sysdeps/unix/sysv/linux/i386/libm.abilist @@ -423,6 +423,7 @@ GLIBC_2.24 nextup F GLIBC_2.24 nextupf F GLIBC_2.24 nextupl F GLIBC_2.25 GLIBC_2.25 A +GLIBC_2.25 __iscanonicall F GLIBC_2.25 fegetmode F GLIBC_2.25 fesetexcept F GLIBC_2.25 fesetmode F diff --git a/sysdeps/unix/sysv/linux/ia64/libm.abilist b/sysdeps/unix/sysv/linux/ia64/libm.abilist index 611a84f..1a32e2c 100644 --- a/sysdeps/unix/sysv/linux/ia64/libm.abilist +++ b/sysdeps/unix/sysv/linux/ia64/libm.abilist @@ -352,6 +352,7 @@ GLIBC_2.24 nextup F GLIBC_2.24 nextupf F GLIBC_2.24 nextupl F GLIBC_2.25 GLIBC_2.25 A +GLIBC_2.25 __iscanonicall F GLIBC_2.25 fegetmode F GLIBC_2.25 fesetexcept F GLIBC_2.25 fesetmode F diff --git a/sysdeps/unix/sysv/linux/m68k/m680x0/libm.abilist b/sysdeps/unix/sysv/linux/m68k/m680x0/libm.abilist index 77f9c7b..2a8cba4 100644 --- a/sysdeps/unix/sysv/linux/m68k/m680x0/libm.abilist +++ b/sysdeps/unix/sysv/linux/m68k/m680x0/libm.abilist @@ -421,6 +421,7 @@ GLIBC_2.24 nextup F GLIBC_2.24 nextupf F GLIBC_2.24 nextupl F GLIBC_2.25 GLIBC_2.25 A +GLIBC_2.25 __iscanonicall F GLIBC_2.25 fegetmode F GLIBC_2.25 fesetexcept F GLIBC_2.25 fesetmode F diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc32/fpu/libm.abilist b/sysdeps/unix/sysv/linux/powerpc/powerpc32/fpu/libm.abilist index 456f47e..155daf5 100644 --- a/sysdeps/unix/sysv/linux/powerpc/powerpc32/fpu/libm.abilist +++ b/sysdeps/unix/sysv/linux/powerpc/powerpc32/fpu/libm.abilist @@ -423,6 +423,7 @@ GLIBC_2.24 nextupf F GLIBC_2.24 nextupl F GLIBC_2.25 GLIBC_2.25 A GLIBC_2.25 __fe_dfl_mode D 0x8 +GLIBC_2.25 __iscanonicall F GLIBC_2.25 fegetmode F GLIBC_2.25 fesetexcept F GLIBC_2.25 fesetmode F diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc32/nofpu/libm.abilist b/sysdeps/unix/sysv/linux/powerpc/powerpc32/nofpu/libm.abilist index b2e4a73..7b38632 100644 --- a/sysdeps/unix/sysv/linux/powerpc/powerpc32/nofpu/libm.abilist +++ b/sysdeps/unix/sysv/linux/powerpc/powerpc32/nofpu/libm.abilist @@ -422,6 +422,7 @@ GLIBC_2.24 nextupf F GLIBC_2.24 nextupl F GLIBC_2.25 GLIBC_2.25 A GLIBC_2.25 __fe_dfl_mode D 0x8 +GLIBC_2.25 __iscanonicall F GLIBC_2.25 fegetmode F GLIBC_2.25 fesetexcept F GLIBC_2.25 fesetmode F diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc64/libm-le.abilist b/sysdeps/unix/sysv/linux/powerpc/powerpc64/libm-le.abilist index 907d8b7..26c8d41 100644 --- a/sysdeps/unix/sysv/linux/powerpc/powerpc64/libm-le.abilist +++ b/sysdeps/unix/sysv/linux/powerpc/powerpc64/libm-le.abilist @@ -417,6 +417,7 @@ GLIBC_2.24 nextupf F GLIBC_2.24 nextupl F GLIBC_2.25 GLIBC_2.25 A GLIBC_2.25 __fe_dfl_mode D 0x8 +GLIBC_2.25 __iscanonicall F GLIBC_2.25 fegetmode F GLIBC_2.25 fesetexcept F GLIBC_2.25 fesetmode F diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc64/libm.abilist b/sysdeps/unix/sysv/linux/powerpc/powerpc64/libm.abilist index 307423d..0e76e88 100644 --- a/sysdeps/unix/sysv/linux/powerpc/powerpc64/libm.abilist +++ b/sysdeps/unix/sysv/linux/powerpc/powerpc64/libm.abilist @@ -98,6 +98,7 @@ GLIBC_2.24 nextupf F GLIBC_2.24 nextupl F GLIBC_2.25 GLIBC_2.25 A GLIBC_2.25 __fe_dfl_mode D 0x8 +GLIBC_2.25 __iscanonicall F GLIBC_2.25 fegetmode F GLIBC_2.25 fesetexcept F GLIBC_2.25 fesetmode F diff --git a/sysdeps/unix/sysv/linux/x86_64/64/libm.abilist b/sysdeps/unix/sysv/linux/x86_64/64/libm.abilist index 65e5bae..1965316 100644 --- a/sysdeps/unix/sysv/linux/x86_64/64/libm.abilist +++ b/sysdeps/unix/sysv/linux/x86_64/64/libm.abilist @@ -412,6 +412,7 @@ GLIBC_2.24 nextup F GLIBC_2.24 nextupf F GLIBC_2.24 nextupl F GLIBC_2.25 GLIBC_2.25 A +GLIBC_2.25 __iscanonicall F GLIBC_2.25 fegetmode F GLIBC_2.25 fesetexcept F GLIBC_2.25 fesetmode F diff --git a/sysdeps/unix/sysv/linux/x86_64/x32/libm.abilist b/sysdeps/unix/sysv/linux/x86_64/x32/libm.abilist index ff520cb..82207c9 100644 --- a/sysdeps/unix/sysv/linux/x86_64/x32/libm.abilist +++ b/sysdeps/unix/sysv/linux/x86_64/x32/libm.abilist @@ -411,6 +411,7 @@ GLIBC_2.24 nextup F GLIBC_2.24 nextupf F GLIBC_2.24 nextupl F GLIBC_2.25 GLIBC_2.25 A +GLIBC_2.25 __iscanonicall F GLIBC_2.25 fegetmode F GLIBC_2.25 fesetexcept F GLIBC_2.25 fesetmode F