Enhance the base/rom extension to generate the symbol file of the rom built.
The symbol file is placed in epoc32/rom/<baseport_name>, along with the rom log and final oby file.
// Copyright (c) 1995-2009 Nokia Corporation and/or its subsidiary(-ies).
// All rights reserved.
// This component and the accompanying materials are made available
// under the terms of the License "Eclipse Public License v1.0"
// which accompanies this distribution, and is available
// at the URL "http://www.eclipse.org/legal/epl-v10.html".
//
// Initial Contributors:
// Nokia Corporation - initial contribution.
//
// Contributors:
//
// Description:
// e32\euser\maths\um_atan.cpp
// Floating point arc tangent
//
//
#include "um_std.h"
#if defined(__USE_VFP_MATH) && !defined(__CPU_HAS_VFP)
#error __USE_VFP_MATH was defined but not __CPU_HAS_VFP - impossible combination, check variant.mmh
#endif
#ifndef __USE_VFP_MATH
LOCAL_D const TUint32 ArctanCoeffs[] =
{
0x00000000,0x80000000,0x7FFF0000, // polynomial approximation to arctan(x)
0xAA84D6EE,0xAAAAAAAA,0x7FFD0001, // for -(sqr2-1) <= x <= (sqr2-1)
0x89C77453,0xCCCCCCCC,0x7FFC0000,
0xEBC0261C,0x9249247B,0x7FFC0001,
0x940BC4DB,0xE38E3121,0x7FFB0000,
0x141C32F1,0xBA2DBF36,0x7FFB0001,
0xA90615E7,0x9D7C807E,0x7FFB0000,
0x1C632E93,0x87F6A873,0x7FFB0001,
0x310FCFFD,0xE8BE5D0A,0x7FFA0000,
0x92289F15,0xB17B930B,0x7FFA0001,
0x546FE7CE,0xABDE562D,0x7FF90000
};
LOCAL_D const TUint32 Sqr2m1data[] = {0xE7799211,0xD413CCCF,0x7FFD0000}; // sqr2-1
LOCAL_D const TUint32 Sqr2p1data[] = {0xFCEF3242,0x9A827999,0x80000000}; // sqr2+1
LOCAL_D const TUint32 Onedata[] = {0x00000000,0x80000000,0x7FFF0000}; // 1.0
LOCAL_D const TUint32 PiBy8data[] = {0x2168C235,0xC90FDAA2,0x7FFD0000}; // pi/8
LOCAL_D const TUint32 PiBy2data[] = {0x2168C235,0xC90FDAA2,0x7FFF0000}; // pi/2
LOCAL_D const TUint32 ThreePiBy8data[] = {0x990E91A8,0x96CBE3F9,0x7FFF0000}; // 3*pi/8
LOCAL_C void Arctan(TRealX& y, TRealX& x)
{
// Calculate arctan(x), write result to y
// Algorithm:
// If x>1, replace x with 1/x and subtract result from pi/2
// ( use identity tan(pi/2-x)=1/tan(x) )
// If x>sqr(2)-1, replace x with (x-(sqr(2)-1))/(1-(sqr2-1)x)
// ( use identity tan(x-a)=(tanx-tana)/(1-tana.tanx)
// where a=pi/8, tan a = sqr2-1
// and add pi/8 to result
// Use polynomial approximation to calculate arctan(x) for
// x in the interval [0,sqr2-1]
const TRealX& Sqr2m1 = *(const TRealX*)Sqr2m1data;
const TRealX& Sqr2p1 = *(const TRealX*)Sqr2p1data;
const TRealX& One = *(const TRealX*)Onedata;
const TRealX& PiBy8 = *(const TRealX*)PiBy8data;
const TRealX& PiBy2 = *(const TRealX*)PiBy2data;
const TRealX& ThreePiBy8 = *(const TRealX*)ThreePiBy8data;
TInt section=0;
TInt8 sign=x.iSign;
x.iSign=0;
if (x>Sqr2p1)
{
x=One/x;
section=3;
}
else if (x>One)
{
x=(One-Sqr2m1*x)/(x+Sqr2m1);
section=2;
}
else if (x>Sqr2m1)
{
x=(x-Sqr2m1)/(One+Sqr2m1*x);
section=1;
}
Math::PolyX(y,x*x,10,(const TRealX*)ArctanCoeffs);
y*=x;
if (section==1)
y+=PiBy8;
else if (section==2)
y=ThreePiBy8-y;
else if (section==3)
y=PiBy2-y;
y.iSign=sign;
}
EXPORT_C TInt Math::ATan(TReal& aTrg, const TReal& aSrc)
/**
Calculates the principal value of the inverse tangent of a number.
@param aTrg A reference containing the result in radians,
a value between -pi/2 and +pi/2.
@param aSrc The argument of the arctan function,
a value between +infinity and +infinity.
@return KErrNone if successful, otherwise another of
the system-wide error codes.
*/
{
TRealX x;
TInt r=x.Set(aSrc);
if (r==KErrNone)
{
TRealX y;
Arctan(y,x);
return y.GetTReal(aTrg);
}
if (r==KErrArgument)
{
SetNaN(aTrg);
return KErrArgument;
}
aTrg=KPiBy2; // arctan(+/- infinity) = +/- pi/2
if (x.iSign&1)
aTrg=-aTrg;
return KErrNone;
}
LOCAL_D const TUint32 Pidata[] = {0x2168C235,0xC90FDAA2,0x80000000};
LOCAL_D const TUint32 PiBy4data[] = {0x2168C235,0xC90FDAA2,0x7FFE0000};
LOCAL_D const TUint32 MinusPiBy4data[] = {0x2168C235,0xC90FDAA2,0x7FFE0001};
LOCAL_D const TUint32 ThreePiBy4data[] = {0x990E91A8,0x96CBE3F9,0x80000000};
LOCAL_D const TUint32 MinusThreePiBy4data[] = {0x990E91A8,0x96CBE3F9,0x80000001};
LOCAL_D const TUint32 Zerodata[] = {0x00000000,0x00000000,0x00000000};
EXPORT_C TInt Math::ATan(TReal &aTrg,const TReal &aY,const TReal &aX)
/**
Calculates the angle between the x-axis and a line drawn from the origin
to a point represented by its (x,y) co-ordinates.
The co-ordinates are passed as arguments to the function.
This function returns the same result as arctan(y/x), but:
1. it adds +/-pi to the result, if x is negative
2. it sets the result to +/-pi/2, if x is zero but y is non-zero.
@param aTrg A reference containing the result in radians,
a value between -pi exclusive and +pi inclusive.
@param aY The y argument of the arctan(y/x) function.
@param aX The x argument of the arctan(y/x) function.
@return KErrNone if successful, otherwise another of
the system-wide error codes.
*/
{
const TRealX& Zero=*(const TRealX*)Zerodata;
const TRealX& Pi=*(const TRealX*)Pidata;
const TRealX& PiBy4=*(const TRealX*)PiBy4data;
const TRealX& MinusPiBy4=*(const TRealX*)MinusPiBy4data;
const TRealX& ThreePiBy4=*(const TRealX*)ThreePiBy4data;
const TRealX& MinusThreePiBy4=*(const TRealX*)MinusThreePiBy4data;
TRealX x, y;
TInt rx=x.Set(aX);
TInt ry=y.Set(aY);
if (rx!=KErrArgument && ry!=KErrArgument)
{
if (x.iExp==0)
x.iSign=0;
TRealX q;
TInt rq=y.Div(q,x);
if (rq!=KErrArgument)
{
TRealX arg;
Arctan(arg,q);
if (x<Zero)
{
if (y>=Zero)
arg+=Pi;
else
arg-=Pi;
}
aTrg=arg;
return KErrNone;
}
if (!x.IsZero())
{
// Both x and y must be infinite
TInt quadrant=((y.iSign & 1)<<1) + (x.iSign&1);
TRealX arg;
if (quadrant==0)
arg=PiBy4;
else if (quadrant==1)
arg=ThreePiBy4;
else if (quadrant==3)
arg=MinusThreePiBy4;
else
arg=MinusPiBy4;
aTrg=(TReal)arg;
return KErrNone;
}
}
SetNaN(aTrg);
return KErrArgument;
}
#else // __USE_VFP_MATH
LOCAL_D const TUint32 PiBy4data[] = {0x54442D18,0x3FE921FB};
LOCAL_D const TUint32 MinusPiBy4data[] = {0x54442D18,0xBFE921FB};
LOCAL_D const TUint32 ThreePiBy4data[] = {0x7F3321D2,0x4002D97C};
LOCAL_D const TUint32 MinusThreePiBy4data[] = {0x7F3321D2,0xC002D97C};
// definitions come from RVCT math library
extern "C" TReal atan(TReal);
extern "C" TReal atan2(TReal,TReal);
EXPORT_C TInt Math::ATan(TReal& aTrg, const TReal& aSrc)
{
aTrg = atan(aSrc);
if (Math::IsFinite(aTrg))
return KErrNone;
SetNaN(aTrg);
return KErrArgument;
}
EXPORT_C TInt Math::ATan(TReal &aTrg,const TReal &aY,const TReal &aX)
{
aTrg = atan2(aY,aX);
if (Math::IsFinite(aTrg))
return KErrNone;
// Return is a NaN, but ARM implementation returns NaN for atan(inf/inf)
// whereas implementation above returns multiples of pi/4 - fix up here
SReal64 *pY=(SReal64 *)&aY;
SReal64 *pX=(SReal64 *)&aX;
if ( pY->msm==0 && pY->lsm==0 && pY->exp==KTReal64SpecialExponent
&& pX->msm==0 && pX->lsm==0 && pX->exp==KTReal64SpecialExponent)
{
TInt quadrant=((pY->sign)<<1) + (pX->sign);
if (quadrant==0)
aTrg=*(const TReal*)PiBy4data;
else if (quadrant==1)
aTrg=*(const TReal*)ThreePiBy4data;
else if (quadrant==3)
aTrg=*(const TReal*)MinusThreePiBy4data;
else
aTrg=*(const TReal*)MinusPiBy4data;
return KErrNone;
}
// If we get here then the args weren't inf/inf so one of them must've
// been a NaN to start with
SetNaN(aTrg);
return KErrArgument;
}
#endif