Macintosh-Tools/macemu
1
0
Fork 0
mirror of https://github.com/kanjitalk755/macemu.git synced 2024-12-20 00:29:56 +00:00
Code Issues Projects Releases Wiki Activity

Merge in FP exceptions support but disable it for now as it is incomplete

Browse Source 
and slower. Implement mcrfs. Fix and optimize fctiw with native rounding.
This commit is contained in:
gbeauche 2004-01-24 16:43:45 +00:00
parent 3de5a15902
commit 82808234fa
6 changed files with 345 additions and 111 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
SheepShaver/src/kpx_cpu/src/cpu/ppc/ppc-config.hpp
View File

@ -63,6 +63,18 @@
#endif
/**
* PPC_ENABLE_FPU_EXCEPTIONS
*
* Define to enable a more precise FPU emulation with support for
* exception bits. This is slower and not fully correct yet.
**/
#ifndef PPC_ENABLE_FPU_EXCEPTIONS
#define PPC_ENABLE_FPU_EXCEPTIONS 0
#endif
/**
* PPC_DECODE_CACHE
*

6
SheepShaver/src/kpx_cpu/src/cpu/ppc/ppc-cpu.hpp
View File

@ -84,8 +84,11 @@ protected:
{ record_cr(0, value); }
void record_cr1()
{ cr().set((cr().get() & ~CR_field<1>::mask()) | ((fpscr() >> 4) & 0x0f000000)); }
void record_fpscr();
void fp_classify(double x);
template< class FP >
void fp_classify(FP x);
void fp_setround(int round);
protected:
@ -365,6 +368,7 @@ private:
template< class RA >
void execute_stwcx(uint32 opcode);
void execute_mcrf(uint32 opcode);
void execute_mcrfs(uint32 opcode);
void execute_mtcrf(uint32 opcode);
template< class FM, class RB, class Rc >
void execute_mtfsf(uint32 opcode);

6
SheepShaver/src/kpx_cpu/src/cpu/ppc/ppc-decode.cpp
View File

@ -727,6 +727,12 @@ const powerpc_cpu::instr_info_t powerpc_cpu::powerpc_ii_table[] = {
PPC_I(MCRF),
XL_form, 19, 0, CFLOW_NORMAL
},
{ "mcrfs",
EXECUTE_0(mcrfs),
NULL,
PPC_I(MCRFS),
X_form, 63, 64, CFLOW_NORMAL
},
{ "mfcr",
EXECUTE_GENERIC_ARITH(nop, RD, CR, NONE, NONE, OE_BIT_0, RC_BIT_0),
NULL,

286
SheepShaver/src/kpx_cpu/src/cpu/ppc/ppc-execute.cpp
View File

@ -21,6 +21,7 @@
#include <stdio.h>
#include <math.h>
#include <time.h>
#include <fenv.h>
#include "sysdeps.h"
#include "cpu/vm.hpp"
@ -51,26 +52,26 @@
* RC Input operand register or immediate (optional: operand_NONE)
**/
template< class OP, class RA, class RB, class RC >
template< class RT, class OP, class RA, class RB, class RC >
struct op_apply {
template< class T >
static inline T apply(T a, T b, T c) {
static inline RT apply(T a, T b, T c) {
return OP::apply(a, b, c);
}
};
template< class OP, class RA, class RB >
struct op_apply<OP, RA, RB, null_operand> {
template< class RT, class OP, class RA, class RB >
struct op_apply<RT, OP, RA, RB, null_operand> {
template< class T >
static inline T apply(T a, T b, T c) {
static inline RT apply(T a, T b, T) {
return OP::apply(a, b);
}
};
template< class OP, class RA >
struct op_apply<OP, RA, null_operand, null_operand> {
template< class RT, class OP, class RA >
struct op_apply<RT, OP, RA, null_operand, null_operand> {
template< class T >
static inline T apply(T a, T b, T c) {
static inline RT apply(T a, T, T) {
return OP::apply(a);
}
};
@ -198,7 +199,7 @@ void powerpc_cpu::execute_generic_arith(uint32 opcode)
const uint32 b = RB::get(this, opcode);
const uint32 c = RC::get(this, opcode);
uint32 d = op_apply<OP, RA, RB, RC>::apply(a, b, c);
uint32 d = op_apply<uint32, OP, RA, RB, RC>::apply(a, b, c);
// Set XER (OV, SO) if instruction has OE set
if (OE::test(opcode))
@ -440,6 +441,37 @@ void powerpc_cpu::execute_multiply(uint32 opcode)
increment_pc(4);
}
/**
* Record FPSCR
*
* Update FP exception bits
**/
void powerpc_cpu::record_fpscr()
{
#if PPC_ENABLE_FPU_EXCEPTIONS
// Always update VX
if (fpscr() & (FPSCR_VXSNAN_field::mask() | FPSCR_VXISI_field::mask() | \
FPSCR_VXISI_field::mask() | FPSCR_VXIDI_field::mask() | \
FPSCR_VXZDZ_field::mask() | FPSCR_VXIMZ_field::mask() | \
FPSCR_VXVC_field::mask() | FPSCR_VXSOFT_field::mask() | \
FPSCR_VXSQRT_field::mask() | FPSCR_VXCVI_field::mask()))
fpscr() |= FPSCR_VX_field::mask();
else
fpscr() &= ~FPSCR_VX_field::mask();
// Always update FEX
if (((fpscr() & FPSCR_VX_field::mask()) && (fpscr() & FPSCR_VE_field::mask())) \
|| ((fpscr() & FPSCR_OX_field::mask()) && (fpscr() & FPSCR_OE_field::mask())) \
|| ((fpscr() & FPSCR_UX_field::mask()) && (fpscr() & FPSCR_UE_field::mask())) \
|| ((fpscr() & FPSCR_ZX_field::mask()) && (fpscr() & FPSCR_ZE_field::mask())) \
|| ((fpscr() & FPSCR_XX_field::mask()) && (fpscr() & FPSCR_XE_field::mask())))
fpscr() |= FPSCR_FEX_field::mask();
else
fpscr() &= ~FPSCR_FEX_field::mask();
#endif
}
/**
* Floating-point arithmetics
*
@ -459,31 +491,52 @@ void powerpc_cpu::execute_fp_arith(uint32 opcode)
const double a = RA::get(this, opcode);
const double b = RB::get(this, opcode);
const double c = RC::get(this, opcode);
FP d = op_apply<OP, RA, RB, RC>::apply(a, b, c);
#if 0
// FIXME: Compute FPSCR bits if instruction requests it
// Check for FP Exception Conditions
#if PPC_ENABLE_FPU_EXCEPTIONS
int exceptions = 0;
if (FPSCR) {
exceptions = op_apply<uint32, fp_exception_condition<OP>, RA, RB, RC>::apply(a, b, c);
feclearexcept(FE_ALL_EXCEPT);
}
#endif
FP d = op_apply<double, OP, RA, RB, RC>::apply(a, b, c);
#if PPC_ENABLE_FPU_EXCEPTIONS
if (FPSCR) {
// Always update VX
if (fpscr() & (FPSCR_VXSNAN_field::mask() | FPSCR_VXISI_field::mask() | \
FPSCR_VXISI_field::mask() | FPSCR_VXIDI_field::mask() | \
FPSCR_VXZDZ_field::mask() | FPSCR_VXIMZ_field::mask() | \
FPSCR_VXVC_field::mask() | FPSCR_VXSOFT_field::mask() | \
FPSCR_VXSQRT_field::mask() | FPSCR_VXCVI_field::mask()))
fpscr() |= FPSCR_VX_field::mask();
else
fpscr() &= ~FPSCR_VX_field::mask();
// Check exceptions raised
int masked = 0xffffffff;
int raised = fetestexcept(FE_ALL_EXCEPT);
if (raised & FE_INEXACT) {
exceptions |= FPSCR_XX_field::mask();
exceptions |= FPSCR_FX_field::mask();
}
if (raised & FE_DIVBYZERO) {
exceptions |= FPSCR_ZX_field::mask();
exceptions |= FPSCR_FX_field::mask();
}
else masked &= ~FPSCR_ZX_field::mask();
if (raised & FE_UNDERFLOW) {
exceptions |= FPSCR_UX_field::mask();
exceptions |= FPSCR_FX_field::mask();
}
else masked &= ~FPSCR_UX_field::mask();
if (raised & FE_OVERFLOW) {
exceptions |= FPSCR_OX_field::mask();
exceptions |= FPSCR_FX_field::mask();
}
else masked &= ~FPSCR_OX_field::mask();
fpscr() &= masked;
fpscr() |= exceptions;
// Always update FEX
if (((fpscr() & FPSCR_VX_field::mask()) && (fpscr() & FPSCR_VE_field::mask())) \
|| ((fpscr() & FPSCR_OX_field::mask()) && (fpscr() & FPSCR_OE_field::mask())) \
|| ((fpscr() & FPSCR_UX_field::mask()) && (fpscr() & FPSCR_UE_field::mask())) \
|| ((fpscr() & FPSCR_ZX_field::mask()) && (fpscr() & FPSCR_ZE_field::mask())) \
|| ((fpscr() & FPSCR_XX_field::mask()) && (fpscr() & FPSCR_XE_field::mask())))
fpscr() |= FPSCR_FEX_field::mask();
else
fpscr() &= ~FPSCR_FEX_field::mask();
// Update FPSCR exception bits
record_fpscr();
// FPSCR[FPRF] is set to the class and sign of the result
if (!exceptions || !FPSCR_VE_field::test(fpscr()))
fp_classify(d);
}
#endif
@ -735,95 +788,50 @@ void powerpc_cpu::execute_stwcx(uint32 opcode)
increment_pc(4);
}
/**
* Basic (and probably incorrect) implementation for missing functions
* in older C libraries
**/
#ifndef signbit
#define signbit(X) my_signbit(X)
#endif
static inline bool my_signbit(double X) {
return X < 0.0;
}
#ifndef isless
#define isless(X, Y) my_isless(X, Y)
#endif
static inline bool my_isless(double X, double Y) {
return X < Y;
}
#ifndef isgreater
#define isgreater(X, Y) my_isgreater(X, Y)
#endif
static inline bool my_isgreater(double X, double Y) {
return X > Y;
}
/**
* Floating-point compare instruction
*
* OC Predicate for ordered compare
**/
static inline bool is_NaN(double v) {
any_register x; x.d = v;
return (((x.j & UVAL64(0x7ff0000000000000)) == UVAL64(0x7ff0000000000000)) &&
((x.j & UVAL64(0x000fffffffffffff)) != 0));
}
static inline bool is_SNaN(double v) {
any_register x; x.d = v;
return is_NaN(v) && !(x.j & UVAL64(0x0008000000000000)) ? signbit(v) : false;
}
static inline bool is_QNaN(double v) {
return is_NaN(v) && !is_SNaN(v);
}
static inline bool is_NaN(float v) {
any_register x; x.f = v;
return (((x.i & 0x7f800000) == 0x7f800000) &&
((x.i & 0x007fffff) != 0));
}
static inline bool is_SNaN(float v) {
any_register x; x.f = v;
return is_NaN(v) && !(x.i & 0x00400000) ? signbit(v) : false;
}
static inline bool is_QNaN(float v) {
return is_NaN(v) && !is_SNaN(v);
}
template< bool OC >
void powerpc_cpu::execute_fp_compare(uint32 opcode)
{
const double a = operand_fp_RA::get(this, opcode);
const double b = operand_fp_RB::get(this, opcode);
const int crfd = crfD_field::extract(opcode);
int c;
if (is_NaN(a) || is_NaN(b))
cr().set(crfd, 1);
c = 1;
else if (isless(a, b))
cr().set(crfd, 8);
c = 8;
else if (isgreater(a, b))
cr().set(crfd, 4);
c = 4;
else
cr().set(crfd, 2);
c = 2;
fpscr() = (fpscr() & ~FPSCR_FPCC_field::mask()) | (cr().get(crfd) << 12);
FPSCR_FPCC_field::insert(fpscr(), c);
cr().set(crfd, c);
// Check for FP exception condition
#if PPC_ENABLE_FPU_EXCEPTIONS
if (is_SNaN(a) || is_SNaN(b)) {
fpscr() |= FPSCR_VXSNAN_field::mask();
if (OC && !FPSCR_VE_field::test(fpscr()))
fpscr() |= FPSCR_FX_field::mask();
if (OC && !FPSCR_VE_field::test(fpscr())) {
fpscr() |= FPSCR_VXVC_field::mask();
fpscr() |= FPSCR_FX_field::mask();
}
}
else if (OC && (is_QNaN(a) || is_QNaN(b)))
else if (OC && (is_QNaN(a) || is_QNaN(b))) {
fpscr() |= FPSCR_VXVC_field::mask();
fpscr() |= FPSCR_FX_field::mask();
}
#endif
// Update FPSCR exception bits
record_fpscr();
increment_pc(4);
}
@ -842,20 +850,16 @@ void powerpc_cpu::execute_fp_int_convert(uint32 opcode)
const uint32 r = RN::get(this, opcode);
any_register d;
switch (r) {
case 0: // Round to nearest
d.j = (int32)(b + 0.5);
break;
case 1: // Round toward zero
// Bounds checking and convert to integer word
if (b > (double)0x7fffffff)
d.j = 0x7fffffff;
else if (b < -(double)0x80000000)
d.j = 0x80000000;
else
d.j = (int32)b;
break;
case 2: // Round toward +infinity
d.j = (int32)ceil(b);
break;
case 3: // Round toward -infinity
d.j = (int32)floor(b);
break;
}
// Update FPSCR exception bits
record_fpscr();
// Set CR1 (FX, FEX, VX, VOX) if instruction has Rc set
if (Rc::test(opcode))
@ -872,7 +876,8 @@ void powerpc_cpu::execute_fp_int_convert(uint32 opcode)
* Rc Predicate to record CR1
**/
void powerpc_cpu::fp_classify(double x)
template< class FP >
void powerpc_cpu::fp_classify(FP x)
{
uint32 c = fpscr() & ~FPSCR_FPRF_field::mask();
uint8 fc = fpclassify(x);
@ -955,6 +960,27 @@ void powerpc_cpu::execute_mcrf(uint32 opcode)
increment_pc(4);
}
void powerpc_cpu::execute_mcrfs(uint32 opcode)
{
const int crfS = crfS_field::extract(opcode);
const int crfD = crfD_field::extract(opcode);
// The contents of FPSCR field crfS are copied to CR field crfD
const uint32 m = 0xf << (28 - 4 * crfS);
cr().set(crfD, (fpscr() & m) >> (28 - 4 * crfS));
// All exception bits copied (except FEX and VX) are cleared in the FPSCR
fpscr() &= ~(m & (FPSCR_FX_field::mask() | FPSCR_OX_field::mask() |
FPSCR_UX_field::mask() | FPSCR_ZX_field::mask() |
FPSCR_XX_field::mask() | FPSCR_VXSNAN_field::mask() |
FPSCR_VXISI_field::mask() | FPSCR_VXIDI_field::mask() |
FPSCR_VXZDZ_field::mask() | FPSCR_VXIMZ_field::mask() |
FPSCR_VXVC_field::mask() | FPSCR_VXSOFT_field::mask() |
FPSCR_VXSQRT_field::mask() | FPSCR_VXCVI_field::mask()));
increment_pc(4);
}
void powerpc_cpu::execute_mtcrf(uint32 opcode)
{
uint32 mask = field2mask[CRM_field::extract(opcode)];
@ -962,6 +988,25 @@ void powerpc_cpu::execute_mtcrf(uint32 opcode)
increment_pc(4);
}
void powerpc_cpu::fp_setround(int round)
{
// Set native rounding mode
switch (round) {
case 0:
fesetround(FE_TONEAREST);
break;
case 1:
fesetround(FE_TOWARDZERO);
break;
case 2:
fesetround(FE_UPWARD);
break;
case 3:
fesetround(FE_DOWNWARD);
break;
}
}
template< class FM, class RB, class Rc >
void powerpc_cpu::execute_mtfsf(uint32 opcode)
{
@ -976,6 +1021,10 @@ void powerpc_cpu::execute_mtfsf(uint32 opcode)
// Move frB bits to FPSCR according to field mask
fpscr() = (fsf & m) | (fpscr() & ~m);
// Update native FP control word
if (m & FPSCR_RN_field::mask())
fp_setround(FPSCR_RN_field::extract(fpscr()));
// Set CR1 (FX, FEX, VX, VOX) if instruction has Rc set
if (Rc::test(opcode))
record_cr1();
@ -996,6 +1045,13 @@ void powerpc_cpu::execute_mtfsfi(uint32 opcode)
// Move immediate to FPSCR according to field crfD
fpscr() = (RB::get(this, opcode) & m) | (fpscr() & ~m);
// Update native FP control word
if (m & FPSCR_RN_field::mask())
fp_setround(FPSCR_RN_field::extract(fpscr()));
// Update FPSCR exception bits
record_fpscr();
// Set CR1 (FX, FEX, VX, VOX) if instruction has Rc set
if (Rc::test(opcode))
record_cr1();
@ -1010,6 +1066,13 @@ void powerpc_cpu::execute_mtfsb(uint32 opcode)
const uint32 crbD = crbD_field::extract(opcode);
fpscr() = (fpscr() & ~(1 << (31 - crbD))) | (RB::get(this, opcode) << (31 - crbD));
// Update native FP control word
if (crbD & FPSCR_RN_field::mask())
fp_setround(FPSCR_RN_field::extract(fpscr()));
// Update FPSCR exception bits
record_fpscr();
// Set CR1 (FX, FEX, VX, VOX) if instruction has Rc set
if (Rc::test(opcode))
record_cr1();
@ -1023,6 +1086,9 @@ void powerpc_cpu::execute_mffs(uint32 opcode)
// Move FPSCR to FPR(FRD)
operand_fp_dw_RD::set(this, opcode, fpscr());
// Update FPSCR exception bits
record_fpscr();
// Set CR1 (FX, FEX, VX, VOX) if instruction has Rc set
if (Rc::test(opcode))
record_cr1();

145
SheepShaver/src/kpx_cpu/src/cpu/ppc/ppc-execute.hpp
View File

@ -131,4 +131,149 @@ DEFINE_HELPER(do_execute_divide, (uint32 RA, uint32 RB))
RETURN(RD);
}
/**
* FP classification
**/
static inline bool is_NaN(double v) {
any_register x; x.d = v;
return (((x.j & UVAL64(0x7ff0000000000000)) == UVAL64(0x7ff0000000000000)) &&
((x.j & UVAL64(0x000fffffffffffff)) != 0));
}
static inline bool is_SNaN(double v) {
any_register x; x.d = v;
return is_NaN(v) && !(x.j & UVAL64(0x0008000000000000)) ? signbit(v) : false;
}
static inline bool is_QNaN(double v) {
return is_NaN(v) && !is_SNaN(v);
}
static inline bool is_NaN(float v) {
any_register x; x.f = v;
return (((x.i & 0x7f800000) == 0x7f800000) &&
((x.i & 0x007fffff) != 0));
}
static inline bool is_SNaN(float v) {
any_register x; x.f = v;
return is_NaN(v) && !(x.i & 0x00400000) ? signbit(v) : false;
}
static inline bool is_QNaN(float v) {
return is_NaN(v) && !is_SNaN(v);
}
/**
* Check for FP Exception Conditions
**/
template< class OP >
struct fp_exception_condition {
static inline uint32 apply(double) {
return 0;
}
static inline uint32 apply(double, double) {
return 0;
}
static inline uint32 apply(double, double, double) {
return 0;
}
};
template< class FP >
struct fp_invalid_operation_condition {
static inline uint32 apply(FP a, FP b, int check, bool negate = false) {
uint32 exceptions = 0;
if (FPSCR_VXSNAN_field::test(check) && (is_SNaN(a) || is_SNaN(b))) {
exceptions |= FPSCR_VXSNAN_field::mask();
exceptions |= FPSCR_FX_field::mask();
}
if (FPSCR_VXISI_field::test(check) && isinf(a) && isinf(b)) {
if ((negate && (signbit(a) == signbit(b))) ||
(!negate && (signbit(a) != signbit(b)))) {
exceptions |= FPSCR_VXISI_field::mask();
exceptions |= FPSCR_FX_field::mask();
}
}
if (FPSCR_VXIDI_field::test(check) && isinf(a) && isinf(b)) {
exceptions |= FPSCR_VXIDI_field::mask();
exceptions |= FPSCR_FX_field::mask();
}
if (FPSCR_VXZDZ_field::test(check) && a == 0 && b == 0) {
exceptions |= FPSCR_VXZDZ_field::mask();
exceptions |= FPSCR_FX_field::mask();
}
if (FPSCR_VXIMZ_field::test(check) && a == 0 && isinf(b)) {
exceptions |= FPSCR_VXIMZ_field::mask();
exceptions |= FPSCR_FX_field::mask();
}
if (FPSCR_VXVC_field::test(check) && (is_NaN(a) || is_NaN(b))) {
exceptions |= FPSCR_VXVC_field::mask();
exceptions |= FPSCR_FX_field::mask();
}
if (FPSCR_VXSOFT_field::test(check)) {
exceptions |= FPSCR_VXSOFT_field::mask();
exceptions |= FPSCR_FX_field::mask();
}
if (FPSCR_VXSQRT_field::test(check) && signbit(a)) {
exceptions |= FPSCR_VXSQRT_field::mask();
exceptions |= FPSCR_FX_field::mask();
}
return exceptions;
}
};
#define DEFINE_FP_INVALID_OPERATION(OP, TYPE, EXCP, NEGATE) \
template<> \
struct fp_exception_condition<OP> { \
static inline uint32 apply(TYPE a, TYPE b) { \
return fp_invalid_operation_condition<TYPE>::apply(a, b, EXCP, NEGATE); \
} \
};
DEFINE_FP_INVALID_OPERATION(op_fadd, double, FPSCR_VXSNAN_field::mask() | FPSCR_VXISI_field::mask(), 0);
DEFINE_FP_INVALID_OPERATION(op_fsub, double, FPSCR_VXSNAN_field::mask() | FPSCR_VXISI_field::mask(), 1);
DEFINE_FP_INVALID_OPERATION(op_fmul, double, FPSCR_VXSNAN_field::mask() | FPSCR_VXIMZ_field::mask(), 0);
template< class FP >
struct fp_divide_exception_condition {
static inline uint32 apply(FP a, FP b) {
int exceptions =
fp_invalid_operation_condition<FP>::
apply(a, b,
FPSCR_VXSNAN_field::mask() | FPSCR_VXIDI_field::mask() | FPSCR_VXZDZ_field::mask());
#if 0
if (!exceptions && b == 0)
exceptions = FPSCR_ZX_field::mask() | FPSCR_FX_field::mask();
#endif
return exceptions;
}
};
template<> struct fp_exception_condition<op_fdiv> : fp_divide_exception_condition<double> { };
template< class FP, bool NG >
struct fp_fma_exception_condition {
static inline uint32 apply(FP a, FP b, FP c) {
#if 1
return fp_invalid_operation_condition<FP>::
apply(a, b, FPSCR_VXSNAN_field::mask() | FPSCR_VXIMZ_field::mask());
#else
// FIXME: we are losing precision
double p = a * b;
return (fp_invalid_operation_condition<FP>::
apply(a, b, FPSCR_VXSNAN_field::mask() | FPSCR_VXIMZ_field::mask(), false) |
fp_invalid_operation_condition<FP>::
apply(p, c, FPSCR_VXSNAN_field::mask() | FPSCR_VXISI_field::mask(), NG));
#endif
}
};
template<> struct fp_exception_condition<op_fmadd> : fp_fma_exception_condition<double, false> { };
template<> struct fp_exception_condition<op_fmsub> : fp_fma_exception_condition<double, true> { };
template<> struct fp_exception_condition<op_fnmadd> : fp_fma_exception_condition<double, false> { };
template<> struct fp_exception_condition<op_fnmsub> : fp_fma_exception_condition<double, true> { };
#endif /* PPC_EXECUTE_H */

1
SheepShaver/src/kpx_cpu/src/cpu/ppc/ppc-instructions.hpp
View File

@ -132,6 +132,7 @@ enum powerpc_instruction {
PPC_I(LWZUX),
PPC_I(LWZX),
PPC_I(MCRF),
PPC_I(MCRFS),
PPC_I(MFCR),
PPC_I(MFFS),
PPC_I(MFMSR),