450 lines
13 KiB
C
450 lines
13 KiB
C
/*
|
|
vfp/vfp.h - ARM VFPv3 emulation unit - SoftFloat lib helper
|
|
Copyright (C) 2003 Skyeye Develop Group
|
|
for help please send mail to <skyeye-developer@lists.gro.clinux.org>
|
|
|
|
This program is free software; you can redistribute it and/or modify
|
|
it under the terms of the GNU General Public License as published by
|
|
the Free Software Foundation; either version 2 of the License, or
|
|
(at your option) any later version.
|
|
|
|
This program is distributed in the hope that it will be useful,
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
GNU General Public License for more details.
|
|
|
|
You should have received a copy of the GNU General Public License
|
|
along with this program; if not, write to the Free Software
|
|
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
|
*/
|
|
|
|
/*
|
|
* The following code is derivative from Linux Android kernel vfp
|
|
* floating point support.
|
|
*
|
|
* Copyright (C) 2004 ARM Limited.
|
|
* Written by Deep Blue Solutions Limited.
|
|
*
|
|
* This program is free software; you can redistribute it and/or modify
|
|
* it under the terms of the GNU General Public License version 2 as
|
|
* published by the Free Software Foundation.
|
|
*/
|
|
|
|
#pragma once
|
|
|
|
#include <cstdio>
|
|
#include "common/common_types.h"
|
|
#include "skyeye_interpreter/skyeye_common/armstate.h"
|
|
#include "skyeye_interpreter/skyeye_common/vfp/asm_vfp.h"
|
|
|
|
#define do_div(n, base) {n/=base;}
|
|
|
|
enum : u32 {
|
|
FOP_MASK = 0x00b00040,
|
|
FOP_FMAC = 0x00000000,
|
|
FOP_FNMAC = 0x00000040,
|
|
FOP_FMSC = 0x00100000,
|
|
FOP_FNMSC = 0x00100040,
|
|
FOP_FMUL = 0x00200000,
|
|
FOP_FNMUL = 0x00200040,
|
|
FOP_FADD = 0x00300000,
|
|
FOP_FSUB = 0x00300040,
|
|
FOP_FDIV = 0x00800000,
|
|
FOP_EXT = 0x00b00040
|
|
};
|
|
|
|
#define FOP_TO_IDX(inst) ((inst & 0x00b00000) >> 20 | (inst & (1 << 6)) >> 4)
|
|
|
|
enum : u32 {
|
|
FEXT_MASK = 0x000f0080,
|
|
FEXT_FCPY = 0x00000000,
|
|
FEXT_FABS = 0x00000080,
|
|
FEXT_FNEG = 0x00010000,
|
|
FEXT_FSQRT = 0x00010080,
|
|
FEXT_FCMP = 0x00040000,
|
|
FEXT_FCMPE = 0x00040080,
|
|
FEXT_FCMPZ = 0x00050000,
|
|
FEXT_FCMPEZ = 0x00050080,
|
|
FEXT_FCVT = 0x00070080,
|
|
FEXT_FUITO = 0x00080000,
|
|
FEXT_FSITO = 0x00080080,
|
|
FEXT_FTOUI = 0x000c0000,
|
|
FEXT_FTOUIZ = 0x000c0080,
|
|
FEXT_FTOSI = 0x000d0000,
|
|
FEXT_FTOSIZ = 0x000d0080
|
|
};
|
|
|
|
#define FEXT_TO_IDX(inst) ((inst & 0x000f0000) >> 15 | (inst & (1 << 7)) >> 7)
|
|
|
|
#define vfp_get_sd(inst) ((inst & 0x0000f000) >> 11 | (inst & (1 << 22)) >> 22)
|
|
#define vfp_get_dd(inst) ((inst & 0x0000f000) >> 12 | (inst & (1 << 22)) >> 18)
|
|
#define vfp_get_sm(inst) ((inst & 0x0000000f) << 1 | (inst & (1 << 5)) >> 5)
|
|
#define vfp_get_dm(inst) ((inst & 0x0000000f) | (inst & (1 << 5)) >> 1)
|
|
#define vfp_get_sn(inst) ((inst & 0x000f0000) >> 15 | (inst & (1 << 7)) >> 7)
|
|
#define vfp_get_dn(inst) ((inst & 0x000f0000) >> 16 | (inst & (1 << 7)) >> 3)
|
|
|
|
#define vfp_single(inst) (((inst) & 0x0000f00) == 0xa00)
|
|
|
|
inline u32 vfp_shiftright32jamming(u32 val, unsigned int shift)
|
|
{
|
|
if (shift) {
|
|
if (shift < 32)
|
|
val = val >> shift | ((val << (32 - shift)) != 0);
|
|
else
|
|
val = val != 0;
|
|
}
|
|
return val;
|
|
}
|
|
|
|
inline u64 vfp_shiftright64jamming(u64 val, unsigned int shift)
|
|
{
|
|
if (shift) {
|
|
if (shift < 64)
|
|
val = val >> shift | ((val << (64 - shift)) != 0);
|
|
else
|
|
val = val != 0;
|
|
}
|
|
return val;
|
|
}
|
|
|
|
inline u32 vfp_hi64to32jamming(u64 val)
|
|
{
|
|
u32 v;
|
|
u32 highval = val >> 32;
|
|
u32 lowval = val & 0xffffffff;
|
|
|
|
if (lowval >= 1)
|
|
v = highval | 1;
|
|
else
|
|
v = highval;
|
|
|
|
return v;
|
|
}
|
|
|
|
inline void add128(u64* resh, u64* resl, u64 nh, u64 nl, u64 mh, u64 ml)
|
|
{
|
|
*resl = nl + ml;
|
|
*resh = nh + mh;
|
|
if (*resl < nl)
|
|
*resh += 1;
|
|
}
|
|
|
|
inline void sub128(u64* resh, u64* resl, u64 nh, u64 nl, u64 mh, u64 ml)
|
|
{
|
|
*resl = nl - ml;
|
|
*resh = nh - mh;
|
|
if (*resl > nl)
|
|
*resh -= 1;
|
|
}
|
|
|
|
inline void mul64to128(u64* resh, u64* resl, u64 n, u64 m)
|
|
{
|
|
u32 nh, nl, mh, ml;
|
|
u64 rh, rma, rmb, rl;
|
|
|
|
nl = static_cast<u32>(n);
|
|
ml = static_cast<u32>(m);
|
|
rl = (u64)nl * ml;
|
|
|
|
nh = n >> 32;
|
|
rma = (u64)nh * ml;
|
|
|
|
mh = m >> 32;
|
|
rmb = (u64)nl * mh;
|
|
rma += rmb;
|
|
|
|
rh = (u64)nh * mh;
|
|
rh += ((u64)(rma < rmb) << 32) + (rma >> 32);
|
|
|
|
rma <<= 32;
|
|
rl += rma;
|
|
rh += (rl < rma);
|
|
|
|
*resl = rl;
|
|
*resh = rh;
|
|
}
|
|
|
|
inline void shift64left(u64* resh, u64* resl, u64 n)
|
|
{
|
|
*resh = n >> 63;
|
|
*resl = n << 1;
|
|
}
|
|
|
|
inline u64 vfp_hi64multiply64(u64 n, u64 m)
|
|
{
|
|
u64 rh, rl;
|
|
mul64to128(&rh, &rl, n, m);
|
|
return rh | (rl != 0);
|
|
}
|
|
|
|
inline u64 vfp_estimate_div128to64(u64 nh, u64 nl, u64 m)
|
|
{
|
|
u64 mh, ml, remh, reml, termh, terml, z;
|
|
|
|
if (nh >= m)
|
|
return ~0ULL;
|
|
mh = m >> 32;
|
|
if (mh << 32 <= nh) {
|
|
z = 0xffffffff00000000ULL;
|
|
} else {
|
|
z = nh;
|
|
do_div(z, mh);
|
|
z <<= 32;
|
|
}
|
|
mul64to128(&termh, &terml, m, z);
|
|
sub128(&remh, &reml, nh, nl, termh, terml);
|
|
ml = m << 32;
|
|
while ((s64)remh < 0) {
|
|
z -= 0x100000000ULL;
|
|
add128(&remh, &reml, remh, reml, mh, ml);
|
|
}
|
|
remh = (remh << 32) | (reml >> 32);
|
|
if (mh << 32 <= remh) {
|
|
z |= 0xffffffff;
|
|
} else {
|
|
do_div(remh, mh);
|
|
z |= remh;
|
|
}
|
|
return z;
|
|
}
|
|
|
|
// Operations on unpacked elements
|
|
#define vfp_sign_negate(sign) (sign ^ 0x8000)
|
|
|
|
// Single-precision
|
|
struct vfp_single {
|
|
s16 exponent;
|
|
u16 sign;
|
|
u32 significand;
|
|
};
|
|
|
|
// VFP_SINGLE_MANTISSA_BITS - number of bits in the mantissa
|
|
// VFP_SINGLE_EXPONENT_BITS - number of bits in the exponent
|
|
// VFP_SINGLE_LOW_BITS - number of low bits in the unpacked significand
|
|
// which are not propagated to the float upon packing.
|
|
#define VFP_SINGLE_MANTISSA_BITS (23)
|
|
#define VFP_SINGLE_EXPONENT_BITS (8)
|
|
#define VFP_SINGLE_LOW_BITS (32 - VFP_SINGLE_MANTISSA_BITS - 2)
|
|
#define VFP_SINGLE_LOW_BITS_MASK ((1 << VFP_SINGLE_LOW_BITS) - 1)
|
|
|
|
// The bit in an unpacked float which indicates that it is a quiet NaN
|
|
#define VFP_SINGLE_SIGNIFICAND_QNAN (1 << (VFP_SINGLE_MANTISSA_BITS - 1 + VFP_SINGLE_LOW_BITS))
|
|
|
|
// Operations on packed single-precision numbers
|
|
#define vfp_single_packed_sign(v) ((v) & 0x80000000)
|
|
#define vfp_single_packed_negate(v) ((v) ^ 0x80000000)
|
|
#define vfp_single_packed_abs(v) ((v) & ~0x80000000)
|
|
#define vfp_single_packed_exponent(v) (((v) >> VFP_SINGLE_MANTISSA_BITS) & ((1 << VFP_SINGLE_EXPONENT_BITS) - 1))
|
|
#define vfp_single_packed_mantissa(v) ((v) & ((1 << VFP_SINGLE_MANTISSA_BITS) - 1))
|
|
|
|
enum : u32 {
|
|
VFP_NUMBER = (1 << 0),
|
|
VFP_ZERO = (1 << 1),
|
|
VFP_DENORMAL = (1 << 2),
|
|
VFP_INFINITY = (1 << 3),
|
|
VFP_NAN = (1 << 4),
|
|
VFP_NAN_SIGNAL = (1 << 5),
|
|
|
|
VFP_QNAN = (VFP_NAN),
|
|
VFP_SNAN = (VFP_NAN|VFP_NAN_SIGNAL)
|
|
};
|
|
|
|
inline int vfp_single_type(const vfp_single* s)
|
|
{
|
|
int type = VFP_NUMBER;
|
|
if (s->exponent == 255) {
|
|
if (s->significand == 0)
|
|
type = VFP_INFINITY;
|
|
else if (s->significand & VFP_SINGLE_SIGNIFICAND_QNAN)
|
|
type = VFP_QNAN;
|
|
else
|
|
type = VFP_SNAN;
|
|
} else if (s->exponent == 0) {
|
|
if (s->significand == 0)
|
|
type |= VFP_ZERO;
|
|
else
|
|
type |= VFP_DENORMAL;
|
|
}
|
|
return type;
|
|
}
|
|
|
|
// Unpack a single-precision float. Note that this returns the magnitude
|
|
// of the single-precision float mantissa with the 1. if necessary,
|
|
// aligned to bit 30.
|
|
inline void vfp_single_unpack(vfp_single* s, s32 val, u32* fpscr)
|
|
{
|
|
s->sign = vfp_single_packed_sign(val) >> 16,
|
|
s->exponent = vfp_single_packed_exponent(val);
|
|
|
|
u32 significand = ((u32)val << (32 - VFP_SINGLE_MANTISSA_BITS)) >> 2;
|
|
if (s->exponent && s->exponent != 255)
|
|
significand |= 0x40000000;
|
|
s->significand = significand;
|
|
|
|
// If flush-to-zero mode is enabled, turn the denormal into zero.
|
|
// On a VFPv2 architecture, the sign of the zero is always positive.
|
|
if ((*fpscr & FPSCR_FLUSH_TO_ZERO) != 0 && (vfp_single_type(s) & VFP_DENORMAL) != 0) {
|
|
s->sign = 0;
|
|
s->exponent = 0;
|
|
s->significand = 0;
|
|
*fpscr |= FPSCR_IDC;
|
|
}
|
|
}
|
|
|
|
// Re-pack a single-precision float. This assumes that the float is
|
|
// already normalised such that the MSB is bit 30, _not_ bit 31.
|
|
inline s32 vfp_single_pack(const vfp_single* s)
|
|
{
|
|
u32 val = (s->sign << 16) +
|
|
(s->exponent << VFP_SINGLE_MANTISSA_BITS) +
|
|
(s->significand >> VFP_SINGLE_LOW_BITS);
|
|
return (s32)val;
|
|
}
|
|
|
|
|
|
u32 vfp_single_normaliseround(ARMul_State* state, int sd, vfp_single* vs, u32 fpscr, u32 exceptions, const char* func);
|
|
|
|
// Double-precision
|
|
struct vfp_double {
|
|
s16 exponent;
|
|
u16 sign;
|
|
u64 significand;
|
|
};
|
|
|
|
// VFP_REG_ZERO is a special register number for vfp_get_double
|
|
// which returns (double)0.0. This is useful for the compare with
|
|
// zero instructions.
|
|
#ifdef CONFIG_VFPv3
|
|
#define VFP_REG_ZERO 32
|
|
#else
|
|
#define VFP_REG_ZERO 16
|
|
#endif
|
|
|
|
#define VFP_DOUBLE_MANTISSA_BITS (52)
|
|
#define VFP_DOUBLE_EXPONENT_BITS (11)
|
|
#define VFP_DOUBLE_LOW_BITS (64 - VFP_DOUBLE_MANTISSA_BITS - 2)
|
|
#define VFP_DOUBLE_LOW_BITS_MASK ((1 << VFP_DOUBLE_LOW_BITS) - 1)
|
|
|
|
// The bit in an unpacked double which indicates that it is a quiet NaN
|
|
#define VFP_DOUBLE_SIGNIFICAND_QNAN (1ULL << (VFP_DOUBLE_MANTISSA_BITS - 1 + VFP_DOUBLE_LOW_BITS))
|
|
|
|
// Operations on packed single-precision numbers
|
|
#define vfp_double_packed_sign(v) ((v) & (1ULL << 63))
|
|
#define vfp_double_packed_negate(v) ((v) ^ (1ULL << 63))
|
|
#define vfp_double_packed_abs(v) ((v) & ~(1ULL << 63))
|
|
#define vfp_double_packed_exponent(v) (((v) >> VFP_DOUBLE_MANTISSA_BITS) & ((1 << VFP_DOUBLE_EXPONENT_BITS) - 1))
|
|
#define vfp_double_packed_mantissa(v) ((v) & ((1ULL << VFP_DOUBLE_MANTISSA_BITS) - 1))
|
|
|
|
inline int vfp_double_type(const vfp_double* s)
|
|
{
|
|
int type = VFP_NUMBER;
|
|
if (s->exponent == 2047) {
|
|
if (s->significand == 0)
|
|
type = VFP_INFINITY;
|
|
else if (s->significand & VFP_DOUBLE_SIGNIFICAND_QNAN)
|
|
type = VFP_QNAN;
|
|
else
|
|
type = VFP_SNAN;
|
|
} else if (s->exponent == 0) {
|
|
if (s->significand == 0)
|
|
type |= VFP_ZERO;
|
|
else
|
|
type |= VFP_DENORMAL;
|
|
}
|
|
return type;
|
|
}
|
|
|
|
// Unpack a double-precision float. Note that this returns the magnitude
|
|
// of the double-precision float mantissa with the 1. if necessary,
|
|
// aligned to bit 62.
|
|
inline void vfp_double_unpack(vfp_double* s, s64 val, u32* fpscr)
|
|
{
|
|
s->sign = vfp_double_packed_sign(val) >> 48;
|
|
s->exponent = vfp_double_packed_exponent(val);
|
|
|
|
u64 significand = ((u64)val << (64 - VFP_DOUBLE_MANTISSA_BITS)) >> 2;
|
|
if (s->exponent && s->exponent != 2047)
|
|
significand |= (1ULL << 62);
|
|
s->significand = significand;
|
|
|
|
// If flush-to-zero mode is enabled, turn the denormal into zero.
|
|
// On a VFPv2 architecture, the sign of the zero is always positive.
|
|
if ((*fpscr & FPSCR_FLUSH_TO_ZERO) != 0 && (vfp_double_type(s) & VFP_DENORMAL) != 0) {
|
|
s->sign = 0;
|
|
s->exponent = 0;
|
|
s->significand = 0;
|
|
*fpscr |= FPSCR_IDC;
|
|
}
|
|
}
|
|
|
|
// Re-pack a double-precision float. This assumes that the float is
|
|
// already normalised such that the MSB is bit 30, _not_ bit 31.
|
|
inline s64 vfp_double_pack(const vfp_double* s)
|
|
{
|
|
u64 val = ((u64)s->sign << 48) +
|
|
((u64)s->exponent << VFP_DOUBLE_MANTISSA_BITS) +
|
|
(s->significand >> VFP_DOUBLE_LOW_BITS);
|
|
return (s64)val;
|
|
}
|
|
|
|
u32 vfp_estimate_sqrt_significand(u32 exponent, u32 significand);
|
|
|
|
// A special flag to tell the normalisation code not to normalise.
|
|
#define VFP_NAN_FLAG 0x100
|
|
|
|
// A bit pattern used to indicate the initial (unset) value of the
|
|
// exception mask, in case nothing handles an instruction. This
|
|
// doesn't include the NAN flag, which get masked out before
|
|
// we check for an error.
|
|
#define VFP_EXCEPTION_ERROR ((u32)-1 & ~VFP_NAN_FLAG)
|
|
|
|
// A flag to tell vfp instruction type.
|
|
// OP_SCALAR - This operation always operates in scalar mode
|
|
// OP_SD - The instruction exceptionally writes to a single precision result.
|
|
// OP_DD - The instruction exceptionally writes to a double precision result.
|
|
// OP_SM - The instruction exceptionally reads from a single precision operand.
|
|
enum : u32 {
|
|
OP_SCALAR = (1 << 0),
|
|
OP_SD = (1 << 1),
|
|
OP_DD = (1 << 1),
|
|
OP_SM = (1 << 2)
|
|
};
|
|
|
|
struct op {
|
|
u32 (* const fn)(ARMul_State* state, int dd, int dn, int dm, u32 fpscr);
|
|
u32 flags;
|
|
};
|
|
|
|
inline u32 fls(u32 x)
|
|
{
|
|
int r = 32;
|
|
|
|
if (!x)
|
|
return 0;
|
|
if (!(x & 0xffff0000u)) {
|
|
x <<= 16;
|
|
r -= 16;
|
|
}
|
|
if (!(x & 0xff000000u)) {
|
|
x <<= 8;
|
|
r -= 8;
|
|
}
|
|
if (!(x & 0xf0000000u)) {
|
|
x <<= 4;
|
|
r -= 4;
|
|
}
|
|
if (!(x & 0xc0000000u)) {
|
|
x <<= 2;
|
|
r -= 2;
|
|
}
|
|
if (!(x & 0x80000000u)) {
|
|
x <<= 1;
|
|
r -= 1;
|
|
}
|
|
return r;
|
|
|
|
}
|
|
|
|
u32 vfp_double_multiply(vfp_double* vdd, vfp_double* vdn, vfp_double* vdm, u32 fpscr);
|
|
u32 vfp_double_add(vfp_double* vdd, vfp_double* vdn, vfp_double *vdm, u32 fpscr);
|
|
u32 vfp_double_normaliseround(ARMul_State* state, int dd, vfp_double* vd, u32 fpscr, u32 exceptions, const char* func);
|