From patchwork Mon Apr 4 20:58:05 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Adhemerval Zanella X-Patchwork-Id: 1613185 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: bilbo.ozlabs.org; dkim=pass (1024-bit key; secure) header.d=sourceware.org header.i=@sourceware.org header.a=rsa-sha256 header.s=default header.b=jt+M5jeo; dkim-atps=neutral Authentication-Results: ozlabs.org; spf=pass (sender SPF authorized) smtp.mailfrom=sourceware.org (client-ip=2620:52:3:1:0:246e:9693:128c; helo=sourceware.org; envelope-from=libc-alpha-bounces+incoming=patchwork.ozlabs.org@sourceware.org; receiver=) Received: from sourceware.org (server2.sourceware.org [IPv6:2620:52:3:1:0:246e:9693:128c]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits) server-digest SHA256) (No client certificate requested) by bilbo.ozlabs.org (Postfix) with ESMTPS id 4KXNRK4n2hz9s5V for ; Tue, 5 Apr 2022 06:58:37 +1000 (AEST) Received: from server2.sourceware.org (localhost [IPv6:::1]) by sourceware.org (Postfix) with ESMTP id CED533858C52 for ; Mon, 4 Apr 2022 20:58:34 +0000 (GMT) DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org CED533858C52 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=sourceware.org; s=default; t=1649105914; bh=FEUe+c+nrhMcBShUBTZQajJ3fY2yYHQqogcT9U/OyPE=; h=To:Subject:Date:List-Id:List-Unsubscribe:List-Archive:List-Post: List-Help:List-Subscribe:From:Reply-To:From; b=jt+M5jeozq5GArcqYe49CxE++nxlIqaPPzv6lh90Zc6EVGZy5QhmlehpGOrNc+Ppx tKU6ldJu0nTHmPDrfqWuF3ei12EDGka/hUxXRCWvG3YfekjupnoOhJQx1YiYMAJacz C9Jv14Za6VHmTH3I2HMJzhmYraPUsUtV9bVUA8SA= X-Original-To: libc-alpha@sourceware.org Delivered-To: libc-alpha@sourceware.org Received: from mail-ot1-x32f.google.com (mail-ot1-x32f.google.com [IPv6:2607:f8b0:4864:20::32f]) by sourceware.org (Postfix) with ESMTPS id 17BA73858D37 for ; Mon, 4 Apr 2022 20:58:19 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.1 sourceware.org 17BA73858D37 Received: by mail-ot1-x32f.google.com with SMTP id b17-20020a0568301df100b005ce0456a9efso8053042otj.9 for ; Mon, 04 Apr 2022 13:58:19 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:from:to:subject:date:message-id:mime-version :content-transfer-encoding; bh=FEUe+c+nrhMcBShUBTZQajJ3fY2yYHQqogcT9U/OyPE=; b=Jt8flVZlTd5hlbiWRuD9mFxWZv+16DHntYR31e+cGxLI4c1Fpx0fBWVBSHZn8iLx9Q qNV0B8HrSevTLVgxGX5cjCv870gTEns43QqPPb4hN7ad30hnZSqBlVRjDLrx+hkdWmu9 YCWla9sopHMQDVrLFBipcm6GjhSxUm0KcGOgEIJ6Gn5irgSG7nVgnt3nllVpJPfGjBR4 mQIlcJR6C1p2XRWLoEX93dsU3u2MQ6tXS9x6kjo2WbRw+nAtmC0rUOGohAXrUIbLxsd9 UK2K47Hjbz1AiKrPsrpiTnksPnsQ2ZXigRlPyctqFrXTkix32ZGIwPAnx0ngzn3bU445 pchA== X-Gm-Message-State: AOAM531fdxkPWh3UANWEVuf+aAH9MPRPa1qRNgsj7hISceytYpjO5MsU 4ZzzNiEC5LHLipxJKA4C2G/nUXLXH2kKuw== X-Google-Smtp-Source: ABdhPJxrQvcTyWv3SVYq5TS5XX1PEl2GVV59MotzC6Pku5gz30RW4mXpQ8tFGWEcLpThj0jsdFHgKQ== X-Received: by 2002:a9d:666:0:b0:5cd:fff5:d71b with SMTP id 93-20020a9d0666000000b005cdfff5d71bmr75448otn.46.1649105897722; Mon, 04 Apr 2022 13:58:17 -0700 (PDT) Received: from birita.. ([2804:431:c7cb:a6c0:94cf:60bc:16d1:2727]) by smtp.gmail.com with ESMTPSA id x18-20020a056830245200b005ce01c28c77sm4936672otr.1.2022.04.04.13.58.16 for (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Mon, 04 Apr 2022 13:58:17 -0700 (PDT) To: libc-alpha@sourceware.org Subject: [PATCH] math: Add math-use-builtins-fabs Date: Mon, 4 Apr 2022 17:58:05 -0300 Message-Id: <20220404205805.701759-1-adhemerval.zanella@linaro.org> X-Mailer: git-send-email 2.32.0 MIME-Version: 1.0 X-Spam-Status: No, score=-11.9 required=5.0 tests=BAYES_00, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, GIT_PATCH_0, KAM_ASCII_DIVIDERS, KAM_SHORT, RCVD_IN_DNSWL_NONE, SPF_HELO_NONE, SPF_PASS, TXREP, T_SCC_BODY_TEXT_LINE autolearn=ham autolearn_force=no version=3.4.4 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on server2.sourceware.org X-BeenThere: libc-alpha@sourceware.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Libc-alpha mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-Patchwork-Original-From: Adhemerval Zanella via Libc-alpha From: Adhemerval Zanella Reply-To: Adhemerval Zanella Errors-To: libc-alpha-bounces+incoming=patchwork.ozlabs.org@sourceware.org Sender: "Libc-alpha" Both float and double are assumed to be supported (since generic implementation uses it without a fallback), only long double and float128 are defined. It allows to remove i686, ia64, x86_64, powerpc, and sparc arch specific implementation. On ia64 it also fixes: math/test-float64x-fabs math/test-ldouble-fabs Checked on x86_64-linux-gnu, i686-linux-gnu, powerpc-linux-gnu, powerpc64-linux-gnu, sparc64-linux-gnu, and ia64-linux-gnu. --- sysdeps/generic/math-use-builtins-fabs.h | 4 + sysdeps/generic/math-use-builtins.h | 1 + sysdeps/i386/fpu/s_fabsl.S | 10 --- sysdeps/ia64/fpu/s_fabsl.S | 82 ------------------- sysdeps/ia64/math-use-builtins-fabs.h | 4 + sysdeps/ieee754/ldbl-128/s_fabsl.c | 5 ++ sysdeps/ieee754/ldbl-128ibm/s_fabsl.c | 5 ++ .../s_fabsl.S => ieee754/ldbl-96/s_fabsl.c} | 22 +++-- sysdeps/powerpc/fpu/math-use-builtins-fabs.h | 9 ++ sysdeps/powerpc/powerpc32/fpu/s_fabsl.S | 52 ------------ sysdeps/powerpc/powerpc64/fpu/s_fabsl.S | 34 -------- sysdeps/sparc/fpu/math-use-builtins-fabs.h | 4 + sysdeps/sparc/sparc32/fpu/s_fabsl.c | 9 -- sysdeps/sparc/sparc64/fpu/s_fabsl.c | 7 -- sysdeps/x86/math-use-builtins-fabs.h | 4 + 15 files changed, 49 insertions(+), 203 deletions(-) create mode 100644 sysdeps/generic/math-use-builtins-fabs.h delete mode 100644 sysdeps/i386/fpu/s_fabsl.S delete mode 100644 sysdeps/ia64/fpu/s_fabsl.S create mode 100644 sysdeps/ia64/math-use-builtins-fabs.h rename sysdeps/{x86_64/fpu/s_fabsl.S => ieee754/ldbl-96/s_fabsl.c} (72%) create mode 100644 sysdeps/powerpc/fpu/math-use-builtins-fabs.h delete mode 100644 sysdeps/powerpc/powerpc32/fpu/s_fabsl.S delete mode 100644 sysdeps/powerpc/powerpc64/fpu/s_fabsl.S create mode 100644 sysdeps/sparc/fpu/math-use-builtins-fabs.h delete mode 100644 sysdeps/sparc/sparc32/fpu/s_fabsl.c delete mode 100644 sysdeps/sparc/sparc64/fpu/s_fabsl.c create mode 100644 sysdeps/x86/math-use-builtins-fabs.h diff --git a/sysdeps/generic/math-use-builtins-fabs.h b/sysdeps/generic/math-use-builtins-fabs.h new file mode 100644 index 0000000000..8f3da3cf21 --- /dev/null +++ b/sysdeps/generic/math-use-builtins-fabs.h @@ -0,0 +1,4 @@ +/* Generic implementation already assume builtin support for + float and double. */ +#define USE_FABSL_BUILTIN 0 +#define USE_FABSF128_BUILTIN 0 diff --git a/sysdeps/generic/math-use-builtins.h b/sysdeps/generic/math-use-builtins.h index 3ec15e59d4..6bd424d900 100644 --- a/sysdeps/generic/math-use-builtins.h +++ b/sysdeps/generic/math-use-builtins.h @@ -36,5 +36,6 @@ #include #include #include +#include #endif /* MATH_USE_BUILTINS_H */ diff --git a/sysdeps/i386/fpu/s_fabsl.S b/sysdeps/i386/fpu/s_fabsl.S deleted file mode 100644 index 6642ed4151..0000000000 --- a/sysdeps/i386/fpu/s_fabsl.S +++ /dev/null @@ -1,10 +0,0 @@ -#include -#include - - .text -ENTRY(__fabsl) - fldt 4(%esp) - fabs - ret -END(__fabsl) -libm_alias_ldouble (__fabs, fabs) diff --git a/sysdeps/ia64/fpu/s_fabsl.S b/sysdeps/ia64/fpu/s_fabsl.S deleted file mode 100644 index 07838b1057..0000000000 --- a/sysdeps/ia64/fpu/s_fabsl.S +++ /dev/null @@ -1,82 +0,0 @@ -.file "fabsl.s" - - -// Copyright (c) 2000 - 2003, Intel Corporation -// All rights reserved. -// -// -// Redistribution and use in source and binary forms, with or without -// modification, are permitted provided that the following conditions are -// met: -// -// * Redistributions of source code must retain the above copyright -// notice, this list of conditions and the following disclaimer. -// -// * Redistributions in binary form must reproduce the above copyright -// notice, this list of conditions and the following disclaimer in the -// documentation and/or other materials provided with the distribution. -// -// * The name of Intel Corporation may not be used to endorse or promote -// products derived from this software without specific prior written -// permission. - -// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS -// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT -// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR -// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL INTEL OR ITS -// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, -// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, -// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR -// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY -// OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY OR TORT (INCLUDING -// NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS -// SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -// -// Intel Corporation is the author of this code, and requests that all -// problem reports or change requests be submitted to it directly at -// http://www.intel.com/software/products/opensource/libraries/num.htm. -// -// History -//============================================================== -// 02/02/00 Initial version -// 02/07/02 Added __libm_fabsl entry point to test in case compiler inlines -// 05/20/02 Cleaned up namespace and sf0 syntax -// 02/10/03 Reordered header: .section, .global, .proc, .align -// -// API -//============================================================== -// long double fabsl (long double x) -// -// Overview of operation -//============================================================== -// returns absolute value of x - -// floating-point registers used: 1 -// f8, input - -.section .text -.global __libm_fabsl# - -.proc __libm_fabsl# -__libm_fabsl: -.endp __libm_fabsl# - -GLOBAL_IEEE754_ENTRY(fabsl) - -// set invalid or denormal flags and take fault if -// necessary - -{ .mfi - nop.m 999 - fcmp.eq.unc.s0 p6,p7 = f8,f1 - nop.i 999 ;; -} - -{ .mfb - nop.m 999 - fmerge.s f8 = f0,f8 - br.ret.sptk b0 ;; -} - -GLOBAL_IEEE754_END(fabsl) -libm_alias_ldouble_other (__fabs, fabs) diff --git a/sysdeps/ia64/math-use-builtins-fabs.h b/sysdeps/ia64/math-use-builtins-fabs.h new file mode 100644 index 0000000000..facf3a6425 --- /dev/null +++ b/sysdeps/ia64/math-use-builtins-fabs.h @@ -0,0 +1,4 @@ +/* Generic implementation already assume builtin support for + float and double. */ +#define USE_FABSL_BUILTIN 1 +#define USE_FABSF128_BUILTIN 1 diff --git a/sysdeps/ieee754/ldbl-128/s_fabsl.c b/sysdeps/ieee754/ldbl-128/s_fabsl.c index 7e08980252..464d62fa1b 100644 --- a/sysdeps/ieee754/ldbl-128/s_fabsl.c +++ b/sysdeps/ieee754/ldbl-128/s_fabsl.c @@ -23,12 +23,17 @@ static char rcsid[] = "$NetBSD: $"; #include #include #include +#include _Float128 __fabsl(_Float128 x) { +#if USE_FABSF128_BUILTIN + return __builtin_fabsf128 (x); +#else uint64_t hx; GET_LDOUBLE_MSW64(hx,x); SET_LDOUBLE_MSW64(x,hx&0x7fffffffffffffffLL); return x; +#endif } libm_alias_ldouble (__fabs, fabs) diff --git a/sysdeps/ieee754/ldbl-128ibm/s_fabsl.c b/sysdeps/ieee754/ldbl-128ibm/s_fabsl.c index cba7de6660..7f26315e84 100644 --- a/sysdeps/ieee754/ldbl-128ibm/s_fabsl.c +++ b/sysdeps/ieee754/ldbl-128ibm/s_fabsl.c @@ -24,9 +24,13 @@ static char rcsid[] = "$NetBSD: $"; #include #include #include +#include long double __fabsl(long double x) { +#if USE_FABSL_BUILTIN + return __builtin_fabsl (x); +#else uint64_t hx, lx; double xhi, xlo; @@ -39,5 +43,6 @@ long double __fabsl(long double x) INSERT_WORDS64 (xlo, lx); x = ldbl_pack (xhi, xlo); return x; +#endif } long_double_symbol (libm, __fabsl, fabsl); diff --git a/sysdeps/x86_64/fpu/s_fabsl.S b/sysdeps/ieee754/ldbl-96/s_fabsl.c similarity index 72% rename from sysdeps/x86_64/fpu/s_fabsl.S rename to sysdeps/ieee754/ldbl-96/s_fabsl.c index cfcf01e027..c8a6a50918 100644 --- a/sysdeps/x86_64/fpu/s_fabsl.S +++ b/sysdeps/ieee754/ldbl-96/s_fabsl.c @@ -1,5 +1,5 @@ -/* Absolute value of floating point number. - Copyright (C) 2002-2022 Free Software Foundation, Inc. +/* Absolute value of floating-point number. + Copyright (C) 2022 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 @@ -16,13 +16,17 @@ License along with the GNU C Library; if not, see . */ -#include +#include #include +#include - .text -ENTRY(__fabsl) - fldt 8(%rsp) - fabs - ret -END(__fabsl) +long double +__fabsl (long double x) +{ +#if USE_FABSL_BUILTIN + return __builtin_fabsl (x); +#else +# error "missing fabsl implementation for lbdl-96" +#endif +} libm_alias_ldouble (__fabs, fabs) diff --git a/sysdeps/powerpc/fpu/math-use-builtins-fabs.h b/sysdeps/powerpc/fpu/math-use-builtins-fabs.h new file mode 100644 index 0000000000..3100a004fe --- /dev/null +++ b/sysdeps/powerpc/fpu/math-use-builtins-fabs.h @@ -0,0 +1,9 @@ +/* Generic implementation already assume builtin support for + float and double. */ +#define USE_FABSL_BUILTIN 1 +/* This is not available for P8 or BE targets. */ +#ifdef __FLOAT128__ +# define USE_FABSF128_BUILTIN 1 +#else +# define USE_FABSF128_BUILTIN 0 +#endif diff --git a/sysdeps/powerpc/powerpc32/fpu/s_fabsl.S b/sysdeps/powerpc/powerpc32/fpu/s_fabsl.S deleted file mode 100644 index 8bdcdc8cfb..0000000000 --- a/sysdeps/powerpc/powerpc32/fpu/s_fabsl.S +++ /dev/null @@ -1,52 +0,0 @@ -/* Copy a sign bit between floating-point values. - IBM extended format long double version. - Copyright (C) 2004-2022 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 - -ENTRY(__fabsl) -/* long double [f1,f2] fabs (long double [f1,f2] x); - fabs(x,y) returns a value with the magnitude of x and - with the sign bit of y. */ -#ifdef _ARCH_PPCGR - /* fsel available. */ - fmr fp0,fp1 -#else - /* Use integer operations to test sign of high part to avoid - exceptions on sNaNs. */ - stwu r1,-16(r1) - cfi_adjust_cfa_offset (16) - stfd fp1,8(r1) -#endif - fabs fp1,fp1 -#ifdef _ARCH_PPCGR - fneg fp3,fp2 - fsel fp2,fp0,fp2,fp3 -#else - lwz r3,8+HIWORD(r1) - cmpwi cr6,r3,0 - addi r1,r1,16 - cfi_adjust_cfa_offset (-16) - bgelr cr6 - fneg fp2,fp2 -#endif - blr -END (__fabsl) - -long_double_symbol (libm, __fabsl, fabsl) diff --git a/sysdeps/powerpc/powerpc64/fpu/s_fabsl.S b/sysdeps/powerpc/powerpc64/fpu/s_fabsl.S deleted file mode 100644 index ec796a1cd3..0000000000 --- a/sysdeps/powerpc/powerpc64/fpu/s_fabsl.S +++ /dev/null @@ -1,34 +0,0 @@ -/* Copy a sign bit between floating-point values. - IBM extended format long double version. - Copyright (C) 2004-2022 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 - -ENTRY_TOCLESS (__fabsl) -/* long double [f1,f2] fabs (long double [f1,f2] x); - fabs(x,y) returns a value with the magnitude of x and - with the sign bit of y. */ - fmr fp0,fp1 - fabs fp1,fp1 - fneg fp3,fp2 - fsel fp2,fp0,fp2,fp3 - blr -END (__fabsl) - -long_double_symbol (libm, __fabsl, fabsl) diff --git a/sysdeps/sparc/fpu/math-use-builtins-fabs.h b/sysdeps/sparc/fpu/math-use-builtins-fabs.h new file mode 100644 index 0000000000..702229ddd4 --- /dev/null +++ b/sysdeps/sparc/fpu/math-use-builtins-fabs.h @@ -0,0 +1,4 @@ +/* Generic implementation already assume builtin support for + float and double. */ +#define USE_FABSL_BUILTIN 1 +#define USE_FABSF128_BUILTIN 0 diff --git a/sysdeps/sparc/sparc32/fpu/s_fabsl.c b/sysdeps/sparc/sparc32/fpu/s_fabsl.c deleted file mode 100644 index 411a2e1511..0000000000 --- a/sysdeps/sparc/sparc32/fpu/s_fabsl.c +++ /dev/null @@ -1,9 +0,0 @@ -#include -#include -#include - -long double __fabsl (long double x) -{ - return __builtin_fabsl (x); -} -libm_alias_ldouble (__fabs, fabs) diff --git a/sysdeps/sparc/sparc64/fpu/s_fabsl.c b/sysdeps/sparc/sparc64/fpu/s_fabsl.c deleted file mode 100644 index 72e1c51340..0000000000 --- a/sysdeps/sparc/sparc64/fpu/s_fabsl.c +++ /dev/null @@ -1,7 +0,0 @@ -#include - -long double __fabsl (long double x) -{ - return __builtin_fabsl (x); -} -libm_alias_ldouble (__fabs, fabs) diff --git a/sysdeps/x86/math-use-builtins-fabs.h b/sysdeps/x86/math-use-builtins-fabs.h new file mode 100644 index 0000000000..facf3a6425 --- /dev/null +++ b/sysdeps/x86/math-use-builtins-fabs.h @@ -0,0 +1,4 @@ +/* Generic implementation already assume builtin support for + float and double. */ +#define USE_FABSL_BUILTIN 1 +#define USE_FABSF128_BUILTIN 1