From patchwork Wed Apr 3 19:39:17 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Adhemerval Zanella Netto X-Patchwork-Id: 1919399 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@legolas.ozlabs.org Authentication-Results: legolas.ozlabs.org; dkim=pass (2048-bit key; unprotected) header.d=linaro.org header.i=@linaro.org header.a=rsa-sha256 header.s=google header.b=DStWMAN6; dkim-atps=neutral Authentication-Results: legolas.ozlabs.org; spf=pass (sender SPF authorized) smtp.mailfrom=sourceware.org (client-ip=8.43.85.97; helo=server2.sourceware.org; envelope-from=libc-alpha-bounces+incoming=patchwork.ozlabs.org@sourceware.org; receiver=patchwork.ozlabs.org) Received: from server2.sourceware.org (server2.sourceware.org [8.43.85.97]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature ECDSA (secp384r1) server-digest SHA384) (No client certificate requested) by legolas.ozlabs.org (Postfix) with ESMTPS id 4V8w7X4DSWz1yYP for ; Thu, 4 Apr 2024 06:39:52 +1100 (AEDT) Received: from server2.sourceware.org (localhost [IPv6:::1]) by sourceware.org (Postfix) with ESMTP id DA25838449C8 for ; Wed, 3 Apr 2024 19:39:50 +0000 (GMT) X-Original-To: libc-alpha@sourceware.org Delivered-To: libc-alpha@sourceware.org Received: from mail-pf1-x42b.google.com (mail-pf1-x42b.google.com [IPv6:2607:f8b0:4864:20::42b]) by sourceware.org (Postfix) with ESMTPS id 0ADB2384772F for ; Wed, 3 Apr 2024 19:39:28 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 0ADB2384772F Authentication-Results: sourceware.org; dmarc=pass (p=none dis=none) header.from=linaro.org Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=linaro.org ARC-Filter: OpenARC Filter v1.0.0 sourceware.org 0ADB2384772F Authentication-Results: server2.sourceware.org; arc=none smtp.remote-ip=2607:f8b0:4864:20::42b ARC-Seal: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1712173171; cv=none; b=vYyAFBrwzaFlgHCQ6gDKIEUzNV8gcaiaclrT4/4PsiHmZqpqQAB/KZB7EUCqQ3kQXnjV67aC/NFyaExCCxhfSVelXPjimtJX1H27RKadRfPZafBxIY8DLgMffaoakluCDhxa5smdNzeWXwKTE7FDDU51DBmDRdDELxqqfEgWz0U= ARC-Message-Signature: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1712173171; c=relaxed/simple; bh=GD3kc8TMA33sMwaxg2v4MjLMVrIy7/c+Nf2HXK9xLN8=; h=DKIM-Signature:From:To:Subject:Date:Message-Id:MIME-Version; b=LBUKNVivDXty1MHaen6QjTRLkTVjThZHw8m2KPfFtnBiglBcmolkcHBtYngQPov1QFCVlL11FJGQ/F3Vnvo614Qe3MY+lRzP2HmkkXonDIzEHtST3vzHByj1zsRwzzLI1FPZNRsh6ATSSpgvoEE04rPFJFDC+K9LvN23mB5LOFE= ARC-Authentication-Results: i=1; server2.sourceware.org Received: by mail-pf1-x42b.google.com with SMTP id d2e1a72fcca58-6ead4093f85so137938b3a.3 for ; Wed, 03 Apr 2024 12:39:27 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linaro.org; s=google; t=1712173166; x=1712777966; darn=sourceware.org; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:from:to:cc:subject:date :message-id:reply-to; bh=Xbhj/mGMrU6+j8Gxv9BiDy9XoEMvBREDMCF5bAm9ILA=; b=DStWMAN6CFECmiKe04kHk11EceaTxIJR09sZPfMTf9CUf8Mhkpv7kxfXyIctv4h4kI Valdv+BsegaKXeeU+QPqpGucAwTEgu4Hygxvf0NXkbv744HiklmY/WHx0C/mlx3OcjII U0DNTQtOJo6KLtMBE8tEX+F4cep1rXywV8uvpCCf3EbUIP3rv4+djASeS/s/vPX6g5YO 2MEMiSi3dpFd7wQ2c5D0oQ/b4LIdteOztJB3jIc+VdkxrmCgyqbU14WP890j9e5riZwO ty2EM0EMa13RXhoiPytpIZXy3BZtHAYjOovL7zpsEKfRRyr4UhjYa+yR2FOY6fVl06xy jp9g== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1712173166; x=1712777966; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:x-gm-message-state:from:to:cc :subject:date:message-id:reply-to; bh=Xbhj/mGMrU6+j8Gxv9BiDy9XoEMvBREDMCF5bAm9ILA=; b=h/nbNGsB7/XNW3od2Qra2PkDrgYPBvmXVmxwyzoAOOoo1ew+FjuZbXgT/7huXyL5rq rCnfbb0nYDcvGL+Y8VrOj/qfdFyox7Jqjey42tlcb+J8RpLTwYWPpCwVwbsa1L9atv4U glZM9P+gBx3FVyRMkq/j1pwa2K9OvJ3ku1Ka/Iki6mvbMKIWvyrMf1Fh8mKkRvNGtXNp e0XAK4vaRHfszTqzT67xBGfRktvVev7zQNgE0hI3OSrZsbxKmoD8ZB3TgntkQo+ciykf vavrH7dp036poroh3Zft1/bVZ08SZClmvwGX+FbUcqKuPvBfvoGjS4BOiczs3dcbFwa2 t7mw== X-Gm-Message-State: AOJu0YzWBGKE5o01RH59xH4LjrXGjRqXiIZSvTLtlWaCYkvLoBnAm+t0 ckUA1agNcFwMWgxr/nAeTE7Lv11DTFNMvlGzyOHJSkhl4TPqw3bD0e4vj2oRrist1zp00Atiexu h X-Google-Smtp-Source: AGHT+IGdCZuW0/hb8o3sFs0esoaQ8WGHOvTyMnA+tJ0dsTWNa+9zY7ZddaLgwSKJ8cxbUhpZEUGczw== X-Received: by 2002:a05:6a00:3991:b0:6e7:117:c5d5 with SMTP id fi17-20020a056a00399100b006e70117c5d5mr571788pfb.23.1712173166224; Wed, 03 Apr 2024 12:39:26 -0700 (PDT) Received: from mandiga.. ([2804:1b3:a7c3:b18e:82b0:6301:7f28:5bcc]) by smtp.gmail.com with ESMTPSA id fd30-20020a056a002e9e00b006eaf43b5982sm8913535pfb.108.2024.04.03.12.39.24 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 03 Apr 2024 12:39:25 -0700 (PDT) From: Adhemerval Zanella To: libc-alpha@sourceware.org Cc: "H . J . Lu" , Joseph Myers Subject: [PATCH 1/3] math: math: x86 ceill traps when FE_INEXACT is enabled (BZ 31600) Date: Wed, 3 Apr 2024 16:39:17 -0300 Message-Id: <20240403193919.1533786-2-adhemerval.zanella@linaro.org> X-Mailer: git-send-email 2.34.1 In-Reply-To: <20240403193919.1533786-1-adhemerval.zanella@linaro.org> References: <20240403193919.1533786-1-adhemerval.zanella@linaro.org> MIME-Version: 1.0 X-Spam-Status: No, score=-12.1 required=5.0 tests=BAYES_00, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, GIT_PATCH_0, KAM_SHORT, RCVD_IN_DNSWL_NONE, SPF_HELO_NONE, SPF_PASS, TXREP autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org X-BeenThere: libc-alpha@sourceware.org X-Mailman-Version: 2.1.30 Precedence: list List-Id: Libc-alpha mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: libc-alpha-bounces+incoming=patchwork.ozlabs.org@sourceware.org The implementations of ceil functions using x87 floating point (i386 and x86_64 long double only) traps when FE_INEXACT is enabled. Although this is a GNU extension outside the scope of the C standard, other architectures that also support traps do not show this behavior. The fix moves the implementation to a common one that holds any exceptions with a 'fnclex' (libc_feholdexcept_setround_387). Checked on x86_64-linux-gnu and i686-linux-gnu. Reviewed-by: H.J. Lu --- math/Makefile | 3 + math/test-ceil-except-2.c | 67 +++++++++++++++++++++ sysdeps/i386/fpu/s_ceil.S | 34 ----------- sysdeps/i386/fpu/s_ceil.c | 25 ++++++++ sysdeps/i386/fpu/s_ceilf.S | 34 ----------- sysdeps/i386/fpu/s_ceilf.c | 25 ++++++++ sysdeps/i386/fpu/s_ceill.S | 39 ------------ sysdeps/x86/fpu/s_ceill.c | 25 ++++++++ sysdeps/x86/fpu/s_nearestint_387_template.c | 36 +++++++++++ sysdeps/x86_64/fpu/s_ceill.S | 34 ----------- 10 files changed, 181 insertions(+), 141 deletions(-) create mode 100644 math/test-ceil-except-2.c delete mode 100644 sysdeps/i386/fpu/s_ceil.S create mode 100644 sysdeps/i386/fpu/s_ceil.c delete mode 100644 sysdeps/i386/fpu/s_ceilf.S create mode 100644 sysdeps/i386/fpu/s_ceilf.c delete mode 100644 sysdeps/i386/fpu/s_ceill.S create mode 100644 sysdeps/x86/fpu/s_ceill.c create mode 100644 sysdeps/x86/fpu/s_nearestint_387_template.c delete mode 100644 sysdeps/x86_64/fpu/s_ceill.S diff --git a/math/Makefile b/math/Makefile index 121a709121..d2a740eebe 100644 --- a/math/Makefile +++ b/math/Makefile @@ -498,6 +498,7 @@ tests = \ bug-nextafter \ bug-nexttoward \ bug-tgmath1 \ + test-ceil-except-2 \ test-femode \ test-femode-traps \ test-fenv basic-test \ @@ -989,6 +990,8 @@ CFLAGS-test-fe-snans-always-signal.c += $(config-cflags-signaling-nans) CFLAGS-test-nan-const.c += -fno-builtin +CFLAGS-test-ceil-except-2.c += -fno-builtin + include ../Rules gen-all-calls = $(gen-libm-calls) $(gen-calls) diff --git a/math/test-ceil-except-2.c b/math/test-ceil-except-2.c new file mode 100644 index 0000000000..394a272d89 --- /dev/null +++ b/math/test-ceil-except-2.c @@ -0,0 +1,67 @@ +/* Test ceil functions do not disable exception traps. + Copyright (C) 2024 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 + +#ifndef FE_INEXACT +# define FE_INEXACT 0 +#endif + +#define TEST_FUNC(NAME, FLOAT, SUFFIX) \ +static int \ +NAME (void) \ +{ \ + int result = 0; \ + volatile FLOAT a, b __attribute__ ((unused)); \ + a = 1.5; \ + /* ceil must work when traps on "inexact" are enabled. */ \ + b = ceil ## SUFFIX (a); \ + /* And it must have left those traps enabled. */ \ + if (fegetexcept () == FE_INEXACT) \ + puts ("PASS: " #FLOAT); \ + else \ + { \ + puts ("FAIL: " #FLOAT); \ + result = 1; \ + } \ + return result; \ +} + +TEST_FUNC (float_test, float, f) +TEST_FUNC (double_test, double, ) +TEST_FUNC (ldouble_test, long double, l) + +static int +do_test (void) +{ + if (feenableexcept (FE_INEXACT) == -1) + { + puts ("enabling FE_INEXACT traps failed, cannot test"); + return 77; + } + int result = float_test (); + feenableexcept (FE_INEXACT); + result |= double_test (); + feenableexcept (FE_INEXACT); + result |= ldouble_test (); + return result; +} + +#include diff --git a/sysdeps/i386/fpu/s_ceil.S b/sysdeps/i386/fpu/s_ceil.S deleted file mode 100644 index 99984f9b8d..0000000000 --- a/sysdeps/i386/fpu/s_ceil.S +++ /dev/null @@ -1,34 +0,0 @@ -/* - * Public domain. - */ - -#include -#include - -RCSID("$NetBSD: s_ceil.S,v 1.4 1995/05/08 23:52:13 jtc Exp $") - -ENTRY(__ceil) - fldl 4(%esp) - subl $32,%esp - cfi_adjust_cfa_offset (32) - - fnstenv 4(%esp) /* store fpu environment */ - - /* We use here %edx although only the low 1 bits are defined. - But none of the operations should care and they are faster - than the 16 bit operations. */ - movl $0x0800,%edx /* round towards +oo */ - orl 4(%esp),%edx - andl $0xfbff,%edx - movl %edx,(%esp) - fldcw (%esp) /* load modified control word */ - - frndint /* round */ - - fldenv 4(%esp) /* restore original environment */ - - addl $32,%esp - cfi_adjust_cfa_offset (-32) - ret -END (__ceil) -libm_alias_double (__ceil, ceil) diff --git a/sysdeps/i386/fpu/s_ceil.c b/sysdeps/i386/fpu/s_ceil.c new file mode 100644 index 0000000000..349135c5d3 --- /dev/null +++ b/sysdeps/i386/fpu/s_ceil.c @@ -0,0 +1,25 @@ +/* Return smallest integral value not less than argument. i386 version. + Copyright (C) 2024 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public + License as published by the Free Software Foundation; either + version 2.1 of the License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public + License along with the GNU C Library; if not, see + . */ + +#include + +#define FUNC __ceil +#define TYPE double +#define FE_OPTION FE_UPWARD +#include "s_nearestint_387_template.c" +libm_alias_double (__ceil, ceil) diff --git a/sysdeps/i386/fpu/s_ceilf.S b/sysdeps/i386/fpu/s_ceilf.S deleted file mode 100644 index 03e8e22609..0000000000 --- a/sysdeps/i386/fpu/s_ceilf.S +++ /dev/null @@ -1,34 +0,0 @@ -/* - * Public domain. - */ - -#include -#include - -RCSID("$NetBSD: s_ceilf.S,v 1.3 1995/05/08 23:52:44 jtc Exp $") - -ENTRY(__ceilf) - flds 4(%esp) - subl $32,%esp - cfi_adjust_cfa_offset (32) - - fnstenv 4(%esp) /* store fpu environment */ - - /* We use here %edx although only the low 1 bits are defined. - But none of the operations should care and they are faster - than the 16 bit operations. */ - movl $0x0800,%edx /* round towards +oo */ - orl 4(%esp),%edx - andl $0xfbff,%edx - movl %edx,(%esp) - fldcw (%esp) /* load modified control word */ - - frndint /* round */ - - fldenv 4(%esp) /* restore original environment */ - - addl $32,%esp - cfi_adjust_cfa_offset (-32) - ret -END (__ceilf) -libm_alias_float (__ceil, ceil) diff --git a/sysdeps/i386/fpu/s_ceilf.c b/sysdeps/i386/fpu/s_ceilf.c new file mode 100644 index 0000000000..e73a20fd71 --- /dev/null +++ b/sysdeps/i386/fpu/s_ceilf.c @@ -0,0 +1,25 @@ +/* Return largest integral value not less than argument. i386 version. + Copyright (C) 2024 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public + License as published by the Free Software Foundation; either + version 2.1 of the License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public + License along with the GNU C Library; if not, see + . */ + +#include + +#define FUNC __ceilf +#define TYPE float +#define FE_OPTION FE_UPWARD +#include "s_nearestint_387_template.c" +libm_alias_float (__ceil, ceil) diff --git a/sysdeps/i386/fpu/s_ceill.S b/sysdeps/i386/fpu/s_ceill.S deleted file mode 100644 index a551fce7f9..0000000000 --- a/sysdeps/i386/fpu/s_ceill.S +++ /dev/null @@ -1,39 +0,0 @@ -/* - * Public domain. - */ - -#include -#include - -RCSID("$NetBSD: $") - -ENTRY(__ceill) - fldt 4(%esp) - subl $32,%esp - cfi_adjust_cfa_offset (32) - - fnstenv 4(%esp) /* store fpu environment */ - - /* We use here %edx although only the low 1 bits are defined. - But none of the operations should care and they are faster - than the 16 bit operations. */ - movl $0x0800,%edx /* round towards +oo */ - orl 4(%esp),%edx - andl $0xfbff,%edx - movl %edx,(%esp) - fldcw (%esp) /* load modified control word */ - - frndint /* round */ - - /* Preserve "invalid" exceptions from sNaN input. */ - fnstsw - andl $0x1, %eax - orl %eax, 8(%esp) - - fldenv 4(%esp) /* restore original environment */ - - addl $32,%esp - cfi_adjust_cfa_offset (-32) - ret -END (__ceill) -libm_alias_ldouble (__ceil, ceil) diff --git a/sysdeps/x86/fpu/s_ceill.c b/sysdeps/x86/fpu/s_ceill.c new file mode 100644 index 0000000000..860dd2c960 --- /dev/null +++ b/sysdeps/x86/fpu/s_ceill.c @@ -0,0 +1,25 @@ +/* Return smallest integral value not less than argument. x86 version. + Copyright (C) 2024 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 + +#define FUNC __ceill +#define TYPE long double +#define FE_OPTION FE_UPWARD +#include "s_nearestint_387_template.c" +libm_alias_ldouble (__ceil, ceil) diff --git a/sysdeps/x86/fpu/s_nearestint_387_template.c b/sysdeps/x86/fpu/s_nearestint_387_template.c new file mode 100644 index 0000000000..95fca93f87 --- /dev/null +++ b/sysdeps/x86/fpu/s_nearestint_387_template.c @@ -0,0 +1,36 @@ +/* Nearest integet template for x86. + Copyright (C) 2024 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 + . */ + +#define NO_MATH_REDIRECT +#include +#include + +TYPE +FUNC (TYPE x) +{ + fenv_t fenv; + TYPE r; + + libc_feholdexcept_setround_387 (&fenv, FE_OPTION); + asm volatile ("frndint" : "=t" (r) : "0" (x)); + /* Preserve "invalid" exceptions from sNaN input. */ + fenv.__status_word |= libc_fetestexcept_387 (FE_INVALID); + libc_fesetenv_387 (&fenv); + + return r; +} diff --git a/sysdeps/x86_64/fpu/s_ceill.S b/sysdeps/x86_64/fpu/s_ceill.S deleted file mode 100644 index 16dbecd56d..0000000000 --- a/sysdeps/x86_64/fpu/s_ceill.S +++ /dev/null @@ -1,34 +0,0 @@ -/* - * Public domain. - */ - -#include -#include - - -ENTRY(__ceill) - fldt 8(%rsp) - - fnstenv -28(%rsp) /* store fpu environment */ - - /* We use here %edx although only the low 1 bits are defined. - But none of the operations should care and they are faster - than the 16 bit operations. */ - movl $0x0800,%edx /* round towards +oo */ - orl -28(%rsp),%edx - andl $0xfbff,%edx - movl %edx,-32(%rsp) - fldcw -32(%rsp) /* load modified control word */ - - frndint /* round */ - - /* Preserve "invalid" exceptions from sNaN input. */ - fnstsw - andl $0x1, %eax - orl %eax, -24(%rsp) - - fldenv -28(%rsp) /* restore original environment */ - - ret -END (__ceill) -libm_alias_ldouble (__ceil, ceil)