1/* 2 * Copyright (C) 1999-2000 Hewlett-Packard Co 3 * Copyright (C) 1999-2000 David Mosberger-Tang <davidm@hpl.hp.com> 4 * 5 * 64-bit integer division. 6 * 7 * This code is based on the application note entitled "Divide, Square Root 8 * and Remainder Algorithms for the IA-64 Architecture". This document 9 * is available as Intel document number 248725-002 or via the web at 10 * http://developer.intel.com/software/opensource/numerics/ 11 * 12 * For more details on the theory behind these algorithms, see "IA-64 13 * and Elementary Functions" by Peter Markstein; HP Professional Books 14 * (http://www.goodreads.com/book/show/2019887.Ia_64_and_Elementary_Functions) 15 */ 16 17#include <asm/asmmacro.h> 18#include <asm/export.h> 19 20#ifdef MODULO 21# define OP mod 22#else 23# define OP div 24#endif 25 26#ifdef UNSIGNED 27# define SGN u 28# define INT_TO_FP(a,b) fcvt.xuf.s1 a=b 29# define FP_TO_INT(a,b) fcvt.fxu.trunc.s1 a=b 30#else 31# define SGN 32# define INT_TO_FP(a,b) fcvt.xf a=b 33# define FP_TO_INT(a,b) fcvt.fx.trunc.s1 a=b 34#endif 35 36#define PASTE1(a,b) a##b 37#define PASTE(a,b) PASTE1(a,b) 38#define NAME PASTE(PASTE(__,SGN),PASTE(OP,di3)) 39 40GLOBAL_ENTRY(NAME) 41 .regstk 2,0,0,0 42 // Transfer inputs to FP registers. 43 setf.sig f8 = in0 44 setf.sig f9 = in1 45 ;; 46 // Convert the inputs to FP, to avoid FP software-assist faults. 47 INT_TO_FP(f8, f8) 48 INT_TO_FP(f9, f9) 49 ;; 50 frcpa.s1 f11, p6 = f8, f9 // y0 = frcpa(b) 51 ;; 52(p6) fmpy.s1 f7 = f8, f11 // q0 = a*y0 53(p6) fnma.s1 f6 = f9, f11, f1 // e0 = -b*y0 + 1 54 ;; 55(p6) fma.s1 f10 = f7, f6, f7 // q1 = q0*e0 + q0 56(p6) fmpy.s1 f7 = f6, f6 // e1 = e0*e0 57 ;; 58#ifdef MODULO 59 sub in1 = r0, in1 // in1 = -b 60#endif 61(p6) fma.s1 f10 = f10, f7, f10 // q2 = q1*e1 + q1 62(p6) fma.s1 f6 = f11, f6, f11 // y1 = y0*e0 + y0 63 ;; 64(p6) fma.s1 f6 = f6, f7, f6 // y2 = y1*e1 + y1 65(p6) fnma.s1 f7 = f9, f10, f8 // r = -b*q2 + a 66 ;; 67#ifdef MODULO 68 setf.sig f8 = in0 // f8 = a 69 setf.sig f9 = in1 // f9 = -b 70#endif 71(p6) fma.s1 f11 = f7, f6, f10 // q3 = r*y2 + q2 72 ;; 73 FP_TO_INT(f11, f11) // q = trunc(q3) 74 ;; 75#ifdef MODULO 76 xma.l f11 = f11, f9, f8 // r = q*(-b) + a 77 ;; 78#endif 79 getf.sig r8 = f11 // transfer result to result register 80 br.ret.sptk.many rp 81END(NAME) 82EXPORT_SYMBOL(NAME) 83