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:
// e32test\math\t_r32.cpp
// T_R32.CPP - Test routines for TReal32
// Also note that these tests do not generally include testing of special values. This is done
// in T_R96 tests and conversions are tested thoroughly, so explicit tests are unnecessary here.
// Overview:
// Test functionality of operations on 32bit real numbers.
// API Information:
// TReal32.
// Details:
// - Test the conversion from TReal to TReal32 is as expected.
// - Check addition, subtraction and multiplication of 32-bit floating point
// numbers are as expected.
// - Check division of 32-bit floating-point numbers and verify that it is
// panicked when divided by zero.
// - Test arithmetic exceptions are raised for
// - overflow error during addition, subtraction.
// - overflow, underflow errors during multiplication.
// - overflow, underflow, divided by zero errors during division.
// - overflow, underflow, invalid operation errors during conversion
// from double to float.
// - Check unary operator, equalities and inequalities operators, pre/post
// increment, decrement operators with TReal32 are as expected.
// Platforms/Drives/Compatibility:
// All
// Assumptions/Requirement/Pre-requisites:
// Failures and causes:
// Base Port information:
//
//
#include "t_math.h"
#include "t_real32.h"
#if defined(__VC32__)
// Some symbols generated by the VC++ compiler for floating point stuff.
extern "C" {
GLDEF_D TInt _adj_fdiv_m32;
}
#endif
// Data for tests from T_R32DTA.cpp
GLREF_D TReal32 addInput[];
GLREF_D TReal32 subInput[];
GLREF_D TReal32 multInput[];
GLREF_D TReal32 divInput[];
GLREF_D TReal32 unaryInput[];
GLREF_D TReal32 incDecInput[];
GLREF_D TInt sizeAdd;
GLREF_D TInt sizeSub;
GLREF_D TInt sizeMult;
GLREF_D TInt sizeDiv;
GLREF_D TInt sizeUnary;
GLREF_D TInt sizeIncDec;
#if defined (__WINS__) || defined (__X86__)
// Functions from EMGCC32.CPP
GLREF_C TReal32 __addsf3(TReal32 a1,TReal32 a2);
GLREF_C TReal32 __subsf3(TReal32 a1,TReal32 a2);
GLREF_C TReal32 __mulsf3(TReal32 a1,TReal32 a2);
GLREF_C TReal32 __divsf3(TReal32 a1,TReal32 a2);
GLREF_C TReal32 __truncdfsf2(TReal64 a1);
GLDEF_D const TReal32 minDenormalTReal32=1.4E-45f;
#endif
GLDEF_D TReal32 NaNTReal32;
GLDEF_D TReal32 posInfTReal32;
GLDEF_D TReal32 negInfTReal32;
GLDEF_D TReal NaNTReal;
GLDEF_D TReal posInfTReal;
GLDEF_D TReal negInfTReal;
enum TOrder
{
ELessThan,
EEqual,
EGreaterThan
};
LOCAL_D RTest test(_L("T_R32"));
LOCAL_C void initSpecialValues()
//
// Initialise special values
//
{
SReal32 *p32=(SReal32*)&NaNTReal32;
p32->sign=0;
p32->exp=KTReal32SpecialExponent;
p32->man=0x7fffff;
p32=(SReal32*)&posInfTReal32;
p32->sign=0;
p32->exp=KTReal32SpecialExponent;
p32->man=0;
p32=(SReal32*)&negInfTReal32;
p32->sign=1;
p32->exp=KTReal32SpecialExponent;
p32->man=0;
SReal64 *p64=(SReal64*)&NaNTReal;
p64->sign=0;
p64->exp=KTReal64SpecialExponent;
p64->lsm=0xffffffffu;
p64->msm=0xfffff;
p64=(SReal64*)&posInfTReal;
p64->sign=0;
p64->exp=KTReal64SpecialExponent;
p64->lsm=0;
p64->msm=0;
p64=(SReal64*)&negInfTReal;
p64->sign=1;
p64->exp=KTReal64SpecialExponent;
p64->lsm=0;
p64->msm=0;
}
LOCAL_C void testConvert()
//
// Conversion tests
//
{
TRealX f;
TReal input[]=
{
KMaxTReal32inTReal,KMinTReal32inTReal,-KMaxTReal32inTReal,-KMinTReal32inTReal,
KMaxTReal32inTReal,KMinTReal32inTReal,-KMaxTReal32inTReal,-KMinTReal32inTReal,
3.4027E+38,1.1755E-38,-3.4027E+38,-1.1755E-38,
0.0,64.5,-64.5,1.54E+18,-1.54E+18,4.72E-22,-4.72E-22,
posInfTReal,negInfTReal,KNegZeroTReal,
1.4E-45,-1.4E-45,2E-41,-2E-41,1E-38,-1E-38
};
TReal32 expect[]=
{
KMaxTReal32,KMinTReal32,-KMaxTReal32,-KMinTReal32,
KMaxTReal32,KMinTReal32,-KMaxTReal32,-KMinTReal32,
3.4027E+38f,1.17550E-38f,-3.40270E+38f,-1.17550E-38f,
0.0f,64.5f,-64.5f,1.54E+18f,-1.54E+18f,4.72E-22f,-4.72E-22f,
posInfTReal32,negInfTReal32,KNegZeroTReal32,
1.4E-45f,-1.4E-45f,2E-41f,-2E-41f,1E-38f,-1E-38f
};
TInt size=sizeof(input)/sizeof(TReal);
for (TInt ii=0; ii<size; ii++)
{
f=TRealX(expect[ii]);
test(f==TRealX(TReal32(input[ii])));
}
// NaN
// TReal a=NaNTReal;
TReal32 b=NaNTReal32;
f=TRealX(b);
// test(f!=TRealX(TReal32(a)));
test(f.IsNaN());
// See EON Software Defects Bug Report no. HA-287
// There is a bug in MSDev compiler which means comparing TReal32's directly
// does not always work, hence...
/*
test(BitTest(TReal32(3.40270E+38),3.40270E+38f)); // this works
// (BitTest() checks for all 32 bits being identical
TReal32 a=TReal32(3.40270E+38);
TReal32 b=3.40270E+38f;
TReal64 c=3.40270E+38;
TReal32 d=TReal32(c);
test(a==b); // this works
test(d==b); // this works
test(TRealX(TReal32(c))==TRealX(b)); // this works
test(TReal64(TReal32(c))==TReal64(b)); // this panics
test(TReal32(c)==b); // this panics
test(TReal32(3.40270E+38)==3.40270E+38f); // this panics
// As expected, all these work fine under ARM.
*/
}
LOCAL_C void testAdd()
//
// Addition tests
//
{
TReal32 f,g,h;
TRealX ff,gg,hh;
for (TInt ii=0; ii<sizeAdd-1; ii++)
{
f=addInput[ii];
g=addInput[ii+1];
ff=TRealX(f);
gg=TRealX(g);
// Test commute
test(f+g == g+f);
// Test PC real addition using fp-hardware same as TRealX addition
test(TRealX(f+g)==TRealX(TReal32(ff+gg)));
test(TRealX(g+f)==TRealX(TReal32(ff+gg)));
// Test hex-encoded constants for TReal32s generated on PC using fp-hardware same as
// TRealX addition
test(TRealX(*(TReal32*)&addArray[ii])==TRealX(f+g));
test(TRealX(*(TReal32*)&addArray[ii])==TRealX(g+f));
// similarly to tests above ...
h=g;
hh=gg;
hh+=ff;
test(TRealX(h+=f)==TRealX(TReal32(hh)));
test(TRealX(h)==TRealX(TReal32(hh)));
test(TRealX(*(TReal32*)&addArray[ii])==TRealX(h));
//
h=f;
hh=ff;
hh+=gg;
test(TRealX(h+=g)==TRealX(TReal32(hh)));
test(h==TReal32(hh));
test(TRealX(*(TReal32*)&addArray[ii])==TRealX(h));
}
}
LOCAL_C void testSubt()
//
// Subtraction tests
//
{
TReal32 f,g,h;
TRealX ff,gg,hh;
for (TInt ii=0; ii<sizeSub-1; ii++)
{
f=subInput[ii];
g=subInput[ii+1];
ff=TRealX(f);
gg=TRealX(g);
//
// This test fails on GCC (with -O1 switch). The reason is that
// comparing two intermediate floats is unpredictable.
// See http://www.parashift.com/c++-faq-lite/newbie.html#faq-29.18
#ifndef __GCC32__
test(f-g == -(g-f));
#endif
//
test(TRealX(f-g)==TRealX(TReal32(ff-gg)));
test(TRealX(g-f)==TRealX(TReal32(gg-ff)));
test(TRealX(*(TReal32*)&subArray[ii])==TRealX(f-g));
test(TRealX(*(TReal32*)&subArray[ii])==TRealX(-(g-f)));
//
h=g;
hh=gg;
hh-=ff;
test(TRealX(h-=f)==TRealX(TReal32(hh)));
test(TRealX(h)==TRealX(TReal32(hh)));
test(TRealX(*(TReal32*)&subArray[ii])==TRealX(-h));
//
h=f;
hh=ff;
hh-=gg;
test(TRealX(h-=g)==TRealX(TReal32(hh)));
test(TRealX(h)==TRealX(TReal32(hh)));
test(TRealX(*(TReal32*)&subArray[ii])==TRealX(h));
}
}
LOCAL_C void testMult()
//
// Multiplication test
//
{
TReal32 f,g,h;
TRealX ff,gg,hh;
for (TInt ii=0; ii<sizeMult-1; ii++)
{
f=multInput[ii];
g=multInput[ii+1];
ff=TRealX(f);
gg=TRealX(g);
//
test(f*g == g*f);
//
test(TRealX(f*g)==TRealX(TReal32(ff*gg)));
test(TRealX(g*f)==TRealX(TReal32(gg*ff)));
test(TRealX(*(TReal32*)&multArray[ii])==TRealX(f*g));
test(TRealX(*(TReal32*)&multArray[ii])==TRealX(g*f));
//
h=f;
hh=ff;
hh*=gg;
test(TRealX(h*=g)==TRealX(TReal32(hh)));
test(TRealX(h)==TRealX(TReal32(hh)));
test(TRealX(*(TReal32*)&multArray[ii])==TRealX(h));
//
h=g;
hh=gg;
hh*=ff;
test(TRealX(h*=f)==TRealX(TReal32(hh)));
test(TRealX(h)==TRealX(TReal32(hh)));
test(TRealX(*(TReal32*)&multArray[ii])==TRealX(h));
}
}
LOCAL_C void testDiv()
//
// Division test
//
{
TReal32 f,g,h;
TRealX ff,gg,hh;
TInt count=0;
// Panic: Divide by Zero
// f=1.0;
// g=0.0;
// f/=g;
for (TInt ii=0; ii<sizeDiv-1; ii++)
{
f=divInput[ii];
g=divInput[ii+1];
ff=TRealX(f);
gg=TRealX(g);
if (g!=0.0)
{
test(TRealX(f/g)==TRealX(TReal32(ff/gg)));
test(TRealX(*(TReal32*)&divArray[count])==TRealX(f/g));
//
h=f;
hh=ff;
hh/=gg;
test(TRealX(h/=g)==TRealX(TReal32(hh)));
test(TRealX(h)==TRealX(TReal32(hh)));
test(TRealX(*(TReal32*)&divArray[count])==TRealX(h));
++count;
}
if (f!=0.0)
{
test(TRealX(g/f)==TRealX(TReal32(gg/ff)));
h=g;
hh=gg;
hh/=ff;
test(TRealX(h/=f)==TRealX(TReal32(hh)));
test(h==TReal32(hh));
}
};
//Additional test
f=3.9999f;
g=KMinTReal32;
ff=TRealX(f);
gg=TRealX(g);
test(TRealX(f/g)==TRealX(TReal32(ff/gg)));
h=f;
hh=ff;
hh/=gg;
test(TRealX(h/=g)==TRealX(TReal32(hh)));
test(TRealX(h)==TRealX(TReal32(hh)));
}
#if defined (__WINS__) || defined (__X86__)
LOCAL_C void testArithmeticExceptionRaising()
//
// Test that UP_GCC.CPP raise exceptions correctly by calling functions from EMGCC32.CPP which
// are copies of those in UP_GCC.CPP. To be used in debugger only.
// Added by AnnW, December 1996
//
{
TReal32 f,g,h;
// Addition - possible errors are overflow, argument or none
// NB no underflow
f=NaNTReal32;
h=__addsf3(f,f); // argument
f=KMaxTReal32;
h=__addsf3(f,f); // overflow
f=1.0f;
g=2.0f;
h=__addsf3(f,g); // none
test(h==3.0f);
// Subtraction - possible errors are overflow, argument or none
// NB no underflow
f=NaNTReal32;
h=__subsf3(f,f); // argument
f=KMaxTReal32;
g=-KMaxTReal32;
h=__subsf3(f,g); // overflow
f=1.0f;
g=2.0f;
h=__subsf3(f,g); // none
test(h==-1.0f);
// Multiplication - possible errors are argument, overflow, underflow or none
f=NaNTReal32;
h=__mulsf3(f,f); // argument
f=KMaxTReal32;
g=2.0f;
h=__mulsf3(f,g); // overflow
f=minDenormalTReal32;
g=0.1f;
h=__mulsf3(f,g); // underflow
f=1.0f;
g=2.0f;
h=__mulsf3(f,g); // none
test(h==2.0f);
// Division - possible errors are overflow, underflow, divide by zero, argument or none
f=KMaxTReal32;
g=0.5f;
h=__divsf3(f,g); // overflow
f=minDenormalTReal32;
g=10.0f;
h=__divsf3(f,g); // underflow
f=4.0f;
g=0.0f;
h=__divsf3(f,g); // divide by zero
f=0.0f;
g=0.0f;
h=__divsf3(f,g); // argument
f=1.0f;
g=2.0f;
h=__divsf3(f,g); // none
test(h==0.5f);
// Converting double to float - possible errors are overflow, underflow, invalid operation or none
TReal64 d;
d=1.0E+50;
f=__truncdfsf2(d); // overflow
d=1.0E-50;
f=__truncdfsf2(d); // underflow
d=KNaNTReal64;
f=__truncdfsf2(d); // invalid operation
d=4.0;
f=__truncdfsf2(d); // none
}
#endif
LOCAL_C void testUnary()
//
// Unary operator tests
//
{
TReal32 f;
TRealX g;
for (TInt ii=0; ii<sizeUnary-1; ii++)
{
f=unaryInput[ii];
g=TRealX(f);
test(TRealX(-f)==TRealX(TReal32(-g)));
test(TRealX(-f)==TRealX(0.0f-f));
test(TRealX(+f)==TRealX(TReal32(g)));
test(TRealX(+f)==TRealX(0.0f+f));
test(TRealX(*(TReal32*)&unaryArray[ii])==TRealX(-f));
}
}
LOCAL_C void testEqualities(const TReal& aA, TOrder aOrder, const TReal& aB)
//
// Test equality/inequality functions on aA and aB
// aOrder specifies the operand's relative sizes
//
{
// Tautologies
test((aA>aA) ==FALSE);
test((aA<aA) ==FALSE);
test((aA>=aA)==TRUE);
test((aA<=aA)==TRUE);
test((aA==aA)==TRUE);
test((aA!=aA)==FALSE);
if (aOrder!=EEqual)
{
test((aA==aB)==FALSE);
test((aA!=aB)==TRUE);
}
if (aOrder==ELessThan)
{
test((aA<aB) ==TRUE);
test((aA<=aB)==TRUE);
test((aA>aB) ==FALSE);
test((aA>=aB)==FALSE);
}
if (aOrder==EEqual)
{
test((aA==aB)==TRUE);
test((aA!=aB)==FALSE);
test((aA>=aB)==TRUE);
test((aA<=aB)==TRUE);
test((aA>aB)==FALSE);
test((aA<aB)==FALSE);
}
if (aOrder==EGreaterThan)
{
test((aA>aB) ==TRUE);
test((aA>=aB)==TRUE);
test((aA<aB) ==FALSE);
test((aA<=aB)==FALSE);
}
}
LOCAL_C void testEqualities()
//
// Test >, <, >=, <=, ==, !=
//
{
TInt i, size;
TReal32 lessThanMax = KMaxTReal32-TReal32(1.0E+32);
TReal32 greaterThanMin = 1.17550E-38f;
TReal32 zero(0.0f);
TReal32 positive[] =
{KMinTReal32,5.3824705E-26f,1.0f,2387501.0f,5.3824705E+28f,KMaxTReal32};
TReal32 large[] =
{2.0f,KMaxTReal32,-lessThanMax,greaterThanMin,-KMinTReal32,10.4058482f,-10.4058482f,
1.2443345E+14f,1.2443345E+14f,-1.3420344E-16f,132435.97f,5.0E-6f,9.6f,-8.0f};
TReal32 small[] =
{1.0f,lessThanMax,-KMaxTReal32,KMinTReal32,-greaterThanMin,10.4058474f,-10.4058496f,
5.0E-10f,1.2443345E+10f,-5.0382470E+25f,-132435.97f,-5.1E-6f,8.0f,-9.6f};
TReal32 equal[] = // Same as large[]
{2.0f,KMaxTReal32,-lessThanMax,greaterThanMin,-KMinTReal32,10.4058482f,-10.4058482f,
1.2443345E+14f,1.2443345E+14f,-1.3420344E-16f,132435.97f,5.0E-6f,9.6f,-8.0f};
// Tests with zero
size = sizeof(positive)/sizeof(TReal32);
test.Start(_L("Zero"));
testEqualities(zero, EEqual, zero);
for (i=0; i<size; i++)
{
testEqualities(positive[i], EGreaterThan, zero);
testEqualities(-positive[i], ELessThan, zero);
testEqualities(zero, ELessThan, positive[i]);
testEqualities(zero, EGreaterThan, -positive[i]);
}
// Test boundary and other numbers
size = sizeof(large)/sizeof(TReal32);
test.Next(_L("Nonzero"));
for (i=0; i<size; i++)
{
testEqualities(large[i], EGreaterThan, small[i]);
testEqualities(small[i], ELessThan, large[i]);
testEqualities(large[i], EEqual, equal[i]);
}
test.End();
}
LOCAL_C void testIncDec()
//
// Test Pre/Post - increment/decrement
//
{
TInt ii;
TReal32 f;
TRealX g;
test.Start(_L("Pre-increment"));
for (ii=0; ii<sizeIncDec; ii++)
{
f=incDecInput[ii];
g=TRealX(f);
test(TRealX(f)==TRealX(TReal32(g)));
test(TRealX(++f)==TRealX(TReal32(++g)));
test(TRealX(*(TReal32*)&preIncArray1[ii])==TRealX(f));
test(TRealX(f)==TRealX(TReal32(g)));
test(TRealX(++f)==TRealX(TReal32(++g)));
test(TRealX(*(TReal32*)&preIncArray2[ii])==TRealX(f));
test(TRealX(f)==TRealX(TReal32(g)));
}
test.Next(_L("Post-increment"));
for (ii=0; ii<sizeIncDec; ii++)
{
f=incDecInput[ii];
g=TRealX(f);
test(TRealX(f)==TRealX(TReal32(g)));
test(TRealX(f++)==TRealX(TReal32(g++)));
test(TRealX(*(TReal32*)&postIncArray1[ii])==TRealX(f));
test(TRealX(f)==TRealX(TReal32(g)));
test(TRealX(f++)==TRealX(TReal32(g++)));
test(TRealX(*(TReal32*)&postIncArray2[ii])==TRealX(f));
test(TRealX(f)==TRealX(TReal32(g)));
}
test.Next(_L("Pre-decrement"));
for (ii=0; ii<sizeIncDec; ii++)
{
f=incDecInput[ii];
g=TRealX(f);
test(TRealX(f)==TRealX(TReal32(g)));
test(TRealX(--f)==TRealX(TReal32(--g)));
test(TRealX(*(TReal32*)&preDecArray1[ii])==TRealX(f));
test(TRealX(f)==TRealX(TReal32(g)));
test(TRealX(--f)==TRealX(TReal32(--g)));
test(TRealX(*(TReal32*)&preDecArray2[ii])==TRealX(f));
test(TRealX(f)==TRealX(TReal32(g)));
}
test.Next(_L("Post-decrement"));
for (ii=0; ii<sizeIncDec; ii++)
{
f=incDecInput[ii];
g=TRealX(f);
test(TRealX(f)==TRealX(TReal32(g)));
test(TRealX(f--)==TRealX(TReal32(g--)));
test(TRealX(*(TReal32*)&postDecArray1[ii])==TRealX(f));
test(TRealX(f)==TRealX(TReal32(g)));
test(TRealX(f--)==TRealX(TReal32(g--)));
test(TRealX(*(TReal32*)&postDecArray2[ii])==TRealX(f));
test(TRealX(f)==TRealX(TReal32(g)));
}
test.End();
}
LOCAL_C void _matherr(TExcType aType)
{
test.Printf(_L("_matherr: Exception type %u handled\n"),TUint(aType));
}
GLDEF_C TInt E32Main()
//
// Test TReal32
//
{
test.Title();
User::SetExceptionHandler(_matherr,KExceptionFpe);
initSpecialValues();
test.Start(_L("Conversion from TReal to TReal32"));
testConvert();
test.Next(_L("Addition"));
testAdd();
test.Next(_L("Subtraction"));
testSubt();
test.Next(_L("Multiplication"));
testMult();
test.Next(_L("Division"));
testDiv();
#if defined (__WINS__) || defined (__X86__)
test.Next(_L("Arithmetic which emulates UP_GCC and raises an exception"));
testArithmeticExceptionRaising();
#endif
test.Next(_L("Unary Operators"));
testUnary();
test.Next(_L("Equalities and Inequalities"));
testEqualities();
test.Next(_L("Increment and Decrement"));
testIncDec();
test.End();
return(KErrNone);
}