/*
 * CDDL HEADER START
 *
 * The contents of this file are subject to the terms of the
 * Common Development and Distribution License (the "License").
 * You may not use this file except in compliance with the License.
 *
 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
 * or http://www.opensolaris.org/os/licensing.
 * See the License for the specific language governing permissions
 * and limitations under the License.
 *
 * When distributing Covered Code, include this CDDL HEADER in each
 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
 * If applicable, add the following below this CDDL HEADER, with the
 * fields enclosed by brackets "[]" replaced with your own identifying
 * information: Portions Copyright [yyyy] [name of copyright owner]
 *
 * CDDL HEADER END
 */
/*
 * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
 */
/*
 * Copyright 2005 Sun Microsystems, Inc.  All rights reserved.
 * Use is subject to license terms.
 */

#pragma weak __atan2f = atan2f

#include "libm.h"

#if defined(__i386) && !defined(__amd64)
extern int __swapRP(int);
#endif

/*
 * For i = 0, ..., 192, let x[i] be the double precision number whose
 * high order 32 bits are 0x3f900000 + (i << 16) and whose low order
 * 32 bits are zero.  Then TBL[i] := atan(x[i]) to double precision.
 */

static const double TBL[] = {
        1.56237286204768313e-02,
        1.66000375562312640e-02,
        1.75763148444955872e-02,
        1.85525586258889763e-02,
        1.95287670414137082e-02,
        2.05049382324763683e-02,
        2.14810703409090559e-02,
        2.24571615089905717e-02,
        2.34332098794675855e-02,
        2.44092135955758099e-02,
        2.53851708010611396e-02,
        2.63610796402007873e-02,
        2.73369382578244127e-02,
        2.83127447993351995e-02,
        2.92884974107309737e-02,
        3.02641942386252458e-02,
        3.12398334302682774e-02,
        3.31909314971115949e-02,
        3.51417768027967800e-02,
        3.70923545503918164e-02,
        3.90426499551669928e-02,
        4.09926482452637811e-02,
        4.29423346623621707e-02,
        4.48916944623464972e-02,
        4.68407129159696539e-02,
        4.87893753095156174e-02,
        5.07376669454602178e-02,
        5.26855731431300420e-02,
        5.46330792393594777e-02,
        5.65801705891457105e-02,
        5.85268325663017702e-02,
        6.04730505641073168e-02,
        6.24188099959573500e-02,
        6.63088949198234884e-02,
        7.01969710718705203e-02,
        7.40829225490337306e-02,
        7.79666338315423008e-02,
        8.18479898030765457e-02,
        8.57268757707448092e-02,
        8.96031774848717461e-02,
        9.34767811585894698e-02,
        9.73475734872236709e-02,
        1.01215441667466668e-01,
        1.05080273416329528e-01,
        1.08941956989865793e-01,
        1.12800381201659389e-01,
        1.16655435441069349e-01,
        1.20507009691224562e-01,
        1.24354994546761438e-01,
        1.32039761614638762e-01,
        1.39708874289163648e-01,
        1.47361481088651630e-01,
        1.54996741923940973e-01,
        1.62613828597948568e-01,
        1.70211925285474408e-01,
        1.77790228992676075e-01,
        1.85347949995694761e-01,
        1.92884312257974672e-01,
        2.00398553825878512e-01,
        2.07889927202262986e-01,
        2.15357699697738048e-01,
        2.22801153759394521e-01,
        2.30219587276843718e-01,
        2.37612313865471242e-01,
        2.44978663126864143e-01,
        2.59629629408257512e-01,
        2.74167451119658789e-01,
        2.88587361894077410e-01,
        3.02884868374971417e-01,
        3.17055753209147029e-01,
        3.31096076704132103e-01,
        3.45002177207105132e-01,
        3.58770670270572245e-01,
        3.72398446676754202e-01,
        3.85882669398073752e-01,
        3.99220769575252543e-01,
        4.12410441597387323e-01,
        4.25449637370042266e-01,
        4.38336559857957830e-01,
        4.51069655988523499e-01,
        4.63647609000806094e-01,
        4.88333951056405535e-01,
        5.12389460310737732e-01,
        5.35811237960463704e-01,
        5.58599315343562441e-01,
        5.80756353567670414e-01,
        6.02287346134964152e-01,
        6.23199329934065904e-01,
        6.43501108793284371e-01,
        6.63202992706093286e-01,
        6.82316554874748071e-01,
        7.00854407884450192e-01,
        7.18829999621624527e-01,
        7.36257428981428097e-01,
        7.53151280962194414e-01,
        7.69526480405658297e-01,
        7.85398163397448279e-01,
        8.15691923316223422e-01,
        8.44153986113171051e-01,
        8.70903457075652976e-01,
        8.96055384571343927e-01,
        9.19719605350416858e-01,
        9.42000040379463610e-01,
        9.62994330680936206e-01,
        9.82793723247329054e-01,
        1.00148313569423464e+00,
        1.01914134426634972e+00,
        1.03584125300880014e+00,
        1.05165021254837376e+00,
        1.06663036531574362e+00,
        1.08083900054116833e+00,
        1.09432890732118993e+00,
        1.10714871779409041e+00,
        1.13095374397916038e+00,
        1.15257199721566761e+00,
        1.17227388112847630e+00,
        1.19028994968253166e+00,
        1.20681737028525249e+00,
        1.22202532321098967e+00,
        1.23605948947808186e+00,
        1.24904577239825443e+00,
        1.26109338225244039e+00,
        1.27229739520871732e+00,
        1.28274087974427076e+00,
        1.29249666778978534e+00,
        1.30162883400919616e+00,
        1.31019393504755555e+00,
        1.31824205101683711e+00,
        1.32581766366803255e+00,
        1.33970565959899957e+00,
        1.35212738092095464e+00,
        1.36330010035969384e+00,
        1.37340076694501589e+00,
        1.38257482149012589e+00,
        1.39094282700241845e+00,
        1.39860551227195762e+00,
        1.40564764938026987e+00,
        1.41214106460849531e+00,
        1.41814699839963154e+00,
        1.42371797140649403e+00,
        1.42889927219073276e+00,
        1.43373015248470903e+00,
        1.43824479449822262e+00,
        1.44247309910910193e+00,
        1.44644133224813509e+00,
        1.45368758222803240e+00,
        1.46013910562100091e+00,
        1.46591938806466282e+00,
        1.47112767430373470e+00,
        1.47584462045214027e+00,
        1.48013643959415142e+00,
        1.48405798811891154e+00,
        1.48765509490645531e+00,
        1.49096634108265924e+00,
        1.49402443552511865e+00,
        1.49685728913695626e+00,
        1.49948886200960629e+00,
        1.50193983749385196e+00,
        1.50422816301907281e+00,
        1.50636948736934317e+00,
        1.50837751679893928e+00,
        1.51204050407917401e+00,
        1.51529782154917969e+00,
        1.51821326518395483e+00,
        1.52083793107295384e+00,
        1.52321322351791322e+00,
        1.52537304737331958e+00,
        1.52734543140336587e+00,
        1.52915374769630819e+00,
        1.53081763967160667e+00,
        1.53235373677370856e+00,
        1.53377621092096650e+00,
        1.53509721411557254e+00,
        1.53632722579538861e+00,
        1.53747533091664934e+00,
        1.53854944435964280e+00,
        1.53955649336462841e+00,
        1.54139303859089161e+00,
        1.54302569020147562e+00,
        1.54448660954197448e+00,
        1.54580153317597646e+00,
        1.54699130060982659e+00,
        1.54807296595325550e+00,
        1.54906061995310385e+00,
        1.54996600675867957e+00,
        1.55079899282174605e+00,
        1.55156792769518947e+00,
        1.55227992472688747e+00,
        1.55294108165534417e+00,
        1.55355665560036682e+00,
        1.55413120308095598e+00,
        1.55466869295126031e+00,
        1.55517259817441977e+00,
};

static const double
        pio4    =  7.8539816339744827900e-01,
        pio2    =  1.5707963267948965580e+00,
        negpi   = -3.1415926535897931160e+00,
        q1      = -3.3333333333296428046e-01,
        q2      =  1.9999999186853752618e-01,
        zero    =  0.0;

static const float two24 = 16777216.0;

float
atan2f(float fy, float fx)
{
        double  a, t, s, dbase;
        float   x, y, base;
        int     i, k, hx, hy, ix, iy, sign;
#if defined(__i386) && !defined(__amd64)
        int     rp;
#endif

        iy = *(int *)&fy;
        ix = *(int *)&fx;
        hy = iy & ~0x80000000;
        hx = ix & ~0x80000000;

        sign = 0;
        if (hy > hx) {
                x = fy;
                y = fx;
                i = hx;
                hx = hy;
                hy = i;
                if (iy < 0) {
                        x = -x;
                        sign = 1;
                }
                if (ix < 0) {
                        y = -y;
                        a = pio2;
                } else {
                        a = -pio2;
                        sign = 1 - sign;
                }
        } else {
                y = fy;
                x = fx;
                if (iy < 0) {
                        y = -y;
                        sign = 1;
                }
                if (ix < 0) {
                        x = -x;
                        a = negpi;
                        sign = 1 - sign;
                } else {
                        a = zero;
                }
        }

        if (hx >= 0x7f800000 || hx - hy >= 0x0c800000) {
                if (hx >= 0x7f800000) {
                        if (hx > 0x7f800000) /* nan */
                                return (x * y);
                        else if (hy >= 0x7f800000)
                                a += pio4;
                } else if ((int)a == 0) {
                        a = (double)y / x;
                }
                return ((float)((sign)? -a : a));
        }

        if (hy < 0x00800000) {
                if (hy == 0)
                        return ((float)((sign)? -a : a));
                /* scale subnormal y */
                y *= two24;
                x *= two24;
                hy = *(int *)&y;
                hx = *(int *)&x;
        }

#if defined(__i386) && !defined(__amd64)
        rp = __swapRP(fp_extended);
#endif
        k = (hy - hx + 0x3f800000) & 0xfff80000;
        if (k >= 0x3c800000) {  /* |y/x| >= 1/64 */
                *(int *)&base = k;
                k = (k - 0x3c800000) >> 19;
                a += TBL[k];
        } else {
                /*
                 * For some reason this is faster on USIII than just
                 * doing t = y/x in this case.
                 */
                *(int *)&base = 0;
        }
        dbase = (double)base;
        t = (y - x * dbase) / (x + y * dbase);
        s = t * t;
        a = (a + t) + t * s * (q1 + s * q2);
#if defined(__i386) && !defined(__amd64)
        if (rp != fp_extended)
                (void) __swapRP(rp);
#endif
        return ((float)((sign)? -a : a));
}