From patchwork Mon Nov 4 15:27:16 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Stefan Liebler X-Patchwork-Id: 35630 Received: (qmail 22302 invoked by alias); 4 Nov 2019 16:28:41 -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 22293 invoked by uid 89); 4 Nov 2019 16:28:41 -0000 Authentication-Results: sourceware.org; auth=none X-Spam-SWARE-Status: No, score=-23.6 required=5.0 tests=AWL, BAYES_00, GIT_PATCH_0, GIT_PATCH_1, GIT_PATCH_2, GIT_PATCH_3, KAM_SHORT, RCVD_IN_DNSWL_LOW, SPF_PASS autolearn=ham version=3.3.1 spammy= X-HELO: mx0a-001b2d01.pphosted.com From: Stefan Liebler To: libc-alpha@sourceware.org Cc: Stefan Liebler Subject: [PATCH 09/17] S390: Use convert-to-fixed instruction for llrint functions. Date: Mon, 4 Nov 2019 16:27:16 +0100 In-Reply-To: <1572881244-6781-1-git-send-email-stli@linux.ibm.com> References: <1572881244-6781-1-git-send-email-stli@linux.ibm.com> x-cbid: 19110415-0016-0000-0000-000002C09A0F X-IBM-AV-DETECTION: SAVI=unused REMOTE=unused XFE=unused x-cbparentid: 19110415-0017-0000-0000-000033220BD3 Message-Id: <1572881244-6781-9-git-send-email-stli@linux.ibm.com> If compiled with z196 zarch support, the convert-to-fixed instruction is used to implement llrint, llrintf, llrintl. Otherwise the common-code implementation is used. --- sysdeps/s390/fpu/s_llrint.c | 50 +++++++++++++++++++++++++++++++++++ sysdeps/s390/fpu/s_llrintf.c | 50 +++++++++++++++++++++++++++++++++++ sysdeps/s390/fpu/s_llrintl.c | 51 ++++++++++++++++++++++++++++++++++++ 3 files changed, 151 insertions(+) create mode 100644 sysdeps/s390/fpu/s_llrint.c create mode 100644 sysdeps/s390/fpu/s_llrintf.c create mode 100644 sysdeps/s390/fpu/s_llrintl.c diff --git a/sysdeps/s390/fpu/s_llrint.c b/sysdeps/s390/fpu/s_llrint.c new file mode 100644 index 0000000000..edd796ae8c --- /dev/null +++ b/sysdeps/s390/fpu/s_llrint.c @@ -0,0 +1,50 @@ +/* llrint() - S390 version. + Copyright (C) 2019 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 + . */ + +#if defined __s390x__ && defined HAVE_S390_MIN_Z196_ZARCH_ASM_SUPPORT +/* We only support s390x as on s390 a long long int refers to a register pair + of two 4byte registers instead of a 8byte register which is produced by the + instruction. + Note: On s390 this instruction would only be used if build with -mzarch. */ +# include +# include + +long long int +__llrint (double x) +{ + long long int y; + /* The z196 zarch "convert to fixed" (cgdbra) instruction is rounding + according to current rounding mode (M3-field: 0). + First convert x with suppressed inexact exception and check if the + resulting value is beyond the target limits (indicated by cc=3; + Note: a nan is also indicated by cc=3). + If the resulting value is within the target limits, redo + without suppressing the inexact exception. */ + __asm__ ("cgdbra %0,0,%1,4 \n\t" + "jo 1f \n\t" + "cgdbra %0,0,%1,0 \n\t" + "1:" + : "=&d" (y) : "f" (x) : "cc"); + return y; +} +libm_alias_double (__llrint, llrint) + +#else +# include +#endif diff --git a/sysdeps/s390/fpu/s_llrintf.c b/sysdeps/s390/fpu/s_llrintf.c new file mode 100644 index 0000000000..3cbe7c581a --- /dev/null +++ b/sysdeps/s390/fpu/s_llrintf.c @@ -0,0 +1,50 @@ +/* llrintf() - S390 version. + Copyright (C) 2019 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 + . */ + +#if defined __s390x__ && defined HAVE_S390_MIN_Z196_ZARCH_ASM_SUPPORT +/* We only support s390x as on s390 a long long int refers to a register pair + of two 4byte registers instead of a 8byte register which is produced by the + instruction. + Note: On s390 this instruction would only be used if build with -mzarch. */ +# include +# include + +long long int +__llrintf (float x) +{ + long long int y; + /* The z196 zarch "convert to fixed" (cgebra) instruction is rounding + according to current rounding mode (M3-field: 0). + First convert x with suppressed inexact exception and check if the + resulting value is beyond the target limits (indicated by cc=3; + Note: a nan is also indicated by cc=3). + If the resulting value is within the target limits, redo + without suppressing the inexact exception. */ + __asm__ ("cgebra %0,0,%1,4 \n\t" + "jo 1f \n\t" + "cgebra %0,0,%1,0 \n\t" + "1:" + : "=&d" (y) : "f" (x) : "cc"); + return y; +} +libm_alias_float (__llrint, llrint) + +#else +# include +#endif diff --git a/sysdeps/s390/fpu/s_llrintl.c b/sysdeps/s390/fpu/s_llrintl.c new file mode 100644 index 0000000000..37eea5914f --- /dev/null +++ b/sysdeps/s390/fpu/s_llrintl.c @@ -0,0 +1,51 @@ +/* llrintl() - S390 version. + Copyright (C) 2019 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 + . */ + +#if defined __s390x__ && defined HAVE_S390_MIN_Z196_ZARCH_ASM_SUPPORT +/* We only support s390x as on s390 a long long int refers to a register pair + of two 4byte registers instead of a 8byte register which is produced by the + instruction. + Note: On s390 this instruction would only be used if build with -mzarch. */ +# include +# include +# include + +long long int +__llrintl (_Float128 x) +{ + long long int y; + /* The z196 zarch "convert to fixed" (cgxbra) instruction is rounding + according to current rounding mode (M3-field: 0). + First convert x with suppressed inexact exception and check if the + resulting value is beyond the target limits (indicated by cc=3; + Note: a nan is also indicated by cc=3). + If the resulting value is within the target limits, redo + without suppressing the inexact exception. */ + __asm__ ("cgxbra %0,0,%1,4 \n\t" + "jo 1f \n\t" + "cgxbra %0,0,%1,0 \n\t" + "1:" + : "=&d" (y) : "f" (x) : "cc"); + return y; +} +libm_alias_ldouble (__llrint, llrint) + +#else +# include +#endif