From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 6973 invoked by alias); 21 Apr 2012 04:45:54 -0000 Received: (qmail 6786 invoked by uid 22791); 21 Apr 2012 04:45:50 -0000 X-SWARE-Spam-Status: No, hits=-4.3 required=5.0 tests=AWL,BAYES_20,KHOP_RCVD_UNTRUST,KHOP_THREADED,RCVD_IN_DNSWL_HI,RCVD_IN_HOSTKARMA_W,TW_XF,T_RP_MATCHES_RCVD X-Spam-Check-By: sourceware.org Received: from smtp.gentoo.org (HELO smtp.gentoo.org) (140.211.166.183) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Sat, 21 Apr 2012 04:45:06 +0000 Received: from localhost.localdomain (localhost [127.0.0.1]) by smtp.gentoo.org (Postfix) with ESMTP id 68AFE1B4014 for ; Sat, 21 Apr 2012 04:34:50 +0000 (UTC) From: Mike Frysinger To: libc-ports@sourceware.org Subject: [PATCH 21/21] ia64: math: rename s_ilogbl.S to e_ilogbl.S Date: Sat, 21 Apr 2012 04:45:00 -0000 Message-Id: <1334982948-12167-18-git-send-email-vapier@gentoo.org> In-Reply-To: <1334982948-12167-1-git-send-email-vapier@gentoo.org> References: <1334982921-12113-1-git-send-email-vapier@gentoo.org> <1334982948-12167-1-git-send-email-vapier@gentoo.org> X-IsSubscribed: yes Mailing-List: contact libc-ports-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Post: List-Help: , Sender: libc-ports-owner@sourceware.org X-SW-Source: 2012-04/txt/msg00114.txt.bz2 The main tree renamed these files as well as the symbols. Update ia64 accordingly. Signed-off-by: Mike Frysinger 2012-04-17 Mike Frysinger * sysdeps/ia64/fpu/s_ilogbl.S: Move to ... * sysdeps/ia64/fpu/e_ilogbl.S: ... here. (__ieee754_ilogbl): Rename from ilogbl. --- sysdeps/ia64/fpu/e_ilogbl.S | 268 +++++++++++++++++++++++++++++++++++++++++++ sysdeps/ia64/fpu/s_ilogbl.S | 268 ------------------------------------------- 2 files changed, 268 insertions(+), 268 deletions(-) create mode 100644 sysdeps/ia64/fpu/e_ilogbl.S delete mode 100644 sysdeps/ia64/fpu/s_ilogbl.S diff --git a/sysdeps/ia64/fpu/e_ilogbl.S b/sysdeps/ia64/fpu/e_ilogbl.S new file mode 100644 index 0000000..8d67d4f --- /dev/null +++ b/sysdeps/ia64/fpu/e_ilogbl.S @@ -0,0 +1,268 @@ +.file "ilogbl.s" + + +// Copyright (c) 2000 - 2003, Intel Corporation +// All rights reserved. +// +// Contributed 2000 by the Intel Numerics Group, Intel Corporation +// +// 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 +//============================================================== +// 02/03/00 Initial version +// 05/26/00 Fix bug when x a double-extended denormal; +// if x=0 call error routine, per C9X +// 08/15/00 Bundle added after call to __libm_error_support to properly +// set [the previously overwritten] GR_Parameter_RESULT. +// 01/20/01 Fixed result for x=0 +// 05/20/02 Cleaned up namespace and sf0 syntax +// 01/20/03 Improved performance +// +// API +//============================================================== +// int ilogbl( long double x ); +// +// Overview of operation +//============================================================== +// The ilogbl function extracts the exponent of x as an integer +// and returns it in r8 +// +// ilogbl is similar to logbl but differs in the following ways: +// +-inf +// ilogbl: returns INT_MAX +// logbl: returns +inf +// Nan returns FP_LOGBNAN (which is either INT_MAX or INT_MIN) +// ilogbl: returns INT_MAX (7fffffff) +// logbl: returns QNAN (quietized SNAN) +// 0 returns FP_ILOGB0 (which is either INT_MIN or -INT_MAX) +// ilogbl: returns -INT_MAX (80000001) +// logbl: returns -inf, raises the divide-by-zero exception, +// and calls libm_error_support to set domain error +// +// Registers used +//============================================================== +// general registers used: +// r26 -> r39 +// r36 -> r39 used as parameters to error path +// +// predicate registers used: +// p6 -> p10 +// floating-point registers used: +// f9, f10, f11 +// f8, input + +rExpBias = r26 +rExpMask = r27 +rSignexp_x = r28 +rExp_x = r29 +rIntMax = r30 +rExp_2to64 = r31 + +GR_SAVE_PFS = r32 +rTrialResult = r33 +GR_SAVE_B0 = r34 +GR_SAVE_GP = r35 + +GR_Parameter_X = r36 +GR_Parameter_Y = r37 +GR_Parameter_RESULT = r38 +GR_Parameter_TAG = r39 + +fTmp = f9 +fNorm_x = f10 +f2to64 = f11 + +.section .text +GLOBAL_LIBM_ENTRY(__ieee754_ilogbl) + +// X NORMAL +// TrueExp_x = exp(f8) - 0xffff +// r8 = TrueExp_x +{ .mfi + getf.exp rSignexp_x = f8 + fclass.m p8,p0 = f8, 0x0b // Test for x unorm + mov rExpBias = 0xffff // Exponent bias +} +{ .mfi + nop.m 0 + fnorm.s1 fNorm_x = f8 + mov rExpMask = 0x1ffff // Exponent mask +} +;; + +// Form signexp of 2^64 in case need to scale denormal +{ .mfb + mov rExp_2to64 = 0x1003f + fclass.m p6,p9 = f8, 0x1e3 // Test x natval, nan, inf +(p8) br.cond.spnt ILOGB_DENORM // Branch if x unorm +} +;; + +ILOGB_COMMON: +// Return here from ILOGB_DENORM +{ .mfi + and rExp_x = rSignexp_x, rExpMask // Get biased exponent + fclass.m p7,p10 = f8, 0x07 // Test x zero + nop.i 0 +} +{ .mlx + nop.m 0 + movl rIntMax = 0x000000007fffffff // Form INT_MAX +} +;; + +.pred.rel "mutex",p6,p9 +{ .mfi +(p9) sub r8 = rExp_x, rExpBias // Get true exponent for normal path +(p6) fma.s0 fTmp = f8, f8, f0 // Dummy to set Invalid flag +(p6) mov r8 = rIntMax // If nan, inf, return INT_MAX +} +{ .mbb + nop.m 0 +(p7) br.cond.spnt ILOGB_ZERO // Branch if x zero +(p10) br.ret.sptk b0 // Exit if x not zero +} +;; + + +ILOGB_DENORM: +// Form 2^64 in case need to scale denormal +// Check to see if double-extended denormal +{ .mfi + setf.exp f2to64 = rExp_2to64 + fclass.m p8,p0 = fNorm_x, 0x0b + nop.i 0 +} +;; + +{ .mfi + nop.m 0 + fcmp.eq.s0 p7,p0 = f8, f0 // Dummy op to set denormal flag + nop.i 0 +} +;; + +// If double-extended denormal add 64 to exponent bias for scaling +// If double-extended denormal form x * 2^64 which is normal +{ .mfi +(p8) add rExpBias = 64, rExpBias +(p8) fmpy.s1 fNorm_x = fNorm_x, f2to64 + nop.i 0 +} +;; + +// Logic is the same as normal path but use normalized input +{ .mib + getf.exp rSignexp_x = fNorm_x + nop.i 0 + br.cond.sptk ILOGB_COMMON // Return to main path +} +;; + +ILOGB_ZERO: +// Here if x zero +// Return INT_MIN, call error support + +{ .mlx + alloc r32=ar.pfs,1,3,4,0 + movl rTrialResult = 0x0000000080000000 +} +{ .mib + mov GR_Parameter_TAG = 156 // Error code + nop.i 0 + br.cond.sptk __libm_error_region // Call error support +} +;; + +GLOBAL_LIBM_END(__ieee754_ilogbl) + + +LOCAL_LIBM_ENTRY(__libm_error_region) +.prologue + +{ .mfi + add GR_Parameter_Y=-32,sp // Parameter 2 value + nop.f 0 +.save ar.pfs,GR_SAVE_PFS + mov GR_SAVE_PFS=ar.pfs // Save ar.pfs +} +{ .mfi +.fframe 64 + add sp=-64,sp // Create new stack + nop.f 0 + mov GR_SAVE_GP=gp // Save gp +};; + +{ .mmi + stfe [GR_Parameter_Y] = f0,16 // STORE Parameter 2 on stack + add GR_Parameter_X = 16,sp // Parameter 1 address +.save b0, GR_SAVE_B0 + mov GR_SAVE_B0=b0 // Save b0 +};; + +.body +{ .mib + stfe [GR_Parameter_X] = f8 // STORE Parameter 1 on stack + add GR_Parameter_RESULT = 0,GR_Parameter_Y // Parameter 3 address + nop.b 0 +} +{ .mib + stfe [GR_Parameter_Y] = f9 // Store Parameter 3 on stack + add GR_Parameter_Y = -16,GR_Parameter_Y + br.call.sptk b0=__libm_error_support# // Call error handling function +};; + +{ .mmi + add GR_Parameter_RESULT = 48,sp + nop.m 0 + nop.i 0 +};; + +{ .mmi + mov r8 = rTrialResult +.restore sp + add sp = 64,sp // Restore stack pointer + mov b0 = GR_SAVE_B0 // Restore return address +};; + +{ .mib + mov gp = GR_SAVE_GP // Restore gp + mov ar.pfs = GR_SAVE_PFS // Restore ar.pfs + br.ret.sptk b0 +};; + +LOCAL_LIBM_END(__libm_error_region) + + +.type __libm_error_support#,@function +.global __libm_error_support# diff --git a/sysdeps/ia64/fpu/s_ilogbl.S b/sysdeps/ia64/fpu/s_ilogbl.S deleted file mode 100644 index 8a6c9dc..0000000 --- a/sysdeps/ia64/fpu/s_ilogbl.S +++ /dev/null @@ -1,268 +0,0 @@ -.file "ilogbl.s" - - -// Copyright (c) 2000 - 2003, Intel Corporation -// All rights reserved. -// -// Contributed 2000 by the Intel Numerics Group, Intel Corporation -// -// 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 -//============================================================== -// 02/03/00 Initial version -// 05/26/00 Fix bug when x a double-extended denormal; -// if x=0 call error routine, per C9X -// 08/15/00 Bundle added after call to __libm_error_support to properly -// set [the previously overwritten] GR_Parameter_RESULT. -// 01/20/01 Fixed result for x=0 -// 05/20/02 Cleaned up namespace and sf0 syntax -// 01/20/03 Improved performance -// -// API -//============================================================== -// int ilogbl( long double x ); -// -// Overview of operation -//============================================================== -// The ilogbl function extracts the exponent of x as an integer -// and returns it in r8 -// -// ilogbl is similar to logbl but differs in the following ways: -// +-inf -// ilogbl: returns INT_MAX -// logbl: returns +inf -// Nan returns FP_LOGBNAN (which is either INT_MAX or INT_MIN) -// ilogbl: returns INT_MAX (7fffffff) -// logbl: returns QNAN (quietized SNAN) -// 0 returns FP_ILOGB0 (which is either INT_MIN or -INT_MAX) -// ilogbl: returns -INT_MAX (80000001) -// logbl: returns -inf, raises the divide-by-zero exception, -// and calls libm_error_support to set domain error -// -// Registers used -//============================================================== -// general registers used: -// r26 -> r39 -// r36 -> r39 used as parameters to error path -// -// predicate registers used: -// p6 -> p10 -// floating-point registers used: -// f9, f10, f11 -// f8, input - -rExpBias = r26 -rExpMask = r27 -rSignexp_x = r28 -rExp_x = r29 -rIntMax = r30 -rExp_2to64 = r31 - -GR_SAVE_PFS = r32 -rTrialResult = r33 -GR_SAVE_B0 = r34 -GR_SAVE_GP = r35 - -GR_Parameter_X = r36 -GR_Parameter_Y = r37 -GR_Parameter_RESULT = r38 -GR_Parameter_TAG = r39 - -fTmp = f9 -fNorm_x = f10 -f2to64 = f11 - -.section .text -GLOBAL_LIBM_ENTRY(ilogbl) - -// X NORMAL -// TrueExp_x = exp(f8) - 0xffff -// r8 = TrueExp_x -{ .mfi - getf.exp rSignexp_x = f8 - fclass.m p8,p0 = f8, 0x0b // Test for x unorm - mov rExpBias = 0xffff // Exponent bias -} -{ .mfi - nop.m 0 - fnorm.s1 fNorm_x = f8 - mov rExpMask = 0x1ffff // Exponent mask -} -;; - -// Form signexp of 2^64 in case need to scale denormal -{ .mfb - mov rExp_2to64 = 0x1003f - fclass.m p6,p9 = f8, 0x1e3 // Test x natval, nan, inf -(p8) br.cond.spnt ILOGB_DENORM // Branch if x unorm -} -;; - -ILOGB_COMMON: -// Return here from ILOGB_DENORM -{ .mfi - and rExp_x = rSignexp_x, rExpMask // Get biased exponent - fclass.m p7,p10 = f8, 0x07 // Test x zero - nop.i 0 -} -{ .mlx - nop.m 0 - movl rIntMax = 0x000000007fffffff // Form INT_MAX -} -;; - -.pred.rel "mutex",p6,p9 -{ .mfi -(p9) sub r8 = rExp_x, rExpBias // Get true exponent for normal path -(p6) fma.s0 fTmp = f8, f8, f0 // Dummy to set Invalid flag -(p6) mov r8 = rIntMax // If nan, inf, return INT_MAX -} -{ .mbb - nop.m 0 -(p7) br.cond.spnt ILOGB_ZERO // Branch if x zero -(p10) br.ret.sptk b0 // Exit if x not zero -} -;; - - -ILOGB_DENORM: -// Form 2^64 in case need to scale denormal -// Check to see if double-extended denormal -{ .mfi - setf.exp f2to64 = rExp_2to64 - fclass.m p8,p0 = fNorm_x, 0x0b - nop.i 0 -} -;; - -{ .mfi - nop.m 0 - fcmp.eq.s0 p7,p0 = f8, f0 // Dummy op to set denormal flag - nop.i 0 -} -;; - -// If double-extended denormal add 64 to exponent bias for scaling -// If double-extended denormal form x * 2^64 which is normal -{ .mfi -(p8) add rExpBias = 64, rExpBias -(p8) fmpy.s1 fNorm_x = fNorm_x, f2to64 - nop.i 0 -} -;; - -// Logic is the same as normal path but use normalized input -{ .mib - getf.exp rSignexp_x = fNorm_x - nop.i 0 - br.cond.sptk ILOGB_COMMON // Return to main path -} -;; - -ILOGB_ZERO: -// Here if x zero -// Return INT_MIN, call error support - -{ .mlx - alloc r32=ar.pfs,1,3,4,0 - movl rTrialResult = 0x0000000080000000 -} -{ .mib - mov GR_Parameter_TAG = 156 // Error code - nop.i 0 - br.cond.sptk __libm_error_region // Call error support -} -;; - -GLOBAL_LIBM_END(ilogbl) - - -LOCAL_LIBM_ENTRY(__libm_error_region) -.prologue - -{ .mfi - add GR_Parameter_Y=-32,sp // Parameter 2 value - nop.f 0 -.save ar.pfs,GR_SAVE_PFS - mov GR_SAVE_PFS=ar.pfs // Save ar.pfs -} -{ .mfi -.fframe 64 - add sp=-64,sp // Create new stack - nop.f 0 - mov GR_SAVE_GP=gp // Save gp -};; - -{ .mmi - stfe [GR_Parameter_Y] = f0,16 // STORE Parameter 2 on stack - add GR_Parameter_X = 16,sp // Parameter 1 address -.save b0, GR_SAVE_B0 - mov GR_SAVE_B0=b0 // Save b0 -};; - -.body -{ .mib - stfe [GR_Parameter_X] = f8 // STORE Parameter 1 on stack - add GR_Parameter_RESULT = 0,GR_Parameter_Y // Parameter 3 address - nop.b 0 -} -{ .mib - stfe [GR_Parameter_Y] = f9 // Store Parameter 3 on stack - add GR_Parameter_Y = -16,GR_Parameter_Y - br.call.sptk b0=__libm_error_support# // Call error handling function -};; - -{ .mmi - add GR_Parameter_RESULT = 48,sp - nop.m 0 - nop.i 0 -};; - -{ .mmi - mov r8 = rTrialResult -.restore sp - add sp = 64,sp // Restore stack pointer - mov b0 = GR_SAVE_B0 // Restore return address -};; - -{ .mib - mov gp = GR_SAVE_GP // Restore gp - mov ar.pfs = GR_SAVE_PFS // Restore ar.pfs - br.ret.sptk b0 -};; - -LOCAL_LIBM_END(__libm_error_region) - - -.type __libm_error_support#,@function -.global __libm_error_support# -- 1.7.8.5